Neuroradiolog y/Head and Neck Imaging • Original Research Yuan et al. CT Analysis of Globe Rupture

Downloaded from www.ajronline.org by East Carolina University on 06/09/14 from IP address 150.216.68.200. Copyright ARRS. For personal use only; all rights reserved

Neuroradiology/Head and Neck Imaging Original Research

CT of Globe Rupture: Analysis and Frequency of Findings Wei-Hsin Yuan1,2,3 Hui-Chen Hsu 4 Hui-Cheng Cheng 4 Wan-Yuo Guo 2,3 Michael Mu-Huo Teng2,5 Shih-Jen Chen 3,6 Tai-Chi Lin 3,6 Yuan WH, Hsu HC, Cheng HC, et al.

Keywords: anterior chamber depth (ACD), CT, globe rupture DOI:10.2214/AJR.13.11010 Received April 2, 2013; accepted after revision September 6, 2013. 1

Division of Radiology, Taipei Municipal Gan-Dau Hospital, Taipei, Taiwan, Republic of China. 2

Department of Radiology, Taipei Veterans General Hospital, Shih-Pai Rd, No. 201, Section 2, Taipei, Taiwan 11217, Republic of China. Address correspondence to W. Y. Guo ([email protected]) and H. C. Hsu ([email protected]). 3 School of Medicine, National Yang Ming University, Taipei, Taiwan, Republic of China. 4 Division of Medical Imaging for Health Management, Cheng Hsin General Hospital, Taipei, Taiwan, Republic of China. 5 Department of Medical Imaging, Cheng Hsin General Hospital, Taipei, Taiwan, Republic of China. 6 Department of Ophthalmology, Taipei Veterans General Hospital, Taipei, Taiwan, Republic of China.

AJR 2014; 202:1100–1107 0361–803X/14/2025–1100 © American Roentgen Ray Society

1100

OBJECTIVE. The objective of our study was to evaluate the CT characteristics of globe rupture. MATERIALS AND METHODS. The medical records of patients seen in the emergency department with blunt, penetrating, or explosive orbit injury were retrospectively reviewed. A total of 75 patients (76 injured globes) were included (56 males and 19 females; average age, 45.1 years; age range, 5–95 years). CT examinations were reviewed by two experienced radiologists without knowledge of ophthalmologic findings, original orbital CT images, or surgical outcomes. RESULTS. Of the 76 globe injuries, 33 (43%) were ruptured and 43 (57%) were nonruptured. There were significant differences between the ruptured and nonruptured globes with respect to intraocular hemorrhage, lens dislocation and destruction, an intraocular foreign body, intraocular gas, anterior chamber depth (ACD), and globe deformity and wall irregularity (p < 0.05). There was good interrater agreement between the two radiologists (kappa value range, 0.63–0.96). The average sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of CT for the detection of globe rupture based on readings by two radiologists were 76%, 85%, 80%, 82%, and 81%, respectively. CONCLUSION. Although CT is extremely useful in the evaluation of ocular trauma, it should not be solely relied on for the diagnosis of globe rupture because of the potentially catastrophic consequences of an undiagnosed injury. A difference in ACD can be diagnostic of globe rupture.

G

lobe rupture is an ophthalmologic emergency that can lead to serious complications such as posttraumatic endophthalmitis, vision loss, sympathetic ophthalmia, meningitis, brain abscess, and even death [1, 2]. Acute globe rupture may occur as a result of blunt trauma or a penetrating injury to the eye, and early diagnosis and prompt treatment are crucial for improving the prognosis [3]. Clinical signs and symptoms of globe rupture (which will vary with the intraocular pressure) include decreased visual acuity, intraocular hemorrhage, low intraocular pressure, irregular pupil shape, and a reduced or enlarged anterior chamber (AC) [4, 5]. The clinical assessment of globe rupture in an emergency department can be challenging because of poor patient cooperation, altered consciousness, and the presence of head trauma and periorbital soft-tissue swelling. Radiography and sonography are inadequate for delineating soft-tissue structures of the trau-

matized eye or depicting the extent of bony deformities and fractures [6, 7]. In addition, sonography is operator-dependent and is contraindicated if a ruptured globe is suspected [7, 8]. MRI is useful for differentiating edema from hemorrhage and for depicting the optic nerve and condition of the globe [6]. However, MRI is insensitive regarding visualization of foreign bodies that may be present in the eye, such as wood, glass, and bone fragments. Intraocular ferromagnetic foreign objects can move during an MRI examination as a result of the magnetic field, and this movement can produce further trauma or blindness [7, 9, 10]. Thus, MRI is contraindicated if ferromagnetic foreign bodies are suspected within or near the orbit. For the aforementioned reasons, MRI is not recommended as an initial assessment for orbital trauma [7]. Thin-section CT offers excellent distinction between normal and abnormal bony and soft-tissue structures, and MDCT can provide high-resolution images and 3D recon-

AJR:202, May 2014

Downloaded from www.ajronline.org by East Carolina University on 06/09/14 from IP address 150.216.68.200. Copyright ARRS. For personal use only; all rights reserved

CT Analysis of Globe Rupture struction. In addition, because of its short image acquisition time, MDCT is advantageous in the unconscious or uncooperative patient [6]. Studies have also shown that CT is the most reliable method for the identification of intraocular foreign bodies [11, 12]. CT is also the primary method for the identification of facial and orbital fractures and alterations in the positions of the rectus muscles [13, 14]. Thus, CT has become the primary imaging modality for the assessment of acute head and globe trauma [6, 7, 15]. CT findings suggestive of globe rupture include scleral discontinuity; the presence of intraocular air, hemorrhage, or foreign body; a change in globe contour; a change in globe volume; and increased or decreased anterior chamber depth (ACD) [1–3, 6, 7, 15, 16]. The purpose of this study was to evaluate the CT characteristics of globe rupture, to determine the relative frequency of the various features of globe rupture on CT, and to assess the significance of a change in the ACD on CT images to improve the diagnosis of globe rupture. Materials and Methods Subjects Taipei Veterans General Hospital, a national medical center, provides veterans, dependents, and the general public with the highest quality care. In this retrospective study, we searched the diagnostic database of the emergency department at Taipei Veterans General Hospital using the index term “eye disease” for cases treated from January 2002 to January 2012. A total of 2134 patients with the diagnosis of eye disease were identified and screened. Patients were included in this study if they had no history of eye disease or trauma and were seen in the emergency department for the treatment of a blunt, penetrating, or explosive orbital injury and had undergone CT of the orbits after having had a complete or incomplete ophthalmologic examination. A complete ophthalmologic examination included visual acuity, intraocular pressure, eye movement, slit-lamp biomicroscopy, funduscopy, or sonography by an in-house ophthalmologist. Of the 2134 patients, 75 patients (76 injured globes) met the inclusion criteria and were included in the analysis. The 75 patients subsequently analyzed included 56 males and 19 females with an average age of 45.1 years (range, 5–95 years). One patient had bilateral globe injuries, whereas all others had a unilateral globe injury. All CT examinations were performed within 6 days after globe injuries. Patients with ruptured globes subsequently underwent orbital surgery for treatment.

CT Technique, Radiation Dose, and Image Interpretation CT images were obtained using either a singledetector CT scanner (Genesis HiSpeed RP, GE Healthcare) or an MDCT scanner (LightSpeed QX/i, GE Healthcare; Somatom Sensation 16, Siemens Healthcare; Brilliance 40, Philips Healthcare; or Aquilion 64, Toshiba Medical Systems). All CT scans were obtained without IV contrast agent. CT was performed on a single-detector unit (n = 16 patients) or on an MDCT unit including 4-slice (n = 3), 16-slice (n = 3), 40-slice (n = 52), and 64-slice (n = 1) MDCT scanners. The following imaging parameters were used for each type of CT scanner: 140 kV, 120–150 mA, 0.8-second gantry rotation time, 3-mm slice thickness, and 1 pitch for the single-detector CT scanner; 120 kV, 89–345 mA, 0.5- to 0.8-second gantry rotation time, 2- or 3-mm slice thickness, and 0.5–0.8 pitch for the MDCT scanners. The axial sections of orbital CT images were performed along the transaxial direction with the sections parallel to the optic nerve along a line from the inferior rim of the maxillary sinus to the middle portion of the frontal region. The coronal sections were taken from the nose to the dorsum sellae. Direct coronal CT images, with a 3-mm slice thickness, were obtained of all patients using single-detector CT and were reformatted with MDCT. The slice thickness for image viewing of axial and coronal images was 3 mm using the single-detector CT scanner and 2 or 3 mm for each type of MDCT scanner. The reconstruction matrix for both single-detector CT and MDCT scans was 512 × 512. In our hospital, CT radiation dosimetry using the volume CT dose index (CTDIvol) and doselength product (DLP) was not performed for any of the 16 patients scanned using the single-detector CT unit or for any of the 38 patients who underwent MDCT from January 2002 to January 2010. The following radiation dosimetry data were recorded for the remaining 21 patients scanned with MDCT scanners: CTDIvol of 12.3–40.5 mGy and DLP of 271.0–625.6 mGy × cm. The approximate range of effective doses to patients (converted from DLP using k conversion coefficient for head [0.0021 mSv/(mGy × cm)]) was 0.6– 1.8 mSv [17, 18]. The lowest effective dose was present in one patient scanned with 16-MDCT (89 mA: corresponding CTDIvol = 21.1 mGy, DLP = 273.0 mGy × cm) and one patient scanned with 40-MDCT (184 mA: corresponding CTDIvol = 12.3 mGy, DLP = 271.0 mGy × cm). All CT examinations were assigned a random order by a researcher not involved in interpretation. Two experienced radiologists (with 17 and 26 years of experience, respectively), who did not have knowledge of the ophthalmologic find-

ings, original orbital CT images, or surgical outcomes, interpreted the CT images. The two radiologists independently evaluated the CT images on a PACS monitor, and each interpretation was recorded individually. The following CT data were obtained and recorded: the contour, dimensions, and content of each globe; altered ACD; and abnormalities of the structures adjacent to the eyes. Patients with lens dislocation or destruction were excluded from the analysis of altered ACD. The two radiologists were also asked to provide an impression of globe rupture versus nonrupture according to their evaluation of contour, dimension, content of each globe, and altered ACD on orbital CT scans. A third experienced radiologist (with 16 years of experience) integrated the ophthalmologic examination results, operative findings, and followup medical records of at least 1 month to obtain the clinical diagnosis of abnormal ocular findings and globe rupture in patients. According to the clinical diagnosis, the presence of a full-thickness laceration or penetrating wound of the cornea, sclera, or both was defined as globe rupture. Two experienced neuroradiologists (with 25 and 35 years of experience, respectively), by consensus, reanalyzed the CT images of the 75 patients on a PACS and revised the original CT orbital radiology reports. These revised reports served as the final radiologic diagnosis. The third radiologist performed the final radiologic-clinical comparison and surgical correlation to serve as the final diagnostic standard for abnormal findings in all 75 patients. Thus, the presence of globe rupture was confirmed by the clinical diagnosis. Any abnormal or normal finding of ocular injuries obtained from the clinical diagnosis was considered as the definite diagnosis. Other clinically unknown findings present in the radiologic interpretation were verified by the final radiologic diagnosis. The third radiologist also compiled the results of the first two radiologists who reviewed the CT images. The third radiologist used electronic calipers on a PACS workstation to measure and record the anteroposterior diameter of ACD for both globes in each patient with shallow or deepened ACs using the final radiologic-clinical comparison or surgical correlation. Each measurement was made at the equator of the globe and along a line perpendicular to the long axis of the lens, and the anteroposterior diameter of the ACD was evaluated from the posterior surface of the cornea to the anterior surface of the lens. The third radiologist measured each anteroposterior diameter twice to obtain the mean anteroposterior diameter for ACD comparison. The second ACD measurement was obtained 1 week after the first measurement.

AJR:202, May 2014 1101

Yuan et al.

Downloaded from www.ajronline.org by East Carolina University on 06/09/14 from IP address 150.216.68.200. Copyright ARRS. For personal use only; all rights reserved

Statistical Analysis Data are presented as frequencies and percentages, and the categoric variables between ruptured and nonruptured globes were compared by the chisquare test or Fisher exact test. Interrater agreement was assessed using kappa statistics with kappa values of greater than 0.75 indicating excellent agreement and values between 0.40 and 0.75 representing fair to good agreement. The final radiologic-clinical comparison or surgical correlation served as the basis for calculating the sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), and accuracy of the CT features assessed by the two interpreting radiologists. The ACD difference between the two eyes in the same patient was compared by the paired Student t test. Data were analyzed using statistics software (SPSS, version 15.0, SPSS), and a value of p < 0.05 was considered statistically significant.

Results Sixty-two of 76 (82%) injured globes received a complete preoperative ophthalmologic examination; the other 14 (18%) had incomplete ophthalmologic examinations

A

because of severe eyelid swelling or lacerations, total hyphema, poor patient cooperation, or unconsciousness. According to the final radiologic-clinical comparison or surgical correlation, of the 76 globes, 29 (38%) exhibited a globe deformity or wall irregularity (Fig. 1A), 31 (41%) exhibited intraocular hemorrhage (Fig. 1A), 18 (24%) had lens dislocation or destruction (Fig. 1A), 54 (71%) exhibited eyelid hematoma or swelling (Fig. 1A), seven (9%) had an intraocular foreign body (Fig. 1B), six (8%) had intraocular gas (Fig. 1B), 10 (13%) had a shallow AC (Fig. 1C), and 21 (28%) were associated with fractures. Fourteen of the 31 (45%) intraocular hemorrhages showed hemorrhage in a single compartment, including three hyphemas, seven vitreous hemorrhages, and four retinal detachments; 17 of 31 (55%) showed hemorrhage in two or three compartments. There was no case with a deepened AC. Of the seven intraocular foreign bodies, five (71%) were metallic and two (29%) were nonmetallic. Among the 29 globe deformities or wall irregularities, 17 globe deformities had volume change and 12 had wall irregularities without volume change. Also, 10 of 29 (35%) globe deformities or wall irregularities showed a shrunken deformity with intraocular hemorrhage, five (17%) exhibited an enlarged deformity with intraocular hem-

Fig. 1—CT findings of globe rupture. A, 75-year-old man with right globe rupture. Axial unenhanced CT scan shows eyelid hematoma (thick straight arrow), lens dislocation (arrowhead), vitreous hemorrhage (thin straight arrow), and irregular scleral wall (curved arrow). B, 80-year-old man with left eyeball rupture. Axial unenhanced CT scan shows gas (short arrow) and metallic foreign body (long arrow) in ruptured globe. C, 34-year-old man with left eyeball rupture. Axial unenhanced CT scan shows shallow anterior chamber depth (ACD) of left globe (single arrow). ACD is evaluated at level of equator of globe from posterior surface of cornea to anterior surface of lens (parallel lines) and is measured along line perpendicular to long axis of lens (double arrows). Difference in anterior-posterior diameters of ACD between two globes is 2.2 mm.

B

1102

orrhage, and two (7%) had a shrunken deformity without intraocular hemorrhage. Twenty-one of the 76 (28%) globes were associated with fractures. Orbital fractures involved the medial walls (12/21, 57%), floor (14/21, 67%), or lateral walls (7/21, 33%). Only one fracture (1/21, 5%) involved the orbital roof. Of the 21 orbital fractures, four (19%) showed obvious herniation of the inferior rectus muscle, two (10%) showed herniation of the medial and inferior rectus muscle, and three (14%) showed herniation of the medial rectus muscle. No optic nerve transection could be identified. Of the 76 globes, there were 13 (17%) cases of iris or uvea prolapse, four (5%) vitreous incarcerations, nine (12%) commotio retinae, and 11 (15%) cases of subconjunctival hemorrhage or edema that were diagnosed by ophthalmologists but were not diagnosed on orbital CT scans. However, four globes (5%) showed obvious retrobulbar hemorrhage on CT scans that was not seen by the ophthalmologists. Incidental CT findings included four intracranial hemorrhages, a suspicious arachnoid cyst in the left middle fossa, and 26 cases of mucus or hemorrhage in the paranasal sinuses. Clinical diagnoses revealed 33 (43%) ruptured globes and 43 (57%) nonruptured globes in 76 injured globes. Differences in eyeball deformity or wall irregularity, shallow AC, lens dislocation or destruction, intraocular

C

AJR:202, May 2014

Downloaded from www.ajronline.org by East Carolina University on 06/09/14 from IP address 150.216.68.200. Copyright ARRS. For personal use only; all rights reserved

CT Analysis of Globe Rupture Fig. 2—False-positive diagnosis of globe rupture on CT scans of 52-year-old man. Axial unenhanced CT scan shows left eyelid hematoma (thick arrow) and wall irregularity of left globe (thin arrow). Subconjunctival hemorrhage and edema in left globe were misinterpreted as globe deformity and wall irregularity on CT scans, which led to false-positive diagnosis of globe rupture by radiologist.

hemorrhage, intraocular foreign body, and intraocular gas (Fig. 1) between the two groups (ruptured vs nonruptured globes) were significant (p < 0.05). The number (percentage) and p value of the abnormal features between ruptured versus nonruptured globes, respectively, were as follows: eight of 33 (24%) versus 13 of 43 (30%) for orbital fractures (p = 0.613); 25 of 33 (76%) versus six of 43 (14%) for intraocular hemorrhage (p < 0.001); 15 of 33 (46%) versus three of 43 (7%) for lens dislocation or destruction (p < 0.001); seven of 33 (21%) versus 0 of 43 (0%) for an intraocular foreign body (p = 0.002); six of 33 (18%) versus 0 of 43 (0%) for intraocular gas (p = 0.005); 10 of 33 (30%) versus 0 of 43 (0%) for a shallow AC (p < 0.001); 29 of 33 (88%) versus 0 of 43 (0%) for a global deformity or wall irregularity (p < 0.001); 22 of 33 (67%) versus 32 of 43 (74%) for an eyelid hematoma or swelling (p = 0.610). Of the 33 ruptured globes, findings in four (12%) globes consisted of only a single feature, including one shallow AC, two intraoc-

ular hemorrhages, and one foreign body. The findings in the other 29 ruptured globes consisted of two or more features. Neither the finding of orbital fracture nor the finding of eyelid hematoma or swelling showed a significant difference between ruptured versus nonruptured globes (p > 0.05). Because the CT readings of the two radiologists were very similar, we present the average figure for each finding and the interrater agreement in Table 1. The kappa value ranged from 0.63 to 0.96, with kappa values of more than 0.75 indicating excellent agreement and those ranging from 0.40 to 0.75 indicating fair to good agreement. For the two radiologists, the average sensitivity, specificity, PPV, NPV, and accuracy of CT for the detection of globe rupture were 76%, 85%, 80%, 82%, and 81%, respectively (Table 1). The readings of intraocular hemorrhage, globe deformity or wall irregularity, and intraocular foreign bodies on the part of both radiologists exhibited a somewhat lower sensitivity. The readings of intraocular hemorrhage, globe deformity or wall

irregularity had a lower accuracy. Neither radiologist had a false-positive interpretation for intraocular hemorrhage or orbital fracture on CT or a false-negative interpretation for intraocular gas. There were 10 orbits with shallow ACs, with a mean anteroposterior diameter of 1.5 mm. The mean anteroposterior diameter of the 10 contralateral normal ACs in the same patients was 3.1 mm. The ACD of the two orbits in the same patient was significantly different (p = 0.001), and the mean anteriorposterior diameter loss was 1.6 mm (95% CI, 0.8–2.4 mm). The case with a false-positive shallow AC interpreted by radiologist A had a left ACD of 3.0 mm and a right ACD of 3.1 mm. The other case with a false-positive shallow AC by radiologist B had a left ACD of 3.1 mm and a right ACD of 3.2 mm. Figures 2 and 3 are examples of false-positive and false-negative readings of globe rupture on CT, respectively. These false-positive and false-negative globe ruptures (read by either a single radiologist or both radiologists) were recognized by the final radiologic-clinical comparison or surgical correlation. The false-positive findings on CT included a lens dislocation, which was misinterpreted because of an upper gaze and large deformity of the globe; a small calcification embedded in the scleral wall, which simulated an intraocular foreign body; small gas bubbles closely attached to the surface of two globes, which simulated intraocular gas; two false-positive shallow ACs because both lenses were not in the same plane on the images; seven cases of subconjunctival hemorrhages or edema or intraocular hemorrhages that were read as false-positive global wall irregularities in nonruptured globes (Fig. 2); and three cases of eyelid hematoma or swell-

TABLE 1: Interrater Agreement Between Two Radiologists and Average Sensitivity, Specificity, Positive Predictive Value (PPV), Negative Predictive Value (NPV), and Accuracy for Each Orbital CT Finding in 76 Injured Globes Agreement (%)

κ

Average Sensitivity (%)

Average Specificity (%)

Orbital fracture

99

0.96

83

100

100

CT Findings

Average PPV (%) Average NPV (%)

Average Accuracy (%)

94

95

Intraocular hemorrhage

96

0.87

50

100

100

74

80

Lens dislocations or destruction

97

0.92

92

99

97

98

97

Intraocular foreign body

97

0.78

64

99

90

96

96

Intraocular gas

97

0.84

100

99

88

100

99 96

Shallow anterior chamber

95

0.75

80

98

89

97

Globe deformity or wall irregularity

89

0.73

57

93

83

78

79

Eyelid hematoma or swelling

86

0.63

77

93

97

63

82

Globe rupture

86

0.70

76

85

80

82

81

AJR:202, May 2014 1103

Downloaded from www.ajronline.org by East Carolina University on 06/09/14 from IP address 150.216.68.200. Copyright ARRS. For personal use only; all rights reserved

Yuan et al.

A

B

Fig. 3—False-negative diagnosis of shallow anterior chamber (AC), globe deformity and irregular wall, intraocular hemorrhage, and globe rupture on CT scans of 40-year-old man. A, Axial unenhanced CT scan shows slight swelling of left eyelid (curved arrow). Ophthalmologic examination shows AC of right globe (straight arrow) is normal but AC of left globe is shallow. B, Shallow AC of left globe (arrow) is missed on CT scan by radiologists because both lenses are not in same plane on images. C, Photograph from ophthalmologic examination shows intraocular hemorrhage (hyphema, white arrows) and iris prolapse from cornea perforation of left globe (black arrow), which were not identifiable on CT images and caused false-negative CT diagnosis of globe rupture.

C ing diagnosed on CT that were not identifiable on clinical examination. There were no false-positive CT results regarding orbital fracture or intraocular hemorrhage. The false-negative findings on CT included four slight depression fractures of the orbital walls that were diagnostic misses; 14 intraocular hemorrhages that were not identifiable on CT scans and three other intraocular hemorrhages that were diagnostic misses; one case of lens destruction that was not identifiable on CT and another case of lens destruction that was a diagnostic miss; two intraocular foreign bodies that were not identifiable on CT scans because of their nonmetallic nature and another foreign body masked by hyperdense intraocu-

1104

lar hemorrhage that was a diagnostic miss; one shallow AC that was not identifiable on CT because of a limitation of CT resolution and two other shallow ACs that were diagnostic misses because both lenses were not in the same plane on the images (Fig. 3); 12 ruptured global wall irregularities (resulting from cornea or sclera full-thickness wounds with an iris or uvea prolapse or vitreous incarceration) that were not identifiable on CT scans and another global wall irregularity that was a diagnostic miss; and seven eyelid hematomas or swellings were not obviously identifiable on CT images and another 11 eyelid hematomas or swellings that were diagnostic misses. There were no false-negative findings due to intraocular gas.

Eleven false-positive globe ruptures diagnosed on CT resulted from three nonruptured globes with true-positive findings of lens dislocation and eight nonruptured globes with false-positive findings of a globe deformity or wall irregularity, intraocular foreign body, intraocular gas, and shallow AC. Nine false-negative globe ruptures on CT resulted from nine ruptured globes with false-negative CT features of a globe deformity or wall irregularity, shallow AC, or intraocular foreign body. Slight depression fractures of the orbital walls and gas bubbles attached to the globe surface were definitively detected by magnification and detailed readings of the orbital CT images.

AJR:202, May 2014

Downloaded from www.ajronline.org by East Carolina University on 06/09/14 from IP address 150.216.68.200. Copyright ARRS. For personal use only; all rights reserved

CT Analysis of Globe Rupture Clinical ophthalmologic examination or surgical correlation detected several falsenegative diagnostic CT features such as intraocular hemorrhage (Fig. 3C), nonmetallic intraocular foreign body, slight destruction at the anterior surface of the lens, shallow ACs, cornea or sclera perforation with iris or uvea prolapse (Fig. 3C), and vitreous incarceration. Discussion Regarding the diagnosis of globe rupture, CT provided an average accuracy of approximately 81%, based on the results of this study. The three most common features of globe rupture were eyelid hematoma or swelling, globe deformity or wall irregularity, and intraocular hemorrhage. In most cases of globe rupture, the presence of one or more of these findings was associated with other abnormalities. Eyelid hematoma or swelling and orbital fractures did not qualify as clinical or CT criteria for globe rupture (p > 0.05). The diagnostic features of globe rupture on CT included intraocular hemorrhage, lens dislocation or destruction, intraocular foreign body, intraocular gas, shallow AC, and globe deformity or wall irregularity (p < 0.05). The diagnostic features of globe rupture on CT had variable distributions with percentages ranging from 18% to 88%. Based on the readings of the two radiologists, the diagnostic CT features with fewer false-negative and fewer false-positive results had higher sensitivity, specificity, and accuracy. An examination of the cases in which both false-negative and false-positive readings were made in this study showed that such errors can be reduced by a thorough ophthalmologic examination and detailed, systematic reading of the CT images. We found an average sensitivity, specificity, PPV, and NPV of CT for the detection of globe rupture of 76%, 85%, 80%, and 82%, respectively (Table 1), and our findings are consistent with those of other studies [3, 19, 20]. Arey et al. [20] reviewed the records of 46 patients (48 eyes), of which 34 eyes were found to have globe rupture on surgical exploration, and reported a sensitivity for determining globe rupture that ranged from 56% to 68% for different observers, a specificity that ranged from 79% to 100%, a PPV that ranged from 86% to 100%, and an NPV that ranged from 42% to 50%. Analysis also revealed that changes in globe contour, obvious volume loss, lens dislocation or absence, vitreous hemorrhage, and retinal detachment were significant predictors of rupture. Similarly, Joseph et al. [19] reviewed the re-

cords of 200 patients who underwent CT for evaluation of ocular trauma and reported a sensitivity and specificity of 75% and 93%, respectively, and a PPV of 88–97% for the diagnosis of globe rupture. Hoffstetter et al. [15], in a study of 59 patients with severe ocular trauma, reported approximately one third of the cases with an unclear clinical diagnosis of globe rupture and in the cases that were accurately diagnosed, globe deformation and volume reduction were the most common findings. These studies indicate that although CT is extremely useful in the case of ocular trauma, it should not be solely relied on for the diagnosis of globe rupture because of the potentially catastrophic consequences of an undiagnosed injury [4]. Early detection of globe rupture after trauma is important so that surgical intervention can be performed promptly. Orbital CT can aid ophthalmologists in the diagnosis of globe rupture. However, orbital CT was not primarily used to triage patients in our hospital. If an ophthalmologist could make a definite diagnosis of ocular or globe injuries independently, orbital CT was not necessary. Hence, the goal of orbital CT in our emergency department was to assist the ophthalmologist in detecting findings (except eyelid hematoma) in patients with a difficult ophthalmologic examination and to provide diagnosis of other lesions. In our study, 14 of 76 (18%) injured globes did not receive a complete ophthalmologic examination and needed orbital CT because of severe eyelid swelling or lacerations, total hyphema, poor patient cooperation, or unconsciousness. Other significant lesions included retrobulbar hemorrhage or edema; facial or orbital fractures; herniation of the inferior or medial rectus muscle; intraocular foreign bodies; paranasal hemorrhage and intracranial hemorrhage; and even optic nerve transection, which might be missed by ophthalmologists and might sometimes require emergent surgical repair. For example, the preoperative determination of the number and location of intraocular foreign bodies on orbital CT assisted in the decision-making process regarding the route of extraction—that is, through the penetrating wound, via an enlarged sclerotomy in vitrectomy; through another limbus incision wound; or even using an “open-sky” procedure. CT assessment of orbital trauma should include axial and coronal images with soft-tissue and bone window settings. Axial images are necessary for assessing the medial and

lateral orbital walls and the lateral and medial rectus muscles. Coronal images are essential in estimating the superior and inferior globe surfaces, the superior and inferior orbital walls, and the superior and inferior rectus muscles and in evaluating for the presence of hematoma of the optic nerve sheath [6]. A direct coronal CT image is usually difficult to obtain in patients with altered consciousness, a head injury, and limited neck mobility and in young children. Coronal reformatted images should be obtained in these patients. A systematic review of the images is also essential so that less apparent abnormalities are not missed [7]. Coronal, sagittal, or even oblique multiplanar images can be generated directly from the axial raw data in MDCT scans obtained with a short acquisition time. In ocular or globe injuries, sagittal and other oblique reformatted CT images can provide additional data to evaluate each abnormal finding to reduce possible false diagnoses. Regarding ACD measurements, in particular, both false-negative and false-positive readings of shallow ACs resulted when both lenses were not in the same plane on the CT images. If sagittal sections or other oblique reformatted images had been provided in the optimal planes, these misinterpretations might have been avoided. However, the technologists would be burdened with taking additional time to perform the multiplanar reconstructions. Radiologists would also be slowed by the expansion in the number of images to be reviewed. Therefore, a better strategy for routine orbital CT would be to provide axial, coronal, and additional sagittal images. The AC is the space between the cornea and the lens that contains aqueous humor, and the iris secretes aqueous humor into the ACs and posterior chambers [2]. The maximum mean depth of the normal AC ranges from 2.4 to 3.5 mm, and the depth of the AC can vary according to patient age and sex and the method of measurement [1, 3]. Our results showed a mean depth of the normal AC of 3.1 mm, a result consistent with prior studies [1, 3]. Joseph et al. [19] graded the altered ACD on CT scans as being flat, shallow, or deepened and reported that the sensitivity of an altered ACD for the diagnosis of globe rupture ranged from 48% to 86%. However, those authors evaluated ACD only qualitatively on CT images without performing direct measurements on the images. Our results indicated that the ACD was signifi-

AJR:202, May 2014 1105

Downloaded from www.ajronline.org by East Carolina University on 06/09/14 from IP address 150.216.68.200. Copyright ARRS. For personal use only; all rights reserved

Yuan et al. cantly different between contralateral eyes when one globe was ruptured and that a difference in ACD of 0.8 mm or greater was diagnostic of globe rupture. Kim et al. [3] reported that a difference in ACD of 0.4 mm or greater between the two eyes was 73% sensitive and 100% specific for the identification of globe rupture. The difference between our findings and those of Kim et al. may be because of variability between observers, different study materials and methods, and different CT parameters. Weissman et al. [1] reported three cases in which a CT finding of a deep AC was useful for the identification of globe rupture at the posterior sclera. A small rupture of the posterior sclera behind the ciliary body may result in prolapsed and decompressed vitreous through the defect, causing the lens to sink backward slightly, which deepens the AC [21]. We did not find a deep AC in any patient in this study for two possible reasons. First, the ACD difference between ruptured and nonruptured globes may have been too small to identify by ophthalmologic examination and CT scans. Second, posterior sclera rupture may combine with destruction or dislocation of the lens rather than retropulsion of the lens. In our study, the 33 ruptured globes included 12 (36%) that were deformed and had shrunk, five (15%) that were deformed and enlarged, and 16 (48%) that were normal size on CT scans. All enlarged ruptured globes showed intraocular hemorrhage. A large amount of intraocular bleeding may cause increased intraocular pressure [3, 19], resulting in an enlarged, ruptured globe. However, our results suggest that a ruptured globe tends to remain a normal size or to decrease in size. A ruptured globe can maintain its normal size if the perforation or laceration is sealed immediately by the cornea or sclera or by iris or uvea herniation after globe rupture. On the other hand, a ruptured globe could become deformed and smaller when the vitreous prolapses through the perforation with decompression of the globe [1, 3, 19]. However, intraocular hemorrhage could sometimes maintain a normal globe size when the vitreous is prolapsed. Concerning radiation-related cancer and genetic effects, the radiologist or technologist should modulate CT parameters to decrease radiation dose in orbital or ocular CT, particularly for patients less than 15 years old [18, 22]. A decrease in CT parameters, such as tube current, kilovoltage, scanning

1106

length, and gantry rotation time (faster rotation time), can reduce radiation dose. An increase in table speed, section thickness, and pitch can also decrease radiation exposure [18, 22]. Because the globe is small, with less contrast on CT scans, synchronous changes of several CT parameters in one orbital or ocular CT study are expected to increase image noise and decrease global lesion conspicuity. Of the 21 patients who had measurements of CTDIvol and DLP, the two patients with the lower tube currents of 89 and 184 mA, respectively, received the lowest effective dose. Decreasing the tube current by 50% will reduce the radiation dose by half [18, 22]. Hence, to promote orbital or ocular CT dose reduction without affecting image quality, the first step should be to restrict the examination or shorten scanning time or length. The next step should be to decrease the tube current with the other parameters being fixed [18, 22]. In addition, application of several modern dose reduction techniques, including automatic exposure control, angular and longitudinal (x-y plane and z-axis) tube current modulation, the organ-based tube current system, and the iterative reconstruction technique, could decrease radiation dose in head CT studies without affecting image quality [23–26]. However, an optimized low-dose orbital or ocular CT protocol still needs further research. Our study had several limitations including its retrospective nature and the relatively small number of patients analyzed. In addition, traumatic mechanisms, treatments, and final visual acuity of globe ruptures were not correlated with CT findings. Last, a detailed analysis of CT dosimetry data could not be performed because of insufficient data. Future studies that improve on these limitations are needed. Conclusions Although globe rupture is uncommon, its consequences are devastating if not promptly diagnosed and treated. The main CT features of a ruptured globe include globe deformity or wall irregularity, destruction or dislocation of the lens, intraocular hemorrhage, intraocular foreign body, shallow AC, and intraocular gas. CT provided an overall accuracy of 81% in the diagnosis of globe rupture. Reference to ophthalmologic examinations and avoidance of CT misinterpretations can improve the diagnostic rate of globe rupture based on CT.

Acknowledgment We thank Cheng-Fen Chiu for her assistance in the preparation of CT brand names and revision of CT parameters in the manuscript. References 1. Weissman JL, Beatty RL, Hirsch WL, Curtin HD. Enlarged anterior chamber: CT finding of a ruptured globe. AJNR 1995; 16:936–938 2. Rahman NU, Jamjoom A, Jamjoom ZA, Abu elAsrar A. Orbito-cranial injury caused by penetrating metallic foreign bodies: report of two cases. Int Ophthalmol 1997; 21:13–17 3. Kim SY, Lee JH, Lee YJ, et al. Diagnostic value of the anterior chamber depth of a globe on CT for detecting open-globe injury. Eur Radiol 2010; 20:1079–1084 4. Bord SP, Linden J. Trauma to the globe and orbit. Emerg Med Clin North Am 2008; 26:97–123 5. Mader TH, Werner RP, Chamberlain DG. Corneal perforation and delayed anterior chamber collapse from a devil’s club thorn. Cornea 2008; 27:961–962 6. Lee HJ, Jilani M, Frohman L, Baker S. CT of orbital trauma. Emerg Radiol 2004; 10:168–172 7. Kubal WS. Imaging of orbital trauma. RadioGraphics 2008; 28:1729–1739 8. Chandra A, Mastrovitch T, Ladner H, Ting V, Radeos MS, Samudre S. The utility of bedside ultrasound in the detection of a ruptured globe in a porcine model. West J Emerg Med 2009; 10:263–266 9. Zhang Y, Cheng J, Bai J, et al. Tiny ferromagnetic intraocular foreign bodies detected by magnetic resonance imaging: a report of two cases. J Magn Reson Imaging 2009; 29:704–707 10. Gunenc U, Maden A, Kaynak S, Pirnar T. Magnetic resonance imaging and computed tomography in the detection and localization of intraocular foreign bodies. Doc Ophthalmol 1992; 81:369–378 11. Patel SN, Langer PD, Zarbin MA, Bhagat N. Diagnostic value of clinical examination and radiographic imaging in identification of intraocular foreign bodies in open globe injury. Eur J Ophthalmol 2012; 22:259–268 12. Lakits A, Prokesch R, Scholda C, Bankier A. Orbital helical computed tomography in the diagnosis and management of eye trauma. Ophthalmology 1999; 106:2330–2335 13. Go JL, Vu VN, Lee KJ, Becker TS. Orbital trauma. Neuroimaging Clin N Am 2002; 12:311–324 14. Hopper RA, Salemy S, Sze RW. Diagnosis of midface fractures with CT: what the surgeon needs to know. RadioGraphics 2006; 26:783–793 15. Hoffstetter P, Schreyer AG, Schreyer CI, et al. Multidetector CT (MD-CT) in the diagnosis of uncertain open globe injuries. Rofo 2010; 182:151–154 16. Osborne DR, Foulks GN. Computed tomographic analysis of deformity and dimensional changes in

AJR:202, May 2014

Downloaded from www.ajronline.org by East Carolina University on 06/09/14 from IP address 150.216.68.200. Copyright ARRS. For personal use only; all rights reserved

CT Analysis of Globe Rupture the eyeball. Radiology 1984; 153:669–674 17. Christner JA, Kofler JM, McCollough CH. Estimating effective dose for CT using dose-length product compared with using organ doses: consequences of adopting International Commission on Radiological Protection publication 103 or dualenergy scanning. AJR 2010; 194:881–889 18. Zacharias C, Alessio AM, Otto RK, et al. Pediatric CT: strategies to lower radiation dose. AJR 2013; 200:950–956 19. Joseph DP, Pieramici DJ, Beauchamp NJ Jr. Computed tomography in the diagnosis and prognosis of open-globe injuries. Ophthalmology 2000; 107:1899–1906

20. Arey ML, Mootha VV, Whittemore AR, Chason DP, Blomquist PH. Computed tomography in the diagnosis of occult open-globe injuries. Ophthalmology 2007; 114:1448–1452 21. Cherry PM. Rupture of the globe. Arch Ophthalmol 1972; 88:498–507 22. Kalra MK, Maher MM, Toth TL, et al. Strategies for CT radiation dose optimization. Radiology 2004; 230:619–628 23. Tan JS, Tan KL, Lee JC, Wan CM, Leong JL, Chan LL. Comparison of eye lens dose on neuroimaging protocols between 16- and 64-section multidetector CT: achieving the lowest possible dose. AJNR 2009; 30:373–377

24. Yamauchi-Kawaura C, Yamauchi M, Imai K, Ikeda M, Aoyama T. Image quality and agespecific dose estimation in head and chest CT examinations with organ-based tube-current modulation. Radiat Prot Dosimetry 2013; 157:193–205 25. Vazquez JL, Pombar MA, Pumar JM, Del Campo VM. Optimised low-dose multidetector CT protocol for children with cranial deformity. Eur Radiol 2013; 23:2279–2287 26. Vorona GA, Zuccoli G, Sutcavage T, Clayton BL, Ceschin RC, Panigrahy A. The use of adaptive statistical iterative reconstruction in pediatric head CT: a feasibility study. AJNR 2013; 34:205–211

F O R YO U R I N F O R M AT I O N

The AJR has made getting the articles you really want really easy with an online tool, Really Simple Syndication, available at www.ajronline.org. It’s simple. Click the RSS button located in the menu on the right side of the page. You’ll be on your way to syndicating your AJR content in no time.

AJR:202, May 2014 1107