Eur Arch Otorhinolaryngol (1991) 248 : 326-330
European Archives of
Oto-RhinoLaryngology © Springer-Verlag1991
Digital image analysis of radio-opacities in the paranasal sinuses using computed radiography H. Suzaki 1, K. Abbey 2, and Y. Nomura 1 1Department of Otolaryngology, University of Tokyo, Tokyo, Japan 2Department of Otolaryngology, Musashino Red Cross Hospital, Tokyo, Japan Received October 18, 1990 / Accepted November 6, 1990
Summary. A c o m p u t e d radiograph system (Toshiba, model TCR-201) was used to investigate the digital image analysis of radio-opacity of the paranasal sinuses. The results of the preliminary p h a n t o m examination for evaluating the exposure technique indicated that the tube voltage should be kept constant. By using conventional radiographic images and tomographic images of normals and cases of sinusitis, the quantum values (Qvalues), Q-value profiles, and Q-value histograms of radio-opacities in the paranasal sinuses were assessed statistically. Findings demonstrated that radio-opacities in the paranasal sinuses could be evaluated quantitatively by these digital image analyses. Key words: C o m p u t e d radiography - Digital image analysis - Paranasal sinuses - Sinus opacifications
Introduction Radiological evaluation is indispensable for the diagnosis of diseases of the paranasal sinuses. H o w e v e r , radio-opacity of a sinus cavity is generally evaluated visually by an examiner. Such evaluations are not objective because of fluctuations in the radiographic density of each X-ray film, depending on various technical conditions in the regular film-screen combination system. Recent advances in m o d e r n computer technology have created a new era in diagnostic imaging, including digital radiography. The Toshiba c o m p u t e d radiography ( T C R ) system uses digital radiography based on the latest computer technologies [9]. High-quality images and digital analyses are obtainable by this system, which is currently used for diagnostic imaging in various fields of medicine [2, 4, 5, 8]. These studies are also of great value in diagnosing head and neck disease. The purpose
of this p a p e r is to show the usefulness of the quantitative analysis of radio-opacity of the paranasal sinuses using the T C R system.
Materials and methods The TCR system. The TCR system (model TCR-201) consists of the components shown in Fig. 1. The system uses a newly developed image receptor, called an imaging plate (IP), which replaces the conventional film-screen system. The IP is a plastic substrate containing barium fluorohalide, europium-doped crystals raised to higher energy levels by X-ray exposure. The energy is stored until it is released during scanning by a helium-neon laser. This property is called "photostimulable fluorescence". The IP is exposed in exactly the same way as a conventional X-ray film cassette for conventional radiographic images, contrast enhancement techniques, tomographic images, etc., but with the added advantage of using a lower dose of X-rays. The fluorescence generated by laser scanning is then converted to an electric current by the photoamplifler, so that the image information can be converted to electrical signals. The IP is first scanned by a weaker "scout" laser beam that does not affect the quality of the image. A histogram is then made representing the distribution of the energy absorbed. A second scanning releases all the energy stored, converting it to fluorescence which is then converted to an electric current. The system converts the amplitude of the current for each point on the IP to a digitalized value by using a function that is controlled to work most efficiently in visualizing the given image. This function is selected when the operator inputs into the system the image type (skull, chest, abdomen, limb, etc.). The system tailors the function according to the histogram made by the scout scanner so that the function best fits the given image. The digital value after conversion is referred to as the quantum value (Q-value). The image processor is a high-speed digital computer that provides image processing capabilities such as gradiation enhancement, spatial frequency enhancement, etc. It also enables analysis of the image in fine detail, making Q-value histograms in given areas, Q-value profiles along given lines, and reviewing all data statistically. Evaluation of optimum exposure technique. The various influences
Offprint requests to: H. Suzaki, Department of Otolaryngology,
Faculty of Medicine, University of Tokyo, 7-3-1 Hongo, Bunkyoku, Tokyo 113, Japan
on exposure due to differing technical conditions in the quantitative analysis of radio-opacity with the TCR system were studied by radiographing a phantom. Radiographs of the same phantom were
H. Suzaki et al. : Digital image analysis
327
Cassette ]~J[ Plate (IP)
Lampll
Eraser
IOver Reusabl e q]C~)E~ 1000 Times) '[l~)~ X- Ray
/
Tube
IP Reader
Image Processor
Image Developer Recorder
taken by the TCR system using various tube voltages ranging from 58 kVp to 98 kVp, and also using various tube currents (milliamperes) and exposure times (seconds) ranging from 10mAs to 20 mAs. Differences in the Q-values, obtained by converting the radio-opacity of the constant region under study, were analyzed by regression analysis to evaluate the optimum exposure technique.
Fig. 1. The Toshiba computed radiography system, model TCR-201
Results
depending on tube voltage was then analyzed using regression analysis. This showed that the regression was significant ( P < 0 . 0 1 ) and the Q-value on the image of the T C R system was influenced by tube voltage. Fluctuations in the Q-values in the constant region of the same p h a n t o m are shown in Fig. 3. Results were obtained with tube currents and exposure times ranging from 1 0 m A s to 2 0 m A s under a fixed tube voltage of 82kVp. The change in the Q-value depending on tube current and exposure time was analyzed using regression analysis, but the regression was not significant. The overall results of the p h a n t o m examination indicated that tube voltage must b e kept constant for quantitative analysis of radio-opacity when using the T C R system.
Phantom examination for evaluating exposure technique
Quantitative analysis of paranasal sinus radio-opacity
The changes in the Q-values in each constant region of the same p h a n t o m were taken with tube voltage ranging from 58 kVp to 98 kVp (Fig. 2). The change in Q-value
Images were taken with tube voltage 82 kVp in conventional radiography and 70 kVp in tomography. Waters'
Quantitative analysis of paranasal sinus radio-opacity. Statistical analysis of the Q-values, Q-value profiles and Q-value histograms of radio-opacifies in the paranasal sinuses were assessed using the TCR system and conventional radiographic and tomographic images. The various conditions of the sinus radio-opacities were also evaluated quantitatively using the Q-value statistics, the pattern of configuration of the Q-value profiles and the Q-value histograms.
QUANTUMVALUE QUANTUMVALUE 390 390"
380
380-
370
370-
360
360-
350 !
t
350340-
340 330 ~
330-
320
t
320-
310 -
310-
5'8
66
7'4
8'2
9'0
98 kVp
TUBE VOLTAGE
Fig. 2. The quantum value (Q-value) in the constant region of interest on each image of the same phantom. These show a lower value depending on tube voltage (ranging from 58 kVp to 98 kVp)
I
i
I
E
~
10
12
14
16
18
TUBE CURRENT (mA)
,
i
20 mAs
EXPOSURETIME (sec)
Fig. 3. The Q-value in the constant region of interest on the image of the same phantom taken with tube currents and exposure times ranging from 10mAs to 20mAs under a fixed tube voltage at 82 kVp
328
H. Suzaki et al.: Digital image analysis
Fig. 4. Waters' projections in a healthy volunteer and in cases with sinusitis. The radio-opacities in each maxillary sinus were converted to the corresponding Q-value
NORMAL
MODERATE
436.4 + 7 . 6 6
Pmfde
1024 Rt.Maxillary Sinus
0 660
364.7 +_ 7.41
Lt.MaxiUary Sinus
r Line (x 0. lrmn)
i
i 1400
Fig. 5. A case of left maxillary sinusitis caused by dental disease. The opacity profile along the line connecting the most lateral margins of the maxillary sinuses was used to produce the corresponding Q-value graph
SEVERE 270.0 ± 5.91
projections of a healthy volunteer and of severe and moderate cases of sinusitis are shown in Fig. 4. The radioopacities in the constant region visualized in the maxillary sinus were converted to corresponding Q-values. The Q-value was 436.4_+ 7.6 in the healthy volunteer, 364.7 + 7.4 in moderate sinusitis and 270.0 + 5.9 in severe sinusitis. As the degree of radio-opacity in the maxillary sinus became more severe, the corresponding Q-value obtained was lower. A case of left maxillary sinusitis is shown in Fig. 5, where the opacity profile along the line connecting the most lateral margins of the maxillary sinuses was reflected onto a Q-value graph. The configuration of the profile of the diseased left maxillary sinus was relatively flat, while that of the healthy (opposite) side showed ups and downs. A case of left sphenoid sinusitis is shown in the tomogram in Fig. 6. The opacity profile along the line connecting the most lateral margins of the sphenoid sinuses was reflected onto a Q-value graph. The configuration of the profile of the left sphenoid sinus showed a lower Qvalue in comparison with the configuration of the healthy right side. Histograms of opacities in cases of ethmoid sinusitis were compared with tomograms of a healthy volunteer (Fig. 7). In the histograms, the vertical axis indicates the frequency, which corresponds to the number of pixels in the region under study. The horizontal axis indicates the Q-value, which corresponds to the degree of opacity present in the image. The histogram of the healthy volunteer showed one sharp peak, while the histogram of the case of moderate ethmoid sinusitis showed multiple peaks and a lower Q-value. The pattern of the histogram of severe ethmoid sinusitis returned to a single peak (as in the healthy control), but the Q-value was the same as that for the case of moderate sinusitis.
H. Suzaki et al.: Digital image analysis
329
787
HISTOGRAM
)Z hi O I,I rf LL I
0
511
1023
QUANTUM VALUE
4~
Profile
1024
I(
HISTOGRAM
hl
Rt. Sphenoid Lt.Sphenoid ) ] .' Sinus ~l Sinus
O
5913
CY
i.
511 1023 QUANTUM VALUE
HISTOGRAM
>O Z Ld
0
[
I
I
600 Line(x 0.lmm) 1050 Fig. 6. A case of left sphenoid sinusitis. The opacity profile alo the line connectingthe most lateral margins was used to prodt the correspondingQ-value graph
Discussion
It is essential for clinicians to be able to properly evaluate radiographic densities in the paranasal sinuses in order to diagnose diseases present and evaluate the efficacy of therapeutic regimens employed. Although a radiographic density can be assessed by light transmission densitometry, the density of each X-ray film is influenced by various technical conditions, including tube voltage and current, exposure time, developer used, developer temperature, etc. Even X-ray films of the same patient may show different radiographic densities under different technical conditions. Therefore, the value of the density itself when assessed by light transmission densitometry cannot be accurately evaluated when compared with other values without correcting for the density fluctuations caused by any variable technical conditions [10]. Computer technology is becoming increasingly applicable to clinical radiology. The TCR system is one such digital computed radiography system that allows image characteristics to be manipulated by various processing options. This system makes it possible to improve image processing as well as to carry out quantitative analysis through the use of digital values. Initial evaluations of computed radiography systems have been reported recently [1, 3, 6, 7]. However, the use of digital analysis in computed radiography systems to diagnose disease has
O i,i n~ h 511
L ,,, 1023
QUANTUM VALUE
Fig.7. Histograms of various opacities obtained from tomograms of a healthy volunteer and cases of ethmoid sinusitisof various degrees of severity
not been investigated thoroughly. As a result, the present study was devised to determine if the TCR system was practical for quantitative analysis of radio-opacities in the paranasal sinuses. According to the results of our phantom study (Fig. 2), tube voltage must be kept constant for quantitative analysis of a radio-opacity when using the TCR system. Different body parts show different changes in X-ray attenuation at various energy levels as a function of their mean atomic number. For example, bone will generally show a change in the X-ray attenuation coefficient when compared to soft tissue when imaged at different X-ray energies. As mentioned above, the TCR system converts the level of energy stored in the IP into a Q-value according to one of a number of predetermined functions. These can be adjusted using a histogram calculated from the result of scout scanning. When a radiogram is taken with this system, some area of the IP must receive X-rays directly to enable the system to determine the maximum intensity and to adjust the upper limit of the converting function. When a skull image is produced, the system uses the peak of the histogram comprised of bone and
H. Suzaki et al.: Digital image analysis
330 soft tissue. Since this peak is not narrow but is relatively broad, the range it occupies influences the adjustment of the lower limit of the system's conversion function. If the radiographed image does not include the patient's shoulders, the tissue mass peak of the scout histogram is narrow and ambiguous. The system can also adjust the lower limit too high, so that the resulting Qvalue for such structures as the maxillary sinuses may become inappropriately low. However, if the image includes most of the shoulders, the system will make the proper adjustments, because the increase in the area of tissue mass above a certain level only influences the height of the peak but not its range. Based on these considerations, we took care to always include some part of our patients' shoulders during T C R imaging, and to achieve the correct angles between the IP and the patient's vertical and horizontal body lines. In making and analyzing T C R images, these two positionings are essential factors. The T C R system has demonstrated that images of different sinus opacities can be analyzed quantitatively and evaluated by assessing the Q-value, its profile configuration and a histogram. As illustrated in Fig. 4, different Q-values can be obtained for different opacities in the maxillary sinuses. The opacity profile along the selected line in the image is t h e n converted into a profile of the Q-value. The different graph patterns of the profile demonstrate different conditions found in the sinuses (Figs. 5, 6). A histogram of the Q-value corresponding to the opacity in the region of interest can then be obtained (Fig. 7). At the present time, these approaches analyze only the degree of an opacification, but do not determine its cause. Nevertheless, our findings show that digital image analysis using the T C R system can be quite
useful for quantitative analysis of opacities in the paranasal sinuses. References 1. Curtis D J, Ayella R J, Whitley J, Moser RP, Rugh KS (1979) Digital radiography in trauma using small-dose exposure. Radiology 132: 587-591 2. Fajardo LL, Hillman BJ, Hunter TB, Claypool HR, Westerman BR, Mockbee B (1987) Excretory urography using computed radiography. Radiology 162: 345-351 3. Goodman LR, Foley WD, Wilson CR, Rimm AA, Lawson TL (1986) Digital and conventional chest images: observer performance with film digital radiography system. Radiology 158: 27-33 4. Ishigaki T, Sakuma S, Ikeda M (1988) One-shot dual-energy subtraction chest imaging with computed radiography: clinical evaluation of film images. Radiology 168 : 67-72 5. Kohda E, Tanaka M, Fujioka M, Miyasaka K (1986) Computed radiography for the major airway in pediatrics. J Jpn Bronchoesophagol Soc 37: 393-399 6. Lams PM, Cocklin ML (1986) Spatial resolution requirements for digital chest radiographs: an ROC study of observer performance in selected cases. Radiology 158 : 11-19 7. MacMahon H, Vyborny CJ, Metz CE, Doi K, Sabeti V, Solomon SL (1986) Digital radiography of subtle pulmonary abnormalities: an ROC study of the effect of pixel size on observer performance. Radiology 158 : 21-26 8. Pond GD, Seeley GW, Ovitt TW, Chernin MM, Yoshino MT, Roehrig H, McIntyre KE (1989) Intraoperative arteriography: comparison of conventional screen-film with photostimulable imaging plate radiographs. Radiology 170: 367-370 9. Sonoda M, Takano M, Miyahara J, Kato H (1983) Computed radiography utilizing scanning laser stimulated luminescence. Radiology 148: 833-838 10. Suzaki H, Abbey K (1988) Quantitative assessment of radioopacity in sinusitis: densitometry using the copper-step-wedge system. ORL (Tokyo) 31 [Suppl 7] : 748-751
European Archives of
Oto-RhinoLaryngology © Springer-Verlag1991
Digital image analysis of radio-opacities in the paranasal sinuses using computed radiography H. Suzaki 1, K. Abbey 2, and Y. Nomura 1 1Department of Otolaryngology, University of Tokyo, Tokyo, Japan 2Department of Otolaryngology, Musashino Red Cross Hospital, Tokyo, Japan Received October 18, 1990 / Accepted November 6, 1990
Summary. A c o m p u t e d radiograph system (Toshiba, model TCR-201) was used to investigate the digital image analysis of radio-opacity of the paranasal sinuses. The results of the preliminary p h a n t o m examination for evaluating the exposure technique indicated that the tube voltage should be kept constant. By using conventional radiographic images and tomographic images of normals and cases of sinusitis, the quantum values (Qvalues), Q-value profiles, and Q-value histograms of radio-opacities in the paranasal sinuses were assessed statistically. Findings demonstrated that radio-opacities in the paranasal sinuses could be evaluated quantitatively by these digital image analyses. Key words: C o m p u t e d radiography - Digital image analysis - Paranasal sinuses - Sinus opacifications
Introduction Radiological evaluation is indispensable for the diagnosis of diseases of the paranasal sinuses. H o w e v e r , radio-opacity of a sinus cavity is generally evaluated visually by an examiner. Such evaluations are not objective because of fluctuations in the radiographic density of each X-ray film, depending on various technical conditions in the regular film-screen combination system. Recent advances in m o d e r n computer technology have created a new era in diagnostic imaging, including digital radiography. The Toshiba c o m p u t e d radiography ( T C R ) system uses digital radiography based on the latest computer technologies [9]. High-quality images and digital analyses are obtainable by this system, which is currently used for diagnostic imaging in various fields of medicine [2, 4, 5, 8]. These studies are also of great value in diagnosing head and neck disease. The purpose
of this p a p e r is to show the usefulness of the quantitative analysis of radio-opacity of the paranasal sinuses using the T C R system.
Materials and methods The TCR system. The TCR system (model TCR-201) consists of the components shown in Fig. 1. The system uses a newly developed image receptor, called an imaging plate (IP), which replaces the conventional film-screen system. The IP is a plastic substrate containing barium fluorohalide, europium-doped crystals raised to higher energy levels by X-ray exposure. The energy is stored until it is released during scanning by a helium-neon laser. This property is called "photostimulable fluorescence". The IP is exposed in exactly the same way as a conventional X-ray film cassette for conventional radiographic images, contrast enhancement techniques, tomographic images, etc., but with the added advantage of using a lower dose of X-rays. The fluorescence generated by laser scanning is then converted to an electric current by the photoamplifler, so that the image information can be converted to electrical signals. The IP is first scanned by a weaker "scout" laser beam that does not affect the quality of the image. A histogram is then made representing the distribution of the energy absorbed. A second scanning releases all the energy stored, converting it to fluorescence which is then converted to an electric current. The system converts the amplitude of the current for each point on the IP to a digitalized value by using a function that is controlled to work most efficiently in visualizing the given image. This function is selected when the operator inputs into the system the image type (skull, chest, abdomen, limb, etc.). The system tailors the function according to the histogram made by the scout scanner so that the function best fits the given image. The digital value after conversion is referred to as the quantum value (Q-value). The image processor is a high-speed digital computer that provides image processing capabilities such as gradiation enhancement, spatial frequency enhancement, etc. It also enables analysis of the image in fine detail, making Q-value histograms in given areas, Q-value profiles along given lines, and reviewing all data statistically. Evaluation of optimum exposure technique. The various influences
Offprint requests to: H. Suzaki, Department of Otolaryngology,
Faculty of Medicine, University of Tokyo, 7-3-1 Hongo, Bunkyoku, Tokyo 113, Japan
on exposure due to differing technical conditions in the quantitative analysis of radio-opacity with the TCR system were studied by radiographing a phantom. Radiographs of the same phantom were
H. Suzaki et al. : Digital image analysis
327
Cassette ]~J[ Plate (IP)
Lampll
Eraser
IOver Reusabl e q]C~)E~ 1000 Times) '[l~)~ X- Ray
/
Tube
IP Reader
Image Processor
Image Developer Recorder
taken by the TCR system using various tube voltages ranging from 58 kVp to 98 kVp, and also using various tube currents (milliamperes) and exposure times (seconds) ranging from 10mAs to 20 mAs. Differences in the Q-values, obtained by converting the radio-opacity of the constant region under study, were analyzed by regression analysis to evaluate the optimum exposure technique.
Fig. 1. The Toshiba computed radiography system, model TCR-201
Results
depending on tube voltage was then analyzed using regression analysis. This showed that the regression was significant ( P < 0 . 0 1 ) and the Q-value on the image of the T C R system was influenced by tube voltage. Fluctuations in the Q-values in the constant region of the same p h a n t o m are shown in Fig. 3. Results were obtained with tube currents and exposure times ranging from 1 0 m A s to 2 0 m A s under a fixed tube voltage of 82kVp. The change in the Q-value depending on tube current and exposure time was analyzed using regression analysis, but the regression was not significant. The overall results of the p h a n t o m examination indicated that tube voltage must b e kept constant for quantitative analysis of radio-opacity when using the T C R system.
Phantom examination for evaluating exposure technique
Quantitative analysis of paranasal sinus radio-opacity
The changes in the Q-values in each constant region of the same p h a n t o m were taken with tube voltage ranging from 58 kVp to 98 kVp (Fig. 2). The change in Q-value
Images were taken with tube voltage 82 kVp in conventional radiography and 70 kVp in tomography. Waters'
Quantitative analysis of paranasal sinus radio-opacity. Statistical analysis of the Q-values, Q-value profiles and Q-value histograms of radio-opacifies in the paranasal sinuses were assessed using the TCR system and conventional radiographic and tomographic images. The various conditions of the sinus radio-opacities were also evaluated quantitatively using the Q-value statistics, the pattern of configuration of the Q-value profiles and the Q-value histograms.
QUANTUMVALUE QUANTUMVALUE 390 390"
380
380-
370
370-
360
360-
350 !
t
350340-
340 330 ~
330-
320
t
320-
310 -
310-
5'8
66
7'4
8'2
9'0
98 kVp
TUBE VOLTAGE
Fig. 2. The quantum value (Q-value) in the constant region of interest on each image of the same phantom. These show a lower value depending on tube voltage (ranging from 58 kVp to 98 kVp)
I
i
I
E
~
10
12
14
16
18
TUBE CURRENT (mA)
,
i
20 mAs
EXPOSURETIME (sec)
Fig. 3. The Q-value in the constant region of interest on the image of the same phantom taken with tube currents and exposure times ranging from 10mAs to 20mAs under a fixed tube voltage at 82 kVp
328
H. Suzaki et al.: Digital image analysis
Fig. 4. Waters' projections in a healthy volunteer and in cases with sinusitis. The radio-opacities in each maxillary sinus were converted to the corresponding Q-value
NORMAL
MODERATE
436.4 + 7 . 6 6
Pmfde
1024 Rt.Maxillary Sinus
0 660
364.7 +_ 7.41
Lt.MaxiUary Sinus
r Line (x 0. lrmn)
i
i 1400
Fig. 5. A case of left maxillary sinusitis caused by dental disease. The opacity profile along the line connecting the most lateral margins of the maxillary sinuses was used to produce the corresponding Q-value graph
SEVERE 270.0 ± 5.91
projections of a healthy volunteer and of severe and moderate cases of sinusitis are shown in Fig. 4. The radioopacities in the constant region visualized in the maxillary sinus were converted to corresponding Q-values. The Q-value was 436.4_+ 7.6 in the healthy volunteer, 364.7 + 7.4 in moderate sinusitis and 270.0 + 5.9 in severe sinusitis. As the degree of radio-opacity in the maxillary sinus became more severe, the corresponding Q-value obtained was lower. A case of left maxillary sinusitis is shown in Fig. 5, where the opacity profile along the line connecting the most lateral margins of the maxillary sinuses was reflected onto a Q-value graph. The configuration of the profile of the diseased left maxillary sinus was relatively flat, while that of the healthy (opposite) side showed ups and downs. A case of left sphenoid sinusitis is shown in the tomogram in Fig. 6. The opacity profile along the line connecting the most lateral margins of the sphenoid sinuses was reflected onto a Q-value graph. The configuration of the profile of the left sphenoid sinus showed a lower Qvalue in comparison with the configuration of the healthy right side. Histograms of opacities in cases of ethmoid sinusitis were compared with tomograms of a healthy volunteer (Fig. 7). In the histograms, the vertical axis indicates the frequency, which corresponds to the number of pixels in the region under study. The horizontal axis indicates the Q-value, which corresponds to the degree of opacity present in the image. The histogram of the healthy volunteer showed one sharp peak, while the histogram of the case of moderate ethmoid sinusitis showed multiple peaks and a lower Q-value. The pattern of the histogram of severe ethmoid sinusitis returned to a single peak (as in the healthy control), but the Q-value was the same as that for the case of moderate sinusitis.
H. Suzaki et al.: Digital image analysis
329
787
HISTOGRAM
)Z hi O I,I rf LL I
0
511
1023
QUANTUM VALUE
4~
Profile
1024
I(
HISTOGRAM
hl
Rt. Sphenoid Lt.Sphenoid ) ] .' Sinus ~l Sinus
O
5913
CY
i.
511 1023 QUANTUM VALUE
HISTOGRAM
>O Z Ld
0
[
I
I
600 Line(x 0.lmm) 1050 Fig. 6. A case of left sphenoid sinusitis. The opacity profile alo the line connectingthe most lateral margins was used to prodt the correspondingQ-value graph
Discussion
It is essential for clinicians to be able to properly evaluate radiographic densities in the paranasal sinuses in order to diagnose diseases present and evaluate the efficacy of therapeutic regimens employed. Although a radiographic density can be assessed by light transmission densitometry, the density of each X-ray film is influenced by various technical conditions, including tube voltage and current, exposure time, developer used, developer temperature, etc. Even X-ray films of the same patient may show different radiographic densities under different technical conditions. Therefore, the value of the density itself when assessed by light transmission densitometry cannot be accurately evaluated when compared with other values without correcting for the density fluctuations caused by any variable technical conditions [10]. Computer technology is becoming increasingly applicable to clinical radiology. The TCR system is one such digital computed radiography system that allows image characteristics to be manipulated by various processing options. This system makes it possible to improve image processing as well as to carry out quantitative analysis through the use of digital values. Initial evaluations of computed radiography systems have been reported recently [1, 3, 6, 7]. However, the use of digital analysis in computed radiography systems to diagnose disease has
O i,i n~ h 511
L ,,, 1023
QUANTUM VALUE
Fig.7. Histograms of various opacities obtained from tomograms of a healthy volunteer and cases of ethmoid sinusitisof various degrees of severity
not been investigated thoroughly. As a result, the present study was devised to determine if the TCR system was practical for quantitative analysis of radio-opacities in the paranasal sinuses. According to the results of our phantom study (Fig. 2), tube voltage must be kept constant for quantitative analysis of a radio-opacity when using the TCR system. Different body parts show different changes in X-ray attenuation at various energy levels as a function of their mean atomic number. For example, bone will generally show a change in the X-ray attenuation coefficient when compared to soft tissue when imaged at different X-ray energies. As mentioned above, the TCR system converts the level of energy stored in the IP into a Q-value according to one of a number of predetermined functions. These can be adjusted using a histogram calculated from the result of scout scanning. When a radiogram is taken with this system, some area of the IP must receive X-rays directly to enable the system to determine the maximum intensity and to adjust the upper limit of the converting function. When a skull image is produced, the system uses the peak of the histogram comprised of bone and
H. Suzaki et al.: Digital image analysis
330 soft tissue. Since this peak is not narrow but is relatively broad, the range it occupies influences the adjustment of the lower limit of the system's conversion function. If the radiographed image does not include the patient's shoulders, the tissue mass peak of the scout histogram is narrow and ambiguous. The system can also adjust the lower limit too high, so that the resulting Qvalue for such structures as the maxillary sinuses may become inappropriately low. However, if the image includes most of the shoulders, the system will make the proper adjustments, because the increase in the area of tissue mass above a certain level only influences the height of the peak but not its range. Based on these considerations, we took care to always include some part of our patients' shoulders during T C R imaging, and to achieve the correct angles between the IP and the patient's vertical and horizontal body lines. In making and analyzing T C R images, these two positionings are essential factors. The T C R system has demonstrated that images of different sinus opacities can be analyzed quantitatively and evaluated by assessing the Q-value, its profile configuration and a histogram. As illustrated in Fig. 4, different Q-values can be obtained for different opacities in the maxillary sinuses. The opacity profile along the selected line in the image is t h e n converted into a profile of the Q-value. The different graph patterns of the profile demonstrate different conditions found in the sinuses (Figs. 5, 6). A histogram of the Q-value corresponding to the opacity in the region of interest can then be obtained (Fig. 7). At the present time, these approaches analyze only the degree of an opacification, but do not determine its cause. Nevertheless, our findings show that digital image analysis using the T C R system can be quite
useful for quantitative analysis of opacities in the paranasal sinuses. References 1. Curtis D J, Ayella R J, Whitley J, Moser RP, Rugh KS (1979) Digital radiography in trauma using small-dose exposure. Radiology 132: 587-591 2. Fajardo LL, Hillman BJ, Hunter TB, Claypool HR, Westerman BR, Mockbee B (1987) Excretory urography using computed radiography. Radiology 162: 345-351 3. Goodman LR, Foley WD, Wilson CR, Rimm AA, Lawson TL (1986) Digital and conventional chest images: observer performance with film digital radiography system. Radiology 158: 27-33 4. Ishigaki T, Sakuma S, Ikeda M (1988) One-shot dual-energy subtraction chest imaging with computed radiography: clinical evaluation of film images. Radiology 168 : 67-72 5. Kohda E, Tanaka M, Fujioka M, Miyasaka K (1986) Computed radiography for the major airway in pediatrics. J Jpn Bronchoesophagol Soc 37: 393-399 6. Lams PM, Cocklin ML (1986) Spatial resolution requirements for digital chest radiographs: an ROC study of observer performance in selected cases. Radiology 158 : 11-19 7. MacMahon H, Vyborny CJ, Metz CE, Doi K, Sabeti V, Solomon SL (1986) Digital radiography of subtle pulmonary abnormalities: an ROC study of the effect of pixel size on observer performance. Radiology 158 : 21-26 8. Pond GD, Seeley GW, Ovitt TW, Chernin MM, Yoshino MT, Roehrig H, McIntyre KE (1989) Intraoperative arteriography: comparison of conventional screen-film with photostimulable imaging plate radiographs. Radiology 170: 367-370 9. Sonoda M, Takano M, Miyahara J, Kato H (1983) Computed radiography utilizing scanning laser stimulated luminescence. Radiology 148: 833-838 10. Suzaki H, Abbey K (1988) Quantitative assessment of radioopacity in sinusitis: densitometry using the copper-step-wedge system. ORL (Tokyo) 31 [Suppl 7] : 748-751
