Acta Psychiatr Scand 2014: 130: 227–237 All rights reserved DOI: 10.1111/acps.12243
© 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd ACTA PSYCHIATRICA SCANDINAVICA
1910s’ brains revisited. Cortical complexity in early 20th century patients with intellectual disability or with dementia praecox Sandu A-L, Paillere Martinot M-L, Artiges E, Martinot J-L. 1910s’ brains revisited. Cortical complexity in early 20th century patients with intellectual disability or with dementia praecox.
A.-L. Sandu1,
M.-L. Paillere Martinot2, E. Artiges1, J.-L. Martinot1,2 Research Unit 1000, Frederic Joliot Hospital Department, INSERM –CEA – Paris Sud University, Orsay, and 2Adolescent Medicine Department, Maison de Solenn, Cochin Hospital, APHP, Paris Descartes University, Paris, France
1
Objective: The idea of cortical surface anomalies in subjects with intellectual disability (mental retardation) and schizophrenia can be traced back to early 20th century qualitative observations. Since it is unknown whether modern quantitative measures of cortical complexity and folding would retrieve those early empirical observations, we measured fractal dimension and sulcal span index in photographs of human brains taken in the 1910’s. Method: Brain photographs were compared between 36 patients with mental retardation and 21 patients with dementia praecox for the fractal dimension and sulcal span index. Also, a mental retardation subgroup with no-or-non-understandable speech (n = 12) was compared with a subgroup with comprehensible speech (n = 23). Results: Mental retardation group had a lower whole-brain fractal dimension than dementia praecox, and a higher sulcal span index in left posterior cortex. The mental retardation subgroup with comprehensible speech had a lower fractal dimension in left hemisphere than the subgroup with no-or-non-understandable speech and a lower sulcal index in left posterior cortex. Conclusion: Measures of cortical complexity and folding suggest differences between mental retardation and dementia praecox, and regional variations according to language abilities in mental retardation. The findings provide a unique picture of cortical surface changes in their original untreated form, one century ago.
Key words: mental retardation; schizophrenia; cortical complexity; sulcal span index; pervasive developmental disorders Jean –Luc Martinot, INSERM – CEA Research Unit 1000 “NeuroImaging & Psychiatry”, Maison des Adolescents – Maison de Solenn, 97 Bd. De Port Royal, 75014 Paris, France. E-mail: [email protected]
Accepted for publication December 4, 2013
Significant outcomes
• Cortical • •
features in patient data acquired one century ago can be quantified using computerized assessment of surface complexity and folding. Language abilities in severe intellectual disability were associated with regional changes in cortical complexity and folding. Historical brain images « sleeping » in scientific libraries, could gain from modern processing methods.
Limitations
• Intellectual disability patients were compared to dementia praecox group, as no healthy controls were •
available. Two – dimensional images measurements of surface complexity and folding cannot be extrapolated to 3 dimensions.
227
Sandu et al. Introduction
The human brain morphology has attracted the interest from the ancient times, but especially in the 19th century, when the era of cortical localization of cognitive function started (1–3). Macroscopic changes in brain cortex structure were putatively associated with some mental disorders during the first decade of the 20th century through qualitative neuropathology case-report approaches (4), without quantitative measures and small sample statistics still to be invented (5). Currently, analyses derived from structural magnetic resonance imaging (MRI) have confirmed morphometry and volume abnormalities in schizophrenia patients (6–8), and significant changes in gross neuroanatomy with multiple absent, duplicated or misshapen gyri (9–13). These changes in cortical folding features have been measured through sulcal index (14, 15) or gyrification index (GI) (16–18). Mental retardation, a term for intellectual disability with poor cognitive performance used in the early 20th century, is another condition associated with macroscopic changes in the cortex. However little research has been done using MRI to quantify these changes. Lower GI was recently reported in multiple cortical regions including frontal lobes (19). Also, neuro-developmental conditions, such as neuronal migration disorders and autism, which involve both mental retardation and speech and language impairments, have been related empirically to macroscopic alterations in the cortex such as simplified gyral pattern (20), cortical dysplasias (21), polymicrogyria (22, 23) or changes in parietal folding (24). In non-syndromic autism, high rates of visible MRI abnormalities have been reported in the temporal lobes (25). Cortical convolutions have recently been assessed using fractal dimension (FD). FD allows cortical complexity to be described as a single numerical value (26), which can be compared among subjects, even for 2D brain images. The higher the irregularities and more detail in the cortical sheet, the higher the FD value. While discrepant changes in mean FD have been reported in schizophrenia patients – higher (27, 28) or lower (29, 30) than healthy subjects -, there is no study of brain FD in mental retardation, except in Williams Syndrome, where Thompson et al. (31) found a higher FD. While cortex differences have been detected in recent studies of 3D MRI in patients compared with controls, it is unknown whether these modern analysis methods of cortical complexity, or folding features, would have yielded analogous results in 228
patient samples gathered at the dawn of the 20th century. Although modern diagnostic criteria became systematized much later, a high degree of agreement has been shown between Kraepelin’s diagnoses of dementia praecox, and the diagnosis of schizophrenia in modern systems (32). During the first two decades of the 20th century the conception of dementia praecox as a disorder of brain functions or of brain structure, was intensively debated (33). Accordingly, early collections of brains had been performed for neuropathology studies. At the time, a peculiar situation existed in Massachusetts because resources available for large studies of patients’ brains could be found nowhere else in such a number of neuropathology laboratories. Many of these laboratories had firstrate pathologists, with full and reliable case-histories and autopsy protocols, and autopsy material routinely prepared (34). In 1906, the leading pathologist at the Boston Psychopathic Hospital Danvers, E. E. Southard, began the collection of brains, and standardised protocols were developed for the collection and preparation of brain samples. These samples were linked to elaborate written clinical descriptions of the case studies, including symptoms and some quantified data such as brain weight and body size. Innovative at that time, a standardised method to photographically record the brains was set up using a fixed camera position over the brain, and a reference scale for each acquired glass plate. By 1914 Southard reported the collection of 500 brains (35) with an associated photograph collection. However, hospitals where he worked have moved location and traces of the photograph glass collection were lost, except for a monograph series: the ‘Waverly researches in the pathology of the feeble-minded’ (36) that includes brain photographs of each reported case. The collection was gathered from 1914 to 1921. After Southard’s death in 1920, his lifelong collaborator Myrtle May Moore Canavan pursued the publication editing of the Waverly series with AE Taft, until case 50 in 1940. Apart from this monograph, another publication with photographs from the brains of ‘criminal’ patients had been collected from the Bridgewater State Hospital for the Criminally Insane from 1914 to 1919 [described by MM Canavan and L Enseinhardt and photographed by HW Taylor; (37)]. Therefore, two datasets of 50 subjects each including 2D high-quality brain photographs collected with standardised methodology by the same team during the same period of time and including case reports, remain available (50 mental retarda-
1910s’ brains revisited tion cases: Waverley series 1918–1940; 50 forensic cases: Canavan and Enseinhardt 1942) (36, 37). These two datasets represent unique collections including brain photographs from patients with mental retardation or with dementia praecox, acquired well before the psychopharmacology era. Advances in medicine over the past century, the current medical treatments for these conditions (38), the social assistance that permits an early clinical detection and better medical care may induce an anatomo-structural compensation in brain structure for those impairments. These collections have the unique advantage that they represent patient groups with the ‘raw face of the illness’. A major limitation however, is the absence of healthy controls, which will restrict the hypotheses to be tested. Still, two hypotheses were raised and tested, applying modern analyses techniques to these century old samples. First, it is unknown if the cortical complexity (FD) and sulcal measurements would differ between the two diagnostic classes. From the modern literature reviewed above, we hypothesised that the mental retardation group might have lower FD and higher sulcal index than the dementia praecox group, if this latter diagnosis overlaps the present diagnosis of schizophrenia. Second, it is unclear whether impaired speech and language abilities within the mental retardation group might correspond to differences in the FD and sulcal measures, an issue that remains pendant since no report is available even from modern studies. From the literature, we hypothesised that depending on speech abilities, differences within the mental retardation group would be marked in the left hemisphere since it is dominant for language, notably its posterior half which includes the areas around the superior temporal sulcus, i.e. classical speech perception areas. Aims of the study
First, the hypothesis that significant between-group differences in (fractal dimension) FD and sulcal measurements exist was tested in the processed images of mental retardation vs. dementia praecox datasets; and second, differences hypothesised in the left hemisphere were investigated in mental retardation subgroups defined according to speech abilities.
Material and methods Image datasets
Brain images acquisition. Copies of the monographs including sets of individual brain post-
mortem photographs [50 mental retardation cases – Waverley series 1918–1940 (36); 50 forensic cases: Canavan and Enseinhardt 1942 (37)], were localized with the assistance of the reference librarian at Center for History of Medicine, F. Countway Library of Medicine, Boston. Copies were available in the Archives and Special Collections Department of the Medical Research Library of Brooklyn at the State University of New York Downstate Medical Center; and in the historical collections of the New York Academy of Medicine Library. All the pages of the books were photographed with the authorization of the reference librarian and archivist of these libraries, using a digital reflex Canon-EOS-40D Camera equipped with a highresolution ESFS 17–85 mm objective. The camera was fixed and maintained at a constant distance from the pages thanks to a Manfrotto293 tripod, and the pages were illuminated with constant artificial light. Brain image selection. The photographs were displayed in the books as plain page plates for the two groups. The first group of plates – 50 forensic cases – displayed the transverse photographs of the whole brain taken from above, one subject’s brain per plate. The second group of plates – 50 patients with mental retardation – displayed several brain views from one subject per plate; they included transverse photographs of the brain taken from above in all cases, and photographs of the brain taken from right and left sides and from below in most, but not all cases; coronal sections were displayed in a few subjects’ plates only. Only the transverse photographs available for all subjects in both groups were retained for analysis. Subjects image quality control, and groups defined for analysis. The 50 mental retardation case-reports – 35 men, 15 women – (Waverly series 1918–1940) (36) included detailed brain descriptions, as well as short clinical vignettes reporting physical examination, family history, personal development history, school progress, social history, moral examination. Autopsy and brain pathology examination were reported. From the survey of all this material, only a few quantitative variables were exploitable for analysis (mean SD): age 28 ( 13) years, height 146.84 ( 24.71) cm, brain weight 1188.88 ( 312.63) grams, mental age assessed with Binet test (39) 5.13 ( 3.51) years. Other usable variables included histopathological reports graded in tables, and short descriptions of speech abilities allowing subgroups. Hence, subgroups were defined for the present analyses with mental retardation
229
Sandu et al. patients reported as producing comprehensible speech (n = 23), for comparison with patients (n = 12) having either no-or-non-understandable speech. No information as regarding speech was available for one subject. The 50 short clinical reports of forensic cases (Canavan and Eisenhardt 1942) (37), all men, were surveyed in order to collect the demographic variables, brain weights, and diagnoses. Were discarded from the analysis patient cases attributed with various diagnoses (staphyloccus encephalitis, prostate cancer, organic dementia and alcohol abuse, senile dementia and alcohol abuse, arteriosclerotic dementia, alcoholism, general paresis, manic depression, epilepsy) because of their heterogeneity and the scarcity of clinical information provided (an example of discarded brain with frontal atrophy in Fig. 1a). Upon inspection, all forensic cases with a diagnosis of dementia praecox (n = 21, all men, age 53 13 years, brain weight 1440 152 g) remained available for analysis (Table 1). There were no brain pictures of healthy subjects within these series. Photographs quality and excluded datasets. The pictures were inspected to discard non-usable plates. From the group of mental retardation patients ten brain photographs displaying meninges were excluded, as well as two with multiple visible acci-
(a)
(b)
(d)
230
dental cuts, one child with microcephalia (brain weight 400 g and age 5 years Fig. 1b) and one with very peculiar shapes (prominent macrogyria Fig. 1c). Photographs from thirty-six mental retardation cases remained available for analysis (Table 1). Image processing. All photographs were processed blindly with respect to the diagnoses and clinical information until they reached the final stage of processing. Preprocessing. The 2D brain photographs were processed using the Mac-Biophotonics-ImageJ software (www.macbiophotonics.ca/imagej/) in order to perform median local threshold corrections. This step was necessary to correct for the differences in light intensity from the original pictures, as their inspection revealed the light fell differently on the brains, although the original pictures were taken by a photographer positioning the camera in a standardised manner over the brains during a long time period spanning from 1914 to 1921. Instead of using a single light intensity threshold for the whole image, local thresholds were calculated using the median local threshold option available in IMAGEJ. Here, for thresholding, we used the median minus a constant C. Both the radius of the window and the constant C are parameters defined according to the method
(c)
(e)
Fig. 1. Examples of 2D brain photographs: (a) brain with frontal atrophy from dementia praecox dataset (b) child brain with microcephalia (brain weight 400 g and age 5 years) (c) brain with prominent macrogyria from mental retardation dataset (d) illustration of a low fractal dimension cortex (FD = 1.7485) in a subject with mental retardation and (e) a high fractal dimension cortex (FD = 1.8815) in a subject with dementia praecox. Brains from panels a, b and c were discarded from the analysis. Panels d and e exemplify the FD difference: the higher the gyrification, the higher the FD.
1910s’ brains revisited Table 1. Group comparisons for the quantitative variables reported in the original descriptions Mental retardation – intragroup comparison
Dementia praecox vs. mental retardation
Age (yrs) Mental age (years) IQ* Brain weight (grams) Height (cm)
Dementia praecox n = 21 mean SD
Mental retarded n = 36 mean SD
Mann Whitney between group P value, U
No or non understandable speech n = 12 mean SD
53.05 13.52 –
30.42 12.26 4.53 3.38
P < 0.001, U = 84 –
28.58 12.36 2.66 1.90
– 1440.62 152.72
15.9 13.8 1213.75 222.43
– P < 0.001, U = 149
–
149.89 23.86
–
Comprehensible speech n = 23 mean SD
Mann Whitney P value, U
30.83 12.38 5.51 3.59
P = 0.745, U = 128.5 P = 0.016, U = 69
11.08 7.26 1180.75 180.63
18.88 15.81 1237.22 45.54
P = 0.141, U = 95.5 P = 0.294, U = 107.5
143.25 29.85
153.26 20.59
P = 0.278, U = 106.5
*IQ = mental age/chronological age 9 100.
described elsewhere (http://fiji.sc/Auto_Local_ Threshold#Median). To obtain anatomically comparable images of sulci shapes, a range of values were tested for the radius and the constant C; from these resulting images only one was chosen for each original picture. To minimize the subjectivity, this procedure was applied on all images by two different observers, blind to each other results and to the subjects’ status. The range of the chosen parameters was the same, i.e. 32–38 for constant C and 23–29 for the radius. Image segmentation. The brains in the resulting images were parcelled using Adobe Photoshop CS3 Extended toolbox (Adobe Systems, San Jose, CA, USA). In each hemisphere, transverse photographs were divided into two regions, the frontal cortex and the posterior cortex, using the central sulcus as a readily visible landmark on each hemisphere. The preprocessing steps are illustrated in Fig. 2. Calculation of sulcal complexity
Fractal dimension. Fractal dimension was determined with the most widely used method: boxcounting (26) in which the object to be analyzed is covered with boxes, each of side length r. The linear size r of the boxes is increased in each iteration. The number of boxes N(r) necessary to cover the whole object structure is then assumed to vary according to N(r)~(1/r)D where D is the fractal dimension. The measurement at a given scale ignores irregularities of the object at a smaller scale. This refers to the fine structure of the fractals; by decreasing the size of ‘the measuring stick’ it is possible to see more details, thus the number of boxes varies in a different way than when dealing with a smooth or with an Euclidean object. Real fractal objects, however, only have this property over a limited range of scales. The HarFa
software package (http://www.fch.vutbr.cz/lectures /imagesci/includes/harfa_download.inc.php) was used to determine the range over which the correlation coefficient of the linear plot of the number of boxes versus box size in the logarithmic scale was the highest. In the current study, the built-in box counting method from ImageJ was used with increment 1, on the range from 19 to 43 pixels as determined using HarFa. An example comparing low and high cortical complexity, namely fractal dimension is illustrated in Fig. 1, d and e. Sulcal span index. The sulcal span index (SSI) was computed as the ratio between the total sulcal area, represented by the sulcal span in 2D images, and the area of the outer cortex in the respective view (14). The number of black pixels describing the sulci, namely the sulcal span in our case and the whole brain area were calculated using a code written in Python 2.6 (www.python.org). SSI was computed for the transverse photographs including left and right hemisphere, left and right frontal region, and left and right posterior parts of the brain. Adjustment by brain weight and age. Since the brain weight and the age of the subjects in the different groups were not matched, an adjustment by brain weight and by age, of FD and SSI was used to allow for corrected comparisons between groups. The adjustment was performed with a covariance procedure (40):
aFD ¼ FD b1 ½brain weight mean ðbrain weightÞ; where aFD is the adjusted value of FD by brain weight, FD is the uncorrected (native) fractal dimension, b1 is the slope of the brain weight regression on FD and the mean (brain weight) is the sample mean. 231
Sandu et al. (a)
(b)
(c)
(d)
(e)
Fig. 2. Individual 2D photographs processing, in three patients. The processing steps were: (a) original image, (b) original image transposed on white background, (c) whole brain extraction of the 2D sulci pixels using the median local threshold method with the IMAGEJ software, (d) segmentation for hemispheres, and (e) frontal and posterior cortex regions; c, d, e, were taken as input images for computation of fractal dimension and sulcal span index.
The aFD value was adjusted once more by age AFD ¼ aFD b2 ½age mean ðageÞ; where AFD is the adjusted value of aFD by age, aFD is the value obtained from previous formula, b2 is the slope of the age regression on aFD and the mean (age) is the sample mean of age. SSI values were also adjusted by brain weight and age using the same covariance procedure. Statistical analysis
Between-group comparisons were performed to test the a priori hypotheses mentioned in the introduction section: i) the dementia praecox group versus the mental retardation group and ii) the mental retardation patients with no-or-non-understandable speech vs. the mental retardation patients with comprehensible speech. Patient characteristics, as well as FD and SSI in the whole brain, in the hemispheres, and in the four cortical regions, were 232
compared between groups using non-parametric Mann–Whitney tests (PASW Statistics 18 package, www.spss.com). Results
Patients’ characteristics (chronological age, mental age assessed with Binet test and brain weight) are reported in Table 1. Subjects with mental retardation were younger and had lighter brains than those with dementia praecox. The mental retardation patients’ mean mental age was lower in those with the most impaired speech. The mental retardation group had lower FD than the dementia praecox group in the whole brain (Fig. 3a) as well as all brain regions, and higher SSI in left frontal and left posterior regions (Fig. 3b) (Table 2). Also, the right frontal SSI tended to be higher in this group. Within the mental retardation group, the subgroup with no-or-non-understandable speech displayed higher FD in the left hemisphere (Fig. 3c)
1910s’ brains revisited (a)
(b)
* **
Left posterior sulcal span index
Whole brain fractal dimension
Dementia praecox Mental retardation
**
(c)
Dementia praecox Mental retardation
*
(d) Left posterior sulcal span index
Left hemisphere fractal dimension
Fig. 3. (a) plot of whole brain fractal dimension in dementia praecox (green) and mental retardation (magenta) patients, *P = 0.003. (b) plot of left posterior sulcal span index in dementia praecox (green) and mental retardation (magenta) patients, *P = 0.014 **the group comparison remained significant with this subject excluded [dementia praecox (0.158 0.028) vs. mental retardation (0.174 0.024), U = 230, P = 0.020] (c) plot of left hemisphere fractal dimension values in mental retardation patients split according to speech abilities (red – no or non-understandable speech, blue – comprehensible speech), **P = 0.001. (d) plot of left posterior sulcal span index values in mental retardation patients split according to speech abilities (red – no or nonunderstandable speech, blue – comprehensible speech), *P = 0.031 (Table 2).
*
No or non Comprehensible understandable speech speech
and its frontal and posterior regions. This subgroup had a higher SSI in the left posterior region than the subgroup with comprehensible speech (Fig. 3d) (Table 2). Discussion
In these early 20th century brain photographs, the mental retardation group had lower cortical complexity as measured with FD in all brain regions, and a higher SSI in the left posterior cortex than the dementia praecox group. Within the mental retardation group, the subgroup with no-or-nonunderstandable speech had a higher FD in the left hemisphere and a higher SSI in the left posterior cortex than the subgroup with comprehensible speech. The data included not only the post-mortem photographs affordable one century ago, but also the diagnoses as known at the beginning of the 20th century. At that time, the recognition of dementia praecox as an entity was debated, and schizophrenia had just been ‘invented’ (4). However, the duration criterion was already introduced, and paranoid, catatonia and hebephrenia symptoms were included in the wider category of dementia praecox; most symptoms are today part of the DSM- description of schizophrenia. The data studied here were obtained from patients with a long duration of illness, precluding the modern
No or non Comprehensible understandable speech speech
schizophreniform or brief psychotic disorders. Moreover, a study applying modern criteria to Kraepelin’s cases has shown up to 91% agreement in diagnostic (32). The patients’ diagnoses in the current sample had been established by psychiatrists who had either been visitors to Kraepelin’s Heidelberg department (e.g. EE Southard in 1902), or by their disciples. Thus, it is likely that their diagnosis would correspond to current schizophrenia diagnoses. Still, we cannot exclude that a few might fulfill criteria for modern schizoaffective disorder. Given the low mental age reported from the Binet test and their other characteristics, the sample with old diagnoses of ‘feeble-mindness’, would likely fulfil the modern criteria for mental retardation, including IQ below 70, significant limitations in two or more areas of adaptive behavior, and evidence that the limitations became apparent before the age of 18. Cortical fractal dimension assessment
Here, the dementia praecox group had higher FD with respect to the mental retardation group. Large brains had disproportionally more cortical surface than smaller brains (41) and the total folding power showed a positive allometric scale, suggesting that large brains are not simply scaled-up versions of the smaller ones (42). A covariance 233
Sandu et al. Table 2. Dementia praecox and mental retardation subgroups comparisons for fractal dimension and sulcal span index Dementia praecox vs. mental retardation: group comparisons Fractal dimension
Brain region Whole brain Left hemisphere Left frontal Left posterior Right hemisphere Right frontal Right posterior
Dementia praecox n = 21 mean SD 1.775 1.784 1.776 1.688 1.732 1.762 1.687
Mental retarded n = 36 mean SD
0.061 0.044 0.072 0.050 0.071 0.091 0.058
1.724 1.694 1.606 1.683 1.676 1.605 1.653
0.070 0.083 0.099 0.076 0.099 0.123 0.095
Mann Whitney between group P value, U P P P P P P P
= < = = = = =
0.003, U 0.001, U 0.001, U 0.009, U 0.001, U 0.012, U 0.001, U
= = = = = = =
196.5 121.5 180.5 221 173 225.5 181
Dementia praecox n = 21 mean SD 0.165 0.158 0.152 0.158 0.167 0.160 0.168
Sulcal span index Mental retarded n = 36 mean SD
0.024 0.023 0.024 0.028 0.032 0.038 0.034
0.171 0.168 0.166 0.176 0.176 0.176 0.182
0.027 0.029 0.036 0.029 0.034 0.041 0.034
Mann Whitney between group P value, U P P P P P P P
= = = = = = =
0.337, U 0.128, U 0.059, U 0.014, U 0.254, U 0.057, U 0.197, U
= = = = = = =
320 286 264 230 309 263 300
Speech abilities and mental retardation: mental retardation subgroups comparisons Fractal dimension Sulcal span index
Brain region Whole brain Left hemisphere Left frontal Left posterior Right hemisphere Right frontal Right posterior
No or non understandable speech abilities n = 12 mean SD 1.719 1.725 1.619 1.693 1.674 1.595 1.649
0.046 0.048 0.071 0.048 0.069 0.081 0.064
Comprehensible speech n = 23 mean SD 1.687 1.639 1.549 1.638 1.638 1.557 1.615
0.070 0.080 0.093 0.080 0.107 0.133 0.100
Mann Whitney between group P value, U P P P P P P P
= = = = = = =
0.362, U 0.001, U 0.034, U 0.023, U 0.420, U 0.482, U 0.362, U
procedure for the correction of brain weight and age was done in the current study, but the allometric power laws which have recently been developed for healthy people (41, 42) were not applied. Diseases, thought to have neuro-developmental origin, would be expected to generate differences in brain allometry laws also. Increased complexity might imply more cortical surface area (43) and higher number of folds rather than extent of folds (44). Reports on FD in modern schizophrenia MRI datasets are scarce, most focused on the border between grey and white matter or on the brain skeleton (28, 29, 45). Only two reports (27, 30) have used an approach consistent with that of the present study, i.e. focused on the outside border of the cortex or on the external surface of the gyri. Their results may point to a progression of cortical surface deficits in schizophrenia, since they point to higher FD in firstepisode and lower FD in chronic schizophrenia patients, with respect to controls. Low FD in the present mental retardation group is an original finding, as, paradoxically, there is no report of FD in this condition neither with respect to controls nor to other pathological conditions. However, Thompson et al. (31) have reported higher FD in Williams syndrome compared with controls for both hemispheres, but the Williams syndrome differs from the mental retardation sample studied here since it is associated with abnormally preserved language abilities. 234
= = = = = = =
111 46 77 73 114 117 111
No or non understandable speech abilities n = 12 mean SD 0.178 0.173 0.168 0.187 0.185 0.189 0.194
0.019 0.016 0.023 0.020 0.027 0.033 0.031
Comprehensible speech n = 23 mean SD 0.167 0.165 0.167 0.171 0.171 0.175 0.175
0.031 0.034 0.041 0.029 0.036 0.046 0.033
Mann Whitney between group P value, U P P P P P P P
= = = = = = =
0.278, U 0.195, U 0.745, U 0.031, U 0.278, U 0.440, U 0.248, U
= = = = = = =
106 100 128 76 106 115 104
Interestingly, FD appears grossly related to brain development in preterm infant cases (46) and during childhood and adolescence (47). FD values increase from preterm to adult, and remain low in cases of preterm development restrictions. Moreover, Im et al. (48) found a relationship between a complex shape of the cortical surface and intelligence and education. Hence, the present reanalysis of photographs in a mental retardation group would support the hypothesis that mental retardation might result from earlier or more marked developmental restrictions than dementia praecox, and strikingly echoes the initial interpretation of the data in the Waverley series, where post-mortem examination findings were mostly regarded as disclosing ‘examples of arrested development’. Cortical folding determination in 2D transverse photographs
Among the methodological approaches for quantifying the complexity of cortical folding geometry, the shape of the convolutions can be characterized with the gyrification index (GI), defined as the ratio between the total cortical outline and the superficially exposed cortical surface (16, 49, 50) and can also be used for regional determinations (17–19, 51). Recently, the sulcal index was introduced for the measurement of the folds in the whole cortex (14, 15, 44, 52). The global sulcal index is defined for each hemisphere as the percentage ratio
1910s’ brains revisited between the total sulcal area and the outer cortex area. A local sulcal index is defined as the percentage ratio between the area of the labeled sulcus and the outer cortical area. In schizophrenia patients, a lower global sulcal index has been found, and the amount of local sulcal index decrease varied across the cortex (14). Even if several authors labeled both GI and sulcal index under the same generic name of gyrification index as a measurement of the sulcation and gyration of cerebral cortex (44), they are not totally equivalent and their values can be independent to each other. The measurement of GI is focused more on the outer surface of the convoluted cortical sheet i.e. the gyri, while the sulcal index measurements rather focuses on the sulci area, the buried part of the cortical sheet. In 3D images, the sulcal index assesses the surface of the sulci walls, which are dependent on the depth of the sulci. In 2D photographs however, the sulcal measurement could assess only the outer surface of the sulci on the cortex, reflecting the sulcal span, i.e. the ‘opening’ of the sulci at the level of the external part of the cortex (53); in order to avoid any confusion we choose the name sulcal span index (SSI) for our results rather than ‘sulcal index’. The present 2D transverse photographs seemed accessible for the measurements of the SSI since the sulcal span of the sulci can be appreciated by black pixels in the processed images. From its formula, the SSI would be high if the area of sulci is large and combined with a small outer-cortex area. This fits previous descriptions of the mental retardation brain, where the largeness of the sulci can be readily visible from postmortem data or in magnetic resonance images (54), even if the sulci are not deeply buried. The crowns of convolutions (outer areas of sulci) are much more flat –smoothed – in mental retardation group in comparison with other disorders. A small brain also is reflected in a reduced outer surface. In the present mental retardation group, we detected higher SSI in left posterior regions compared to the dementia praecox group consistent with a larger sulcal span. Also, a trend for higher SSI was detected in the mental retardation patients in the frontal regions bilaterally. While there is no previous report on sulcal measurements in mental retardation, this finding might be seen in line with a recent report of generally degraded cognitive performance with wider sulcal span (55). Findings in mental retardation subgroups and speech abilities
Early 20th century mental retardation classification is now largely obsolete, but the patients were
assessed with the first adaptative test (Binet and Simon 1905, translated in United States of America in 1910). The reported mental age, 4.53 ( 3.38) years, of the sample reported to its chronological age, 30.42 ( 12.26) years, yields an IQ of 15.9 ( 13.8) denoting profound to severe retardation, although the concept of mental age was later criticized. Interestingly, the 1905 Binet-Simon test focused on verbal abilities (oral language and vocabulary) (39). These abilities were also briefly described in the mental retardation sample’s original tables, allowing the definition of mental retardation subgroups’ according to their impaired or non-impaired speech abilities. The subjects with no-or-non-understandable speech could have had aphasia, progressive non-fluent aphasia, dyslexia, pervasive developmental disorders, autism or even other categories of speech disabilities since at the beginning of the 20th century the differential diagnosis for different categories were not so contoured like nowadays. Whatever their diagnoses, comparison of brain images in mental retardation subgroups, taking into account the speech abilities were never reported previously. The mental retardation subgroup with noor-non-understandable speech had higher FD and SSI in the left hemisphere, notably in its left posterior part, classically involved in speech perception (56, 57), suggesting that increased irregularities might be a characteristic for such speech disabilities. The only modern study of FD in a language-related pathology, dyslexia (58) highlighted consistently higher FD in the left hemisphere. Limitations
The major limitation is the absence of a healthy control group. The reason that no photographs were available in healthy subjects is mainly because they were mostly hospitalized in general hospitals, precluding their availability for pathology laboratories within asylums. Even if such photographs did exist they were not published and may have been destroyed. In the available MRI studies assessing both FD and sulcal measurements, the case of mental retardation was not treated precluding extrapolation from modern literature. Even if a few articles assessing FD or sulcal measurements are available for schizophrenia, it is difficult to compare their normative data with the current work, where the comparison is made between dementia praecox and mental retardation or within mental retardation subgroups. Only Bonici et al. (59) analysed gyrification indexes comparing patients with mental retardation, schizophrenia and controls. The 235
Sandu et al. results for whole hemispheres classified the schizophrenia group between mental retardation and control group. Also, from a qualitative point of view, the inspection of the transverse photographs in the present mental retardation datasets yielded a number of cases displaying abnormally simplified cortical geometry, at variance with the dementia praecox photographs. On a cautious note, we thus speculate that dementia praecox could be closer to a ‘control’ group than the mental retardation sample. Also, the brain measures were adjusted by brain weight and age, in order to rule out these effects for between-group comparisons. A second limitation is the 2-dimensional nature of photographs and the lack of information in comparison with 3D data. However this limitation, namely the projection of 3D folding in 2D is made under the same technical circumstances, suffering the same transformation for all data and thus it still allows us a comparison between the 2D photographs from the investigated groups with a reduced bias. In conclusion the findings suggest differences in cortical complexity between dementia praecox and mental retardation, and regional variations in mental retardation according to a phenotype such as language abilities. The current post-mortem data are a precious collection even nowadays, since modern studies in profound mental retardation are scarce as the cooperation of patients is impossible for magnetic resonance acquisition. Through revisited analyses, the century-old photographs might gain a ‘second life’ and the findings be used as a reference, as these photographs provide a unique picture of cortex changes in their original, untreated form. Acknowledgements Yanko Roa and Rafik Kheffache are acknowledged for their help in image processing (image segmentation and Python programming, respectively). Jack Eckert, reference librarian, Harvard Medical Library, Francis A. Countway Library of Medicine, Boston, Mass.; Jack E Termine, archivist, Medical Research Library of Brooklyn, State University of New-York, Downstate Medical Center, Brooklyn; Arlene Shaner, assistant curator and reference librarian, the New York Academy of Medicine, New York; are acknowledged for their help in localizing and photographing the original books displaying the image series. Dr. Gordon D. Waiter is acknowledged for reading the manuscript and providing valuable comments.
Declaration of interests
None. References 1. Clarke E, O’Malley CD. The human brain and spinal cord. San Francesco: Norman Publishing, 1996.
236
2. Finger S. Origins of neuroscience: a history of explorations into brain function. Oxford: University Press, 2001. 3. Catani M, Thiebaut De Schotten M. Atlas of human brain connections. Oxford: University Press, 2012. 4. Noll R. American madness: the rise and fall of dementia praecox. Harvard: Harvard University Press, 2011. 5. Box JF. Guinness, gosset, fisher, and small samples. Stat Sci 1987;2:45–52. 6. Levitt JJ, Bobrow L, Lucia D, Srinivasan P. A selective review of volumetric and morphometric imaging in schizophrenia. Curr Top Behav Neurosci 2010;4:243–28. 7. Malhi GS, Lagopoulos J. Making sense of neuroimaging in psychiatry. Acta Psychiatr Scand 2008;117:100–117. 8. Perlini C, Bellani M, Brambilla P. Structural imaging techniques in schizophrenia. Acta Psychiatr Scand 2012;126:235–242. 9. Crow T. Temporolimbic or transcallosal connections: where is the primary lesion in schizophrenia and what is its nature? Schizophr Bull 1997;23:521–523. 10. Green MF. Schizophrenia from a neurocognitive perspective – probing the impenetrable darkness. Boston, London: Allyn and Bacon, 1998. 11. Leonard CM, Kuldau JM, Breier JI, et al. Cumulative effect of anatomical risk factors for schizophrenia: an MRI study. Biol Psychiatry 1999;46:374–382. 12. le Provost JB, Bartres-Faz D, Paillere-Martinot ML, et al. Paracingulate sulcus morphology in men with early-onset schizophrenia. Br J Psychiatry 2003;182: 228–232. 13. Narr KL, Thompson PM, Sharma T, et al. Three-dimensional mapping of gyral shape and cortical surface asymmetries in schizophrenia: gender effects. Am J Psychiatry 2001;158:244–255. 14. Cachia A, Paillere-Martinot ML, Galinowski A, et al. Cortical folding abnormalities in schizophrenia patients with resistant auditory hallucinations. Neuroimage 2008;39: 927–935. 15. Penttila J, Paillere-Martinot ML, Martinot JL, et al. Global and temporal cortical folding in patients with earlyonset schizophrenia. J Am Acad Child Adolesc Psychiatry 2008;10:1125–1132. 16. Kulynych JJ, Luevano LF, Jones DW, Weinberger DR. Cortical abnormality in schizophrenia: an in vivo application of the gyrification index. Biol Psychiatry 1997;41:995– 999. 17. Vogeley K, Schneider-Axmann T, Pfeiffer U, et al. Disturbed gyrification of the prefrontal region in male schizophrenic patients: a morphometric postmortem study. Am J Psychiatry 2000;157:34–39. 18. Vogeley K, Tepest R, Pfeiffer U, et al. Right frontal hypergyria differentiation in affected and unaffected siblings from families multiply affected with schizophrenia: a morphometric mri study. Am J Psychiatry 2001;158:494– 496. 19. Zhang Y, Zhou Y, Yu C, Lin L, Li C, Jiang T. Reduced cortical folding in mental retardation. AJNR Am J Neuroradiol 2010;31:1063–1067. 20. Torniero C, Dalla Bernardina B, Novara F, et al. Dysmorphic features, simplified gyral pattern and 7q11.23 duplication reciprocal to the Williams-Beuren deletion. Eur J Hum Genet 2008;16:880–887. 21. Barkovich AJ, Kjos BO. Nonlissencephalic cortical dysplasias: correlation of imaging findings with clinical deficits. Am J Neuroradiol 1992;13:95–103. 22. Guerrini R, Parrini E. Neuronal migration disorders. Neurobiol Dis 2010;38:154–166.
1910s’ brains revisited 23. Saletti V, Bulgheroni S, D’incerti L, et al. Verbal and gestural communication in children with bilateral perisylvian polymicrogyria. J Child Neurol 2007;22:1090–1098. 24. Kates WR, Ikuta I, Burnette CP. Gyrification patterns in monozygotic twin pairs varying in discordance for autism. Autism Res 2009;2:267–278. 25. Boddaert N, Zibovicius M, Philipe A, et al. MRI findings in 77 children with non-syndromic autistic disorder. PLoS ONE 2009;4:e4415. 26. Mandelbrot BB. Fractals: form, change and dimension. San Francisco: W.H. Freeman and co, 1977. 27. Narr KL, Bilder RM, Kim S, et al. Abnormal gyral complexity in first-episode schizophrenia. Biol Psychiatry 2004;55:859–867. 28. Sandu AL, Rasmussen IA Jr, Lundervold A, et al. Fractal dimension analysis of MR images reveals grey matter structure irregularities in schizophrenia. Comput Med Imaging Graph 2008;32:150–158. 29. Ha TH, Yoon U, Lee KJ, et al. Fractal dimension of cerebral cortical surface in schizophrenia and obsessive-compulsive disorder. Neurosci Lett 2005;384: 172–176. 30. Yotter RA, Nemandic I, Ziegler G, Thompson PM, Gasser C. Local cortical surface complexity maps from spherical harmonic reconstructions. Neuroimage 2011;56:961–973. 31. Thompson PM, Lee AD, Dutton RA, et al. Abnormal cortical complexity and thickness profiles mapped in Williams syndrome. J Neurosci 2005;25:4146–4158. 32. James RL, May PR. Diagnosing schizophrenia: Professor Kraepelin and the research diagnostic criteria. Am J Psychiatry 1981;138:501–504. 33. Southard EE. A study of the dementia praecox group in the light of certain cases showing anomalies or sclerosis in particular brain regions. Am J Insanity 1912;67:119–176. 34. Alford LB. Dr. E. E. Southard’s scientific contributions to psychiatry. An Appreciation After Twenty Years. Am J Psychiatry 1935;92:675–694. 35. Southard EE. On the topographical distribution of cortex lesions and anomalies in dementia praecox, with some account of their functional significance. Am J Insanity 1915; 71:603–671. 36. WAVERLEY researches in the pathology of the feebleminded (1918 — 1940) Research series, cases I to L. Waverley, Massachusetts School for the Feeble-Minded. Editors: [1st series] W. E. Fernald, E. E. Southard and Annie E. Taft; —[2d series] W. E. Fernald, E. E. Southard, Myrtelle M. Canavan, O. J. Raeder and Annie E. Taft; — [3d–4th series] Myrtelle M. Canavan and Annie E. Taft; — [5th series] Myrtelle M. Canavan and Louise Eisenhardt. Boston: The American Academy of Arts and Sciences. 37. Canavan MM, Eisenstadt L. The brains of fifty insane criminals: shapes and patterns. Boston: Massachusetts, 1942. 38. Nielsen J, Munk-Jørgensen P. Changes in hospitalization and suicide rates in schizophrenia. In Tatarelli R, Pompili M, eds. Suicide in schizophrenia. New York, USA: Nova Science Publishers, 2007: 225–237. 39. Binet A, Simon TA. New method for the diagnostic of the intellectual level of abnormals. Annee Psychol 1905; 11:1191–1244. (in french). 40. Fotenos AF, Mintun MA, Snyder AZ, Morris JC, Buckner RL. Brain volume decline in aging. Arch Neurol 2008;65:113–120.
41. Toro R, Perron M, Pike B, et al. Brain size and folding of the human cerebral cortex. Cereb Cortex 2008;18: 2352–2357. 42. Germanaud D, Lefevre J, Toro R, et al. Larger is twistier: spectral analysis of gyrification (SPANGY) applied to adult brain size polymorphism. Neuroimage 2012;63: 1257–1272. 43. Luders E, Narr KL, Thompson PM, et al. Gender differences in cortical complexity. Nat Neurosci 2004;7: 799–800. 44. Mangin JF, Jouvent E, Cachia A. In-vivo measurement of cortical morphology: means and meanings. Curr Opin Neurol 2010;23:359–367. 45. Bulmore E, Brammer M, Harvey I, Persaud R, Murray R, Ron M. Fractal analyses of the boundary between white matter and cerebral cortex in magnetic resonance images: a controlled study of schizophrenic and manic-depressive patients. Psychol Med 1994;24:771–781. 46. Esteban FJ, Padilla N, Sanz-Cortez M, et al. Fractaldimension analysis detects cerebral changes in preterm infants with and without intrauterine growth restriction. Neuroimage 2010;53:1225–1232. 47. Blanton RE, Levit JG, Thompson PM, et al. Mapping cortical asymmetry and complexity patterns in normal children. Psychiatry Res 2001;107:29–43. 48. Im K, Lee JM, Yoon U, et al. Fractal dimension in human cortical surface: multiple regression analysis with cortical thickness, sulcal depth, and folding area. Hum Brain Mapp 2006;27:994–1003. 49. Rilling JK, Insel TR. The primate neocortex in comparative perspective using magnetic resonance imaging. J Hum Evol 1999;37:191–223. 50. Zilles K, Armstrong E, Schleicher A, Kretschmann HJ. The human pattern of gyrification in the cerebral cortex. Anat Embryol (Berl) 1988;179:173–179. 51. Moorhead TW, Harris JM, Standfield AC, et al. Automated computation of the Gyrification Index in prefrontal lobes: methods and comparison with manual implementation. Neuroimage 2006;31:1560–1566. 52. Mangin JF, Riviere D, Cachia A, et al. A framework to study the cortical folding patterns. Neuroimage 2004;23: S129–S138. 53. Kochunov P, Thompson PM, Coyle TR, et al. Relationship among neuroimaging indices of cerebral health during normal aging. Hum Brain Mapp 2008;29:36–45. 54. Decobert F, Grabar S, Merzoug Merzoug V, et al. Unexplained mental retardation: is brain MRI useful? Pediatr Radiol 2005;35:587–596. 55. Liu T, Wen W, Zhu W, et al. The relationship between cortical sulcal variability and cognitive performance in the elderly. Neuroimage 2011;56:865–873. 56. Hugdahl K, Davidson RJ. The asymmetrical brain. Cambridge: MIT Press, 2004. 57. Sommer I, Kahn R. Language lateralisation and psychosis. Cambridge: Cambridge University Press, 2009. 58. Sandu AL, Specht K, Beneventi H, Lundervol A, Hugdahl K. Sex-differences in grey-white matter structure in normal-reading and dyslexic adolescents. Neurosci Lett 2008;438:80–84. 59. Bonnici HM, William T, Moorhead J, et al. Pre-frontal lobe gyrification index in schizophrenia, mental retardation and comorbid groups: an automated study. Neuroimage 2007;35:648–654.
237
© 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd ACTA PSYCHIATRICA SCANDINAVICA
1910s’ brains revisited. Cortical complexity in early 20th century patients with intellectual disability or with dementia praecox Sandu A-L, Paillere Martinot M-L, Artiges E, Martinot J-L. 1910s’ brains revisited. Cortical complexity in early 20th century patients with intellectual disability or with dementia praecox.
A.-L. Sandu1,
M.-L. Paillere Martinot2, E. Artiges1, J.-L. Martinot1,2 Research Unit 1000, Frederic Joliot Hospital Department, INSERM –CEA – Paris Sud University, Orsay, and 2Adolescent Medicine Department, Maison de Solenn, Cochin Hospital, APHP, Paris Descartes University, Paris, France
1
Objective: The idea of cortical surface anomalies in subjects with intellectual disability (mental retardation) and schizophrenia can be traced back to early 20th century qualitative observations. Since it is unknown whether modern quantitative measures of cortical complexity and folding would retrieve those early empirical observations, we measured fractal dimension and sulcal span index in photographs of human brains taken in the 1910’s. Method: Brain photographs were compared between 36 patients with mental retardation and 21 patients with dementia praecox for the fractal dimension and sulcal span index. Also, a mental retardation subgroup with no-or-non-understandable speech (n = 12) was compared with a subgroup with comprehensible speech (n = 23). Results: Mental retardation group had a lower whole-brain fractal dimension than dementia praecox, and a higher sulcal span index in left posterior cortex. The mental retardation subgroup with comprehensible speech had a lower fractal dimension in left hemisphere than the subgroup with no-or-non-understandable speech and a lower sulcal index in left posterior cortex. Conclusion: Measures of cortical complexity and folding suggest differences between mental retardation and dementia praecox, and regional variations according to language abilities in mental retardation. The findings provide a unique picture of cortical surface changes in their original untreated form, one century ago.
Key words: mental retardation; schizophrenia; cortical complexity; sulcal span index; pervasive developmental disorders Jean –Luc Martinot, INSERM – CEA Research Unit 1000 “NeuroImaging & Psychiatry”, Maison des Adolescents – Maison de Solenn, 97 Bd. De Port Royal, 75014 Paris, France. E-mail: [email protected]
Accepted for publication December 4, 2013
Significant outcomes
• Cortical • •
features in patient data acquired one century ago can be quantified using computerized assessment of surface complexity and folding. Language abilities in severe intellectual disability were associated with regional changes in cortical complexity and folding. Historical brain images « sleeping » in scientific libraries, could gain from modern processing methods.
Limitations
• Intellectual disability patients were compared to dementia praecox group, as no healthy controls were •
available. Two – dimensional images measurements of surface complexity and folding cannot be extrapolated to 3 dimensions.
227
Sandu et al. Introduction
The human brain morphology has attracted the interest from the ancient times, but especially in the 19th century, when the era of cortical localization of cognitive function started (1–3). Macroscopic changes in brain cortex structure were putatively associated with some mental disorders during the first decade of the 20th century through qualitative neuropathology case-report approaches (4), without quantitative measures and small sample statistics still to be invented (5). Currently, analyses derived from structural magnetic resonance imaging (MRI) have confirmed morphometry and volume abnormalities in schizophrenia patients (6–8), and significant changes in gross neuroanatomy with multiple absent, duplicated or misshapen gyri (9–13). These changes in cortical folding features have been measured through sulcal index (14, 15) or gyrification index (GI) (16–18). Mental retardation, a term for intellectual disability with poor cognitive performance used in the early 20th century, is another condition associated with macroscopic changes in the cortex. However little research has been done using MRI to quantify these changes. Lower GI was recently reported in multiple cortical regions including frontal lobes (19). Also, neuro-developmental conditions, such as neuronal migration disorders and autism, which involve both mental retardation and speech and language impairments, have been related empirically to macroscopic alterations in the cortex such as simplified gyral pattern (20), cortical dysplasias (21), polymicrogyria (22, 23) or changes in parietal folding (24). In non-syndromic autism, high rates of visible MRI abnormalities have been reported in the temporal lobes (25). Cortical convolutions have recently been assessed using fractal dimension (FD). FD allows cortical complexity to be described as a single numerical value (26), which can be compared among subjects, even for 2D brain images. The higher the irregularities and more detail in the cortical sheet, the higher the FD value. While discrepant changes in mean FD have been reported in schizophrenia patients – higher (27, 28) or lower (29, 30) than healthy subjects -, there is no study of brain FD in mental retardation, except in Williams Syndrome, where Thompson et al. (31) found a higher FD. While cortex differences have been detected in recent studies of 3D MRI in patients compared with controls, it is unknown whether these modern analysis methods of cortical complexity, or folding features, would have yielded analogous results in 228
patient samples gathered at the dawn of the 20th century. Although modern diagnostic criteria became systematized much later, a high degree of agreement has been shown between Kraepelin’s diagnoses of dementia praecox, and the diagnosis of schizophrenia in modern systems (32). During the first two decades of the 20th century the conception of dementia praecox as a disorder of brain functions or of brain structure, was intensively debated (33). Accordingly, early collections of brains had been performed for neuropathology studies. At the time, a peculiar situation existed in Massachusetts because resources available for large studies of patients’ brains could be found nowhere else in such a number of neuropathology laboratories. Many of these laboratories had firstrate pathologists, with full and reliable case-histories and autopsy protocols, and autopsy material routinely prepared (34). In 1906, the leading pathologist at the Boston Psychopathic Hospital Danvers, E. E. Southard, began the collection of brains, and standardised protocols were developed for the collection and preparation of brain samples. These samples were linked to elaborate written clinical descriptions of the case studies, including symptoms and some quantified data such as brain weight and body size. Innovative at that time, a standardised method to photographically record the brains was set up using a fixed camera position over the brain, and a reference scale for each acquired glass plate. By 1914 Southard reported the collection of 500 brains (35) with an associated photograph collection. However, hospitals where he worked have moved location and traces of the photograph glass collection were lost, except for a monograph series: the ‘Waverly researches in the pathology of the feeble-minded’ (36) that includes brain photographs of each reported case. The collection was gathered from 1914 to 1921. After Southard’s death in 1920, his lifelong collaborator Myrtle May Moore Canavan pursued the publication editing of the Waverly series with AE Taft, until case 50 in 1940. Apart from this monograph, another publication with photographs from the brains of ‘criminal’ patients had been collected from the Bridgewater State Hospital for the Criminally Insane from 1914 to 1919 [described by MM Canavan and L Enseinhardt and photographed by HW Taylor; (37)]. Therefore, two datasets of 50 subjects each including 2D high-quality brain photographs collected with standardised methodology by the same team during the same period of time and including case reports, remain available (50 mental retarda-
1910s’ brains revisited tion cases: Waverley series 1918–1940; 50 forensic cases: Canavan and Enseinhardt 1942) (36, 37). These two datasets represent unique collections including brain photographs from patients with mental retardation or with dementia praecox, acquired well before the psychopharmacology era. Advances in medicine over the past century, the current medical treatments for these conditions (38), the social assistance that permits an early clinical detection and better medical care may induce an anatomo-structural compensation in brain structure for those impairments. These collections have the unique advantage that they represent patient groups with the ‘raw face of the illness’. A major limitation however, is the absence of healthy controls, which will restrict the hypotheses to be tested. Still, two hypotheses were raised and tested, applying modern analyses techniques to these century old samples. First, it is unknown if the cortical complexity (FD) and sulcal measurements would differ between the two diagnostic classes. From the modern literature reviewed above, we hypothesised that the mental retardation group might have lower FD and higher sulcal index than the dementia praecox group, if this latter diagnosis overlaps the present diagnosis of schizophrenia. Second, it is unclear whether impaired speech and language abilities within the mental retardation group might correspond to differences in the FD and sulcal measures, an issue that remains pendant since no report is available even from modern studies. From the literature, we hypothesised that depending on speech abilities, differences within the mental retardation group would be marked in the left hemisphere since it is dominant for language, notably its posterior half which includes the areas around the superior temporal sulcus, i.e. classical speech perception areas. Aims of the study
First, the hypothesis that significant between-group differences in (fractal dimension) FD and sulcal measurements exist was tested in the processed images of mental retardation vs. dementia praecox datasets; and second, differences hypothesised in the left hemisphere were investigated in mental retardation subgroups defined according to speech abilities.
Material and methods Image datasets
Brain images acquisition. Copies of the monographs including sets of individual brain post-
mortem photographs [50 mental retardation cases – Waverley series 1918–1940 (36); 50 forensic cases: Canavan and Enseinhardt 1942 (37)], were localized with the assistance of the reference librarian at Center for History of Medicine, F. Countway Library of Medicine, Boston. Copies were available in the Archives and Special Collections Department of the Medical Research Library of Brooklyn at the State University of New York Downstate Medical Center; and in the historical collections of the New York Academy of Medicine Library. All the pages of the books were photographed with the authorization of the reference librarian and archivist of these libraries, using a digital reflex Canon-EOS-40D Camera equipped with a highresolution ESFS 17–85 mm objective. The camera was fixed and maintained at a constant distance from the pages thanks to a Manfrotto293 tripod, and the pages were illuminated with constant artificial light. Brain image selection. The photographs were displayed in the books as plain page plates for the two groups. The first group of plates – 50 forensic cases – displayed the transverse photographs of the whole brain taken from above, one subject’s brain per plate. The second group of plates – 50 patients with mental retardation – displayed several brain views from one subject per plate; they included transverse photographs of the brain taken from above in all cases, and photographs of the brain taken from right and left sides and from below in most, but not all cases; coronal sections were displayed in a few subjects’ plates only. Only the transverse photographs available for all subjects in both groups were retained for analysis. Subjects image quality control, and groups defined for analysis. The 50 mental retardation case-reports – 35 men, 15 women – (Waverly series 1918–1940) (36) included detailed brain descriptions, as well as short clinical vignettes reporting physical examination, family history, personal development history, school progress, social history, moral examination. Autopsy and brain pathology examination were reported. From the survey of all this material, only a few quantitative variables were exploitable for analysis (mean SD): age 28 ( 13) years, height 146.84 ( 24.71) cm, brain weight 1188.88 ( 312.63) grams, mental age assessed with Binet test (39) 5.13 ( 3.51) years. Other usable variables included histopathological reports graded in tables, and short descriptions of speech abilities allowing subgroups. Hence, subgroups were defined for the present analyses with mental retardation
229
Sandu et al. patients reported as producing comprehensible speech (n = 23), for comparison with patients (n = 12) having either no-or-non-understandable speech. No information as regarding speech was available for one subject. The 50 short clinical reports of forensic cases (Canavan and Eisenhardt 1942) (37), all men, were surveyed in order to collect the demographic variables, brain weights, and diagnoses. Were discarded from the analysis patient cases attributed with various diagnoses (staphyloccus encephalitis, prostate cancer, organic dementia and alcohol abuse, senile dementia and alcohol abuse, arteriosclerotic dementia, alcoholism, general paresis, manic depression, epilepsy) because of their heterogeneity and the scarcity of clinical information provided (an example of discarded brain with frontal atrophy in Fig. 1a). Upon inspection, all forensic cases with a diagnosis of dementia praecox (n = 21, all men, age 53 13 years, brain weight 1440 152 g) remained available for analysis (Table 1). There were no brain pictures of healthy subjects within these series. Photographs quality and excluded datasets. The pictures were inspected to discard non-usable plates. From the group of mental retardation patients ten brain photographs displaying meninges were excluded, as well as two with multiple visible acci-
(a)
(b)
(d)
230
dental cuts, one child with microcephalia (brain weight 400 g and age 5 years Fig. 1b) and one with very peculiar shapes (prominent macrogyria Fig. 1c). Photographs from thirty-six mental retardation cases remained available for analysis (Table 1). Image processing. All photographs were processed blindly with respect to the diagnoses and clinical information until they reached the final stage of processing. Preprocessing. The 2D brain photographs were processed using the Mac-Biophotonics-ImageJ software (www.macbiophotonics.ca/imagej/) in order to perform median local threshold corrections. This step was necessary to correct for the differences in light intensity from the original pictures, as their inspection revealed the light fell differently on the brains, although the original pictures were taken by a photographer positioning the camera in a standardised manner over the brains during a long time period spanning from 1914 to 1921. Instead of using a single light intensity threshold for the whole image, local thresholds were calculated using the median local threshold option available in IMAGEJ. Here, for thresholding, we used the median minus a constant C. Both the radius of the window and the constant C are parameters defined according to the method
(c)
(e)
Fig. 1. Examples of 2D brain photographs: (a) brain with frontal atrophy from dementia praecox dataset (b) child brain with microcephalia (brain weight 400 g and age 5 years) (c) brain with prominent macrogyria from mental retardation dataset (d) illustration of a low fractal dimension cortex (FD = 1.7485) in a subject with mental retardation and (e) a high fractal dimension cortex (FD = 1.8815) in a subject with dementia praecox. Brains from panels a, b and c were discarded from the analysis. Panels d and e exemplify the FD difference: the higher the gyrification, the higher the FD.
1910s’ brains revisited Table 1. Group comparisons for the quantitative variables reported in the original descriptions Mental retardation – intragroup comparison
Dementia praecox vs. mental retardation
Age (yrs) Mental age (years) IQ* Brain weight (grams) Height (cm)
Dementia praecox n = 21 mean SD
Mental retarded n = 36 mean SD
Mann Whitney between group P value, U
No or non understandable speech n = 12 mean SD
53.05 13.52 –
30.42 12.26 4.53 3.38
P < 0.001, U = 84 –
28.58 12.36 2.66 1.90
– 1440.62 152.72
15.9 13.8 1213.75 222.43
– P < 0.001, U = 149
–
149.89 23.86
–
Comprehensible speech n = 23 mean SD
Mann Whitney P value, U
30.83 12.38 5.51 3.59
P = 0.745, U = 128.5 P = 0.016, U = 69
11.08 7.26 1180.75 180.63
18.88 15.81 1237.22 45.54
P = 0.141, U = 95.5 P = 0.294, U = 107.5
143.25 29.85
153.26 20.59
P = 0.278, U = 106.5
*IQ = mental age/chronological age 9 100.
described elsewhere (http://fiji.sc/Auto_Local_ Threshold#Median). To obtain anatomically comparable images of sulci shapes, a range of values were tested for the radius and the constant C; from these resulting images only one was chosen for each original picture. To minimize the subjectivity, this procedure was applied on all images by two different observers, blind to each other results and to the subjects’ status. The range of the chosen parameters was the same, i.e. 32–38 for constant C and 23–29 for the radius. Image segmentation. The brains in the resulting images were parcelled using Adobe Photoshop CS3 Extended toolbox (Adobe Systems, San Jose, CA, USA). In each hemisphere, transverse photographs were divided into two regions, the frontal cortex and the posterior cortex, using the central sulcus as a readily visible landmark on each hemisphere. The preprocessing steps are illustrated in Fig. 2. Calculation of sulcal complexity
Fractal dimension. Fractal dimension was determined with the most widely used method: boxcounting (26) in which the object to be analyzed is covered with boxes, each of side length r. The linear size r of the boxes is increased in each iteration. The number of boxes N(r) necessary to cover the whole object structure is then assumed to vary according to N(r)~(1/r)D where D is the fractal dimension. The measurement at a given scale ignores irregularities of the object at a smaller scale. This refers to the fine structure of the fractals; by decreasing the size of ‘the measuring stick’ it is possible to see more details, thus the number of boxes varies in a different way than when dealing with a smooth or with an Euclidean object. Real fractal objects, however, only have this property over a limited range of scales. The HarFa
software package (http://www.fch.vutbr.cz/lectures /imagesci/includes/harfa_download.inc.php) was used to determine the range over which the correlation coefficient of the linear plot of the number of boxes versus box size in the logarithmic scale was the highest. In the current study, the built-in box counting method from ImageJ was used with increment 1, on the range from 19 to 43 pixels as determined using HarFa. An example comparing low and high cortical complexity, namely fractal dimension is illustrated in Fig. 1, d and e. Sulcal span index. The sulcal span index (SSI) was computed as the ratio between the total sulcal area, represented by the sulcal span in 2D images, and the area of the outer cortex in the respective view (14). The number of black pixels describing the sulci, namely the sulcal span in our case and the whole brain area were calculated using a code written in Python 2.6 (www.python.org). SSI was computed for the transverse photographs including left and right hemisphere, left and right frontal region, and left and right posterior parts of the brain. Adjustment by brain weight and age. Since the brain weight and the age of the subjects in the different groups were not matched, an adjustment by brain weight and by age, of FD and SSI was used to allow for corrected comparisons between groups. The adjustment was performed with a covariance procedure (40):
aFD ¼ FD b1 ½brain weight mean ðbrain weightÞ; where aFD is the adjusted value of FD by brain weight, FD is the uncorrected (native) fractal dimension, b1 is the slope of the brain weight regression on FD and the mean (brain weight) is the sample mean. 231
Sandu et al. (a)
(b)
(c)
(d)
(e)
Fig. 2. Individual 2D photographs processing, in three patients. The processing steps were: (a) original image, (b) original image transposed on white background, (c) whole brain extraction of the 2D sulci pixels using the median local threshold method with the IMAGEJ software, (d) segmentation for hemispheres, and (e) frontal and posterior cortex regions; c, d, e, were taken as input images for computation of fractal dimension and sulcal span index.
The aFD value was adjusted once more by age AFD ¼ aFD b2 ½age mean ðageÞ; where AFD is the adjusted value of aFD by age, aFD is the value obtained from previous formula, b2 is the slope of the age regression on aFD and the mean (age) is the sample mean of age. SSI values were also adjusted by brain weight and age using the same covariance procedure. Statistical analysis
Between-group comparisons were performed to test the a priori hypotheses mentioned in the introduction section: i) the dementia praecox group versus the mental retardation group and ii) the mental retardation patients with no-or-non-understandable speech vs. the mental retardation patients with comprehensible speech. Patient characteristics, as well as FD and SSI in the whole brain, in the hemispheres, and in the four cortical regions, were 232
compared between groups using non-parametric Mann–Whitney tests (PASW Statistics 18 package, www.spss.com). Results
Patients’ characteristics (chronological age, mental age assessed with Binet test and brain weight) are reported in Table 1. Subjects with mental retardation were younger and had lighter brains than those with dementia praecox. The mental retardation patients’ mean mental age was lower in those with the most impaired speech. The mental retardation group had lower FD than the dementia praecox group in the whole brain (Fig. 3a) as well as all brain regions, and higher SSI in left frontal and left posterior regions (Fig. 3b) (Table 2). Also, the right frontal SSI tended to be higher in this group. Within the mental retardation group, the subgroup with no-or-non-understandable speech displayed higher FD in the left hemisphere (Fig. 3c)
1910s’ brains revisited (a)
(b)
* **
Left posterior sulcal span index
Whole brain fractal dimension
Dementia praecox Mental retardation
**
(c)
Dementia praecox Mental retardation
*
(d) Left posterior sulcal span index
Left hemisphere fractal dimension
Fig. 3. (a) plot of whole brain fractal dimension in dementia praecox (green) and mental retardation (magenta) patients, *P = 0.003. (b) plot of left posterior sulcal span index in dementia praecox (green) and mental retardation (magenta) patients, *P = 0.014 **the group comparison remained significant with this subject excluded [dementia praecox (0.158 0.028) vs. mental retardation (0.174 0.024), U = 230, P = 0.020] (c) plot of left hemisphere fractal dimension values in mental retardation patients split according to speech abilities (red – no or non-understandable speech, blue – comprehensible speech), **P = 0.001. (d) plot of left posterior sulcal span index values in mental retardation patients split according to speech abilities (red – no or nonunderstandable speech, blue – comprehensible speech), *P = 0.031 (Table 2).
*
No or non Comprehensible understandable speech speech
and its frontal and posterior regions. This subgroup had a higher SSI in the left posterior region than the subgroup with comprehensible speech (Fig. 3d) (Table 2). Discussion
In these early 20th century brain photographs, the mental retardation group had lower cortical complexity as measured with FD in all brain regions, and a higher SSI in the left posterior cortex than the dementia praecox group. Within the mental retardation group, the subgroup with no-or-nonunderstandable speech had a higher FD in the left hemisphere and a higher SSI in the left posterior cortex than the subgroup with comprehensible speech. The data included not only the post-mortem photographs affordable one century ago, but also the diagnoses as known at the beginning of the 20th century. At that time, the recognition of dementia praecox as an entity was debated, and schizophrenia had just been ‘invented’ (4). However, the duration criterion was already introduced, and paranoid, catatonia and hebephrenia symptoms were included in the wider category of dementia praecox; most symptoms are today part of the DSM- description of schizophrenia. The data studied here were obtained from patients with a long duration of illness, precluding the modern
No or non Comprehensible understandable speech speech
schizophreniform or brief psychotic disorders. Moreover, a study applying modern criteria to Kraepelin’s cases has shown up to 91% agreement in diagnostic (32). The patients’ diagnoses in the current sample had been established by psychiatrists who had either been visitors to Kraepelin’s Heidelberg department (e.g. EE Southard in 1902), or by their disciples. Thus, it is likely that their diagnosis would correspond to current schizophrenia diagnoses. Still, we cannot exclude that a few might fulfill criteria for modern schizoaffective disorder. Given the low mental age reported from the Binet test and their other characteristics, the sample with old diagnoses of ‘feeble-mindness’, would likely fulfil the modern criteria for mental retardation, including IQ below 70, significant limitations in two or more areas of adaptive behavior, and evidence that the limitations became apparent before the age of 18. Cortical fractal dimension assessment
Here, the dementia praecox group had higher FD with respect to the mental retardation group. Large brains had disproportionally more cortical surface than smaller brains (41) and the total folding power showed a positive allometric scale, suggesting that large brains are not simply scaled-up versions of the smaller ones (42). A covariance 233
Sandu et al. Table 2. Dementia praecox and mental retardation subgroups comparisons for fractal dimension and sulcal span index Dementia praecox vs. mental retardation: group comparisons Fractal dimension
Brain region Whole brain Left hemisphere Left frontal Left posterior Right hemisphere Right frontal Right posterior
Dementia praecox n = 21 mean SD 1.775 1.784 1.776 1.688 1.732 1.762 1.687
Mental retarded n = 36 mean SD
0.061 0.044 0.072 0.050 0.071 0.091 0.058
1.724 1.694 1.606 1.683 1.676 1.605 1.653
0.070 0.083 0.099 0.076 0.099 0.123 0.095
Mann Whitney between group P value, U P P P P P P P
= < = = = = =
0.003, U 0.001, U 0.001, U 0.009, U 0.001, U 0.012, U 0.001, U
= = = = = = =
196.5 121.5 180.5 221 173 225.5 181
Dementia praecox n = 21 mean SD 0.165 0.158 0.152 0.158 0.167 0.160 0.168
Sulcal span index Mental retarded n = 36 mean SD
0.024 0.023 0.024 0.028 0.032 0.038 0.034
0.171 0.168 0.166 0.176 0.176 0.176 0.182
0.027 0.029 0.036 0.029 0.034 0.041 0.034
Mann Whitney between group P value, U P P P P P P P
= = = = = = =
0.337, U 0.128, U 0.059, U 0.014, U 0.254, U 0.057, U 0.197, U
= = = = = = =
320 286 264 230 309 263 300
Speech abilities and mental retardation: mental retardation subgroups comparisons Fractal dimension Sulcal span index
Brain region Whole brain Left hemisphere Left frontal Left posterior Right hemisphere Right frontal Right posterior
No or non understandable speech abilities n = 12 mean SD 1.719 1.725 1.619 1.693 1.674 1.595 1.649
0.046 0.048 0.071 0.048 0.069 0.081 0.064
Comprehensible speech n = 23 mean SD 1.687 1.639 1.549 1.638 1.638 1.557 1.615
0.070 0.080 0.093 0.080 0.107 0.133 0.100
Mann Whitney between group P value, U P P P P P P P
= = = = = = =
0.362, U 0.001, U 0.034, U 0.023, U 0.420, U 0.482, U 0.362, U
procedure for the correction of brain weight and age was done in the current study, but the allometric power laws which have recently been developed for healthy people (41, 42) were not applied. Diseases, thought to have neuro-developmental origin, would be expected to generate differences in brain allometry laws also. Increased complexity might imply more cortical surface area (43) and higher number of folds rather than extent of folds (44). Reports on FD in modern schizophrenia MRI datasets are scarce, most focused on the border between grey and white matter or on the brain skeleton (28, 29, 45). Only two reports (27, 30) have used an approach consistent with that of the present study, i.e. focused on the outside border of the cortex or on the external surface of the gyri. Their results may point to a progression of cortical surface deficits in schizophrenia, since they point to higher FD in firstepisode and lower FD in chronic schizophrenia patients, with respect to controls. Low FD in the present mental retardation group is an original finding, as, paradoxically, there is no report of FD in this condition neither with respect to controls nor to other pathological conditions. However, Thompson et al. (31) have reported higher FD in Williams syndrome compared with controls for both hemispheres, but the Williams syndrome differs from the mental retardation sample studied here since it is associated with abnormally preserved language abilities. 234
= = = = = = =
111 46 77 73 114 117 111
No or non understandable speech abilities n = 12 mean SD 0.178 0.173 0.168 0.187 0.185 0.189 0.194
0.019 0.016 0.023 0.020 0.027 0.033 0.031
Comprehensible speech n = 23 mean SD 0.167 0.165 0.167 0.171 0.171 0.175 0.175
0.031 0.034 0.041 0.029 0.036 0.046 0.033
Mann Whitney between group P value, U P P P P P P P
= = = = = = =
0.278, U 0.195, U 0.745, U 0.031, U 0.278, U 0.440, U 0.248, U
= = = = = = =
106 100 128 76 106 115 104
Interestingly, FD appears grossly related to brain development in preterm infant cases (46) and during childhood and adolescence (47). FD values increase from preterm to adult, and remain low in cases of preterm development restrictions. Moreover, Im et al. (48) found a relationship between a complex shape of the cortical surface and intelligence and education. Hence, the present reanalysis of photographs in a mental retardation group would support the hypothesis that mental retardation might result from earlier or more marked developmental restrictions than dementia praecox, and strikingly echoes the initial interpretation of the data in the Waverley series, where post-mortem examination findings were mostly regarded as disclosing ‘examples of arrested development’. Cortical folding determination in 2D transverse photographs
Among the methodological approaches for quantifying the complexity of cortical folding geometry, the shape of the convolutions can be characterized with the gyrification index (GI), defined as the ratio between the total cortical outline and the superficially exposed cortical surface (16, 49, 50) and can also be used for regional determinations (17–19, 51). Recently, the sulcal index was introduced for the measurement of the folds in the whole cortex (14, 15, 44, 52). The global sulcal index is defined for each hemisphere as the percentage ratio
1910s’ brains revisited between the total sulcal area and the outer cortex area. A local sulcal index is defined as the percentage ratio between the area of the labeled sulcus and the outer cortical area. In schizophrenia patients, a lower global sulcal index has been found, and the amount of local sulcal index decrease varied across the cortex (14). Even if several authors labeled both GI and sulcal index under the same generic name of gyrification index as a measurement of the sulcation and gyration of cerebral cortex (44), they are not totally equivalent and their values can be independent to each other. The measurement of GI is focused more on the outer surface of the convoluted cortical sheet i.e. the gyri, while the sulcal index measurements rather focuses on the sulci area, the buried part of the cortical sheet. In 3D images, the sulcal index assesses the surface of the sulci walls, which are dependent on the depth of the sulci. In 2D photographs however, the sulcal measurement could assess only the outer surface of the sulci on the cortex, reflecting the sulcal span, i.e. the ‘opening’ of the sulci at the level of the external part of the cortex (53); in order to avoid any confusion we choose the name sulcal span index (SSI) for our results rather than ‘sulcal index’. The present 2D transverse photographs seemed accessible for the measurements of the SSI since the sulcal span of the sulci can be appreciated by black pixels in the processed images. From its formula, the SSI would be high if the area of sulci is large and combined with a small outer-cortex area. This fits previous descriptions of the mental retardation brain, where the largeness of the sulci can be readily visible from postmortem data or in magnetic resonance images (54), even if the sulci are not deeply buried. The crowns of convolutions (outer areas of sulci) are much more flat –smoothed – in mental retardation group in comparison with other disorders. A small brain also is reflected in a reduced outer surface. In the present mental retardation group, we detected higher SSI in left posterior regions compared to the dementia praecox group consistent with a larger sulcal span. Also, a trend for higher SSI was detected in the mental retardation patients in the frontal regions bilaterally. While there is no previous report on sulcal measurements in mental retardation, this finding might be seen in line with a recent report of generally degraded cognitive performance with wider sulcal span (55). Findings in mental retardation subgroups and speech abilities
Early 20th century mental retardation classification is now largely obsolete, but the patients were
assessed with the first adaptative test (Binet and Simon 1905, translated in United States of America in 1910). The reported mental age, 4.53 ( 3.38) years, of the sample reported to its chronological age, 30.42 ( 12.26) years, yields an IQ of 15.9 ( 13.8) denoting profound to severe retardation, although the concept of mental age was later criticized. Interestingly, the 1905 Binet-Simon test focused on verbal abilities (oral language and vocabulary) (39). These abilities were also briefly described in the mental retardation sample’s original tables, allowing the definition of mental retardation subgroups’ according to their impaired or non-impaired speech abilities. The subjects with no-or-non-understandable speech could have had aphasia, progressive non-fluent aphasia, dyslexia, pervasive developmental disorders, autism or even other categories of speech disabilities since at the beginning of the 20th century the differential diagnosis for different categories were not so contoured like nowadays. Whatever their diagnoses, comparison of brain images in mental retardation subgroups, taking into account the speech abilities were never reported previously. The mental retardation subgroup with noor-non-understandable speech had higher FD and SSI in the left hemisphere, notably in its left posterior part, classically involved in speech perception (56, 57), suggesting that increased irregularities might be a characteristic for such speech disabilities. The only modern study of FD in a language-related pathology, dyslexia (58) highlighted consistently higher FD in the left hemisphere. Limitations
The major limitation is the absence of a healthy control group. The reason that no photographs were available in healthy subjects is mainly because they were mostly hospitalized in general hospitals, precluding their availability for pathology laboratories within asylums. Even if such photographs did exist they were not published and may have been destroyed. In the available MRI studies assessing both FD and sulcal measurements, the case of mental retardation was not treated precluding extrapolation from modern literature. Even if a few articles assessing FD or sulcal measurements are available for schizophrenia, it is difficult to compare their normative data with the current work, where the comparison is made between dementia praecox and mental retardation or within mental retardation subgroups. Only Bonici et al. (59) analysed gyrification indexes comparing patients with mental retardation, schizophrenia and controls. The 235
Sandu et al. results for whole hemispheres classified the schizophrenia group between mental retardation and control group. Also, from a qualitative point of view, the inspection of the transverse photographs in the present mental retardation datasets yielded a number of cases displaying abnormally simplified cortical geometry, at variance with the dementia praecox photographs. On a cautious note, we thus speculate that dementia praecox could be closer to a ‘control’ group than the mental retardation sample. Also, the brain measures were adjusted by brain weight and age, in order to rule out these effects for between-group comparisons. A second limitation is the 2-dimensional nature of photographs and the lack of information in comparison with 3D data. However this limitation, namely the projection of 3D folding in 2D is made under the same technical circumstances, suffering the same transformation for all data and thus it still allows us a comparison between the 2D photographs from the investigated groups with a reduced bias. In conclusion the findings suggest differences in cortical complexity between dementia praecox and mental retardation, and regional variations in mental retardation according to a phenotype such as language abilities. The current post-mortem data are a precious collection even nowadays, since modern studies in profound mental retardation are scarce as the cooperation of patients is impossible for magnetic resonance acquisition. Through revisited analyses, the century-old photographs might gain a ‘second life’ and the findings be used as a reference, as these photographs provide a unique picture of cortex changes in their original, untreated form. Acknowledgements Yanko Roa and Rafik Kheffache are acknowledged for their help in image processing (image segmentation and Python programming, respectively). Jack Eckert, reference librarian, Harvard Medical Library, Francis A. Countway Library of Medicine, Boston, Mass.; Jack E Termine, archivist, Medical Research Library of Brooklyn, State University of New-York, Downstate Medical Center, Brooklyn; Arlene Shaner, assistant curator and reference librarian, the New York Academy of Medicine, New York; are acknowledged for their help in localizing and photographing the original books displaying the image series. Dr. Gordon D. Waiter is acknowledged for reading the manuscript and providing valuable comments.
Declaration of interests
None. References 1. Clarke E, O’Malley CD. The human brain and spinal cord. San Francesco: Norman Publishing, 1996.
236
2. Finger S. Origins of neuroscience: a history of explorations into brain function. Oxford: University Press, 2001. 3. Catani M, Thiebaut De Schotten M. Atlas of human brain connections. Oxford: University Press, 2012. 4. Noll R. American madness: the rise and fall of dementia praecox. Harvard: Harvard University Press, 2011. 5. Box JF. Guinness, gosset, fisher, and small samples. Stat Sci 1987;2:45–52. 6. Levitt JJ, Bobrow L, Lucia D, Srinivasan P. A selective review of volumetric and morphometric imaging in schizophrenia. Curr Top Behav Neurosci 2010;4:243–28. 7. Malhi GS, Lagopoulos J. Making sense of neuroimaging in psychiatry. Acta Psychiatr Scand 2008;117:100–117. 8. Perlini C, Bellani M, Brambilla P. Structural imaging techniques in schizophrenia. Acta Psychiatr Scand 2012;126:235–242. 9. Crow T. Temporolimbic or transcallosal connections: where is the primary lesion in schizophrenia and what is its nature? Schizophr Bull 1997;23:521–523. 10. Green MF. Schizophrenia from a neurocognitive perspective – probing the impenetrable darkness. Boston, London: Allyn and Bacon, 1998. 11. Leonard CM, Kuldau JM, Breier JI, et al. Cumulative effect of anatomical risk factors for schizophrenia: an MRI study. Biol Psychiatry 1999;46:374–382. 12. le Provost JB, Bartres-Faz D, Paillere-Martinot ML, et al. Paracingulate sulcus morphology in men with early-onset schizophrenia. Br J Psychiatry 2003;182: 228–232. 13. Narr KL, Thompson PM, Sharma T, et al. Three-dimensional mapping of gyral shape and cortical surface asymmetries in schizophrenia: gender effects. Am J Psychiatry 2001;158:244–255. 14. Cachia A, Paillere-Martinot ML, Galinowski A, et al. Cortical folding abnormalities in schizophrenia patients with resistant auditory hallucinations. Neuroimage 2008;39: 927–935. 15. Penttila J, Paillere-Martinot ML, Martinot JL, et al. Global and temporal cortical folding in patients with earlyonset schizophrenia. J Am Acad Child Adolesc Psychiatry 2008;10:1125–1132. 16. Kulynych JJ, Luevano LF, Jones DW, Weinberger DR. Cortical abnormality in schizophrenia: an in vivo application of the gyrification index. Biol Psychiatry 1997;41:995– 999. 17. Vogeley K, Schneider-Axmann T, Pfeiffer U, et al. Disturbed gyrification of the prefrontal region in male schizophrenic patients: a morphometric postmortem study. Am J Psychiatry 2000;157:34–39. 18. Vogeley K, Tepest R, Pfeiffer U, et al. Right frontal hypergyria differentiation in affected and unaffected siblings from families multiply affected with schizophrenia: a morphometric mri study. Am J Psychiatry 2001;158:494– 496. 19. Zhang Y, Zhou Y, Yu C, Lin L, Li C, Jiang T. Reduced cortical folding in mental retardation. AJNR Am J Neuroradiol 2010;31:1063–1067. 20. Torniero C, Dalla Bernardina B, Novara F, et al. Dysmorphic features, simplified gyral pattern and 7q11.23 duplication reciprocal to the Williams-Beuren deletion. Eur J Hum Genet 2008;16:880–887. 21. Barkovich AJ, Kjos BO. Nonlissencephalic cortical dysplasias: correlation of imaging findings with clinical deficits. Am J Neuroradiol 1992;13:95–103. 22. Guerrini R, Parrini E. Neuronal migration disorders. Neurobiol Dis 2010;38:154–166.
1910s’ brains revisited 23. Saletti V, Bulgheroni S, D’incerti L, et al. Verbal and gestural communication in children with bilateral perisylvian polymicrogyria. J Child Neurol 2007;22:1090–1098. 24. Kates WR, Ikuta I, Burnette CP. Gyrification patterns in monozygotic twin pairs varying in discordance for autism. Autism Res 2009;2:267–278. 25. Boddaert N, Zibovicius M, Philipe A, et al. MRI findings in 77 children with non-syndromic autistic disorder. PLoS ONE 2009;4:e4415. 26. Mandelbrot BB. Fractals: form, change and dimension. San Francisco: W.H. Freeman and co, 1977. 27. Narr KL, Bilder RM, Kim S, et al. Abnormal gyral complexity in first-episode schizophrenia. Biol Psychiatry 2004;55:859–867. 28. Sandu AL, Rasmussen IA Jr, Lundervold A, et al. Fractal dimension analysis of MR images reveals grey matter structure irregularities in schizophrenia. Comput Med Imaging Graph 2008;32:150–158. 29. Ha TH, Yoon U, Lee KJ, et al. Fractal dimension of cerebral cortical surface in schizophrenia and obsessive-compulsive disorder. Neurosci Lett 2005;384: 172–176. 30. Yotter RA, Nemandic I, Ziegler G, Thompson PM, Gasser C. Local cortical surface complexity maps from spherical harmonic reconstructions. Neuroimage 2011;56:961–973. 31. Thompson PM, Lee AD, Dutton RA, et al. Abnormal cortical complexity and thickness profiles mapped in Williams syndrome. J Neurosci 2005;25:4146–4158. 32. James RL, May PR. Diagnosing schizophrenia: Professor Kraepelin and the research diagnostic criteria. Am J Psychiatry 1981;138:501–504. 33. Southard EE. A study of the dementia praecox group in the light of certain cases showing anomalies or sclerosis in particular brain regions. Am J Insanity 1912;67:119–176. 34. Alford LB. Dr. E. E. Southard’s scientific contributions to psychiatry. An Appreciation After Twenty Years. Am J Psychiatry 1935;92:675–694. 35. Southard EE. On the topographical distribution of cortex lesions and anomalies in dementia praecox, with some account of their functional significance. Am J Insanity 1915; 71:603–671. 36. WAVERLEY researches in the pathology of the feebleminded (1918 — 1940) Research series, cases I to L. Waverley, Massachusetts School for the Feeble-Minded. Editors: [1st series] W. E. Fernald, E. E. Southard and Annie E. Taft; —[2d series] W. E. Fernald, E. E. Southard, Myrtelle M. Canavan, O. J. Raeder and Annie E. Taft; — [3d–4th series] Myrtelle M. Canavan and Annie E. Taft; — [5th series] Myrtelle M. Canavan and Louise Eisenhardt. Boston: The American Academy of Arts and Sciences. 37. Canavan MM, Eisenstadt L. The brains of fifty insane criminals: shapes and patterns. Boston: Massachusetts, 1942. 38. Nielsen J, Munk-Jørgensen P. Changes in hospitalization and suicide rates in schizophrenia. In Tatarelli R, Pompili M, eds. Suicide in schizophrenia. New York, USA: Nova Science Publishers, 2007: 225–237. 39. Binet A, Simon TA. New method for the diagnostic of the intellectual level of abnormals. Annee Psychol 1905; 11:1191–1244. (in french). 40. Fotenos AF, Mintun MA, Snyder AZ, Morris JC, Buckner RL. Brain volume decline in aging. Arch Neurol 2008;65:113–120.
41. Toro R, Perron M, Pike B, et al. Brain size and folding of the human cerebral cortex. Cereb Cortex 2008;18: 2352–2357. 42. Germanaud D, Lefevre J, Toro R, et al. Larger is twistier: spectral analysis of gyrification (SPANGY) applied to adult brain size polymorphism. Neuroimage 2012;63: 1257–1272. 43. Luders E, Narr KL, Thompson PM, et al. Gender differences in cortical complexity. Nat Neurosci 2004;7: 799–800. 44. Mangin JF, Jouvent E, Cachia A. In-vivo measurement of cortical morphology: means and meanings. Curr Opin Neurol 2010;23:359–367. 45. Bulmore E, Brammer M, Harvey I, Persaud R, Murray R, Ron M. Fractal analyses of the boundary between white matter and cerebral cortex in magnetic resonance images: a controlled study of schizophrenic and manic-depressive patients. Psychol Med 1994;24:771–781. 46. Esteban FJ, Padilla N, Sanz-Cortez M, et al. Fractaldimension analysis detects cerebral changes in preterm infants with and without intrauterine growth restriction. Neuroimage 2010;53:1225–1232. 47. Blanton RE, Levit JG, Thompson PM, et al. Mapping cortical asymmetry and complexity patterns in normal children. Psychiatry Res 2001;107:29–43. 48. Im K, Lee JM, Yoon U, et al. Fractal dimension in human cortical surface: multiple regression analysis with cortical thickness, sulcal depth, and folding area. Hum Brain Mapp 2006;27:994–1003. 49. Rilling JK, Insel TR. The primate neocortex in comparative perspective using magnetic resonance imaging. J Hum Evol 1999;37:191–223. 50. Zilles K, Armstrong E, Schleicher A, Kretschmann HJ. The human pattern of gyrification in the cerebral cortex. Anat Embryol (Berl) 1988;179:173–179. 51. Moorhead TW, Harris JM, Standfield AC, et al. Automated computation of the Gyrification Index in prefrontal lobes: methods and comparison with manual implementation. Neuroimage 2006;31:1560–1566. 52. Mangin JF, Riviere D, Cachia A, et al. A framework to study the cortical folding patterns. Neuroimage 2004;23: S129–S138. 53. Kochunov P, Thompson PM, Coyle TR, et al. Relationship among neuroimaging indices of cerebral health during normal aging. Hum Brain Mapp 2008;29:36–45. 54. Decobert F, Grabar S, Merzoug Merzoug V, et al. Unexplained mental retardation: is brain MRI useful? Pediatr Radiol 2005;35:587–596. 55. Liu T, Wen W, Zhu W, et al. The relationship between cortical sulcal variability and cognitive performance in the elderly. Neuroimage 2011;56:865–873. 56. Hugdahl K, Davidson RJ. The asymmetrical brain. Cambridge: MIT Press, 2004. 57. Sommer I, Kahn R. Language lateralisation and psychosis. Cambridge: Cambridge University Press, 2009. 58. Sandu AL, Specht K, Beneventi H, Lundervol A, Hugdahl K. Sex-differences in grey-white matter structure in normal-reading and dyslexic adolescents. Neurosci Lett 2008;438:80–84. 59. Bonnici HM, William T, Moorhead J, et al. Pre-frontal lobe gyrification index in schizophrenia, mental retardation and comorbid groups: an automated study. Neuroimage 2007;35:648–654.
237
