Computerized
Pergamon
Medical Imaging and Graphics, Vol. 18, No. 6, pp. 423-427, 1994 Copyright 0 1994 Elsevier Science Ltd Printed in the USA. All rights reserved 0895-61 I l/94 $6.00 + .OO
08956111(94)00026-3
COMPARISON OF SPIRAL CI- AND CONVENTIONAL CT IN 3D VISUALIZATION OF FACIAL TRAUMA: WORK IN PROGRESS Richard
l’ello’, James Suojanen,
Department
Philip Costello,
and Anne McGinnes
of Radiological Sciences, Harvard Medical School, Deaconess 185 Pilgrim Road, Boston, MA 022 15, USA
Hospital,
(Received 8 July 1994; revised 22 February 1994) Abstract--Spiral computed tomography (SCT) differs from conventional CI (CCT) in that regions of the body can be rapidly imaged via continuous scanning. This is accompanied by simultaneous advancement of the patient, thus allowing volumetric data acquisition of up to 60 cm in less than a minute. Thus motion is minimized and slice misregistration is minimized when multiplanar and three dimensional reconstructions are performed. This paper compares the utility of SCC and CCT in facial trauma. A total of six patients were studied with either CCT or SCT of the face after trauma. !XT scans were obtained using a Siemens Somatom Plus-S CT scanner (2 mm thick collimation and 3 mm,/sec table feed for 32 s). Three-dimensional (3D) and multiplanar reconstruction images of the facial bones were generated after appropriate thresholds were selected by the radiologist; similar images generated with a CCT (GE quick 9800) were compared using a three point scale with kappa analysis. XX is able to generate axial and reformatted images of comparable quality to CCT (k = 0.47-0.89) in less than half the time to perform an examination (26 min vs 63 min). SCT can rapidly produce (3D) and multiplanar reformatted images of facial trauma with minimal motion, or misregistration artifact when compared to CCT. Key Words: Computed
tomography,
Technology,
Computed
tomography,
Facial trauma,
CT, Spiral CT
short time periods are possible. This eliminates the interscan delay for table translation and repositioning that occurs with conventional CT. Acquired data can be reconstructed in the axial plane at variable section intervals at any point in the scan cycle, with no apparent difference in spatial resolution between conventional and volumetric scans (3). In addition the rapid acquisition minimizes the opportunity for motion artifact by reducing imaging time to less than a minute. The purpose of this paper is to report our experience and technique with SCT in the assessment of facial fractures and compare with conventional facial CT.
INTRODUCHON Open reduction and internal fixation of facial fractures demand a detailed understanding of the three dimensional pattern of injury. Three dimensional CT can provide superior definition of fracture lines and the extent of comminution. Tlhis added information makes preoperative planning more accurate and thus facilitates surgical intervention (1). Similarly three-dimensional (3D) reconstruction can be efficacious in planning facial surgery for other craniofacial malformations (2). Spiral computed tomography (SCT) differs from conventional CT (CCT) in that the entire head can be continuously imaged in the axial plane without a time gap in slice acquisition. SCT involves continuous scanning due to slip ring technology while simultaneously advancing the patient through the gantry thus allowing a volumetric acquisition up to 60 cm long. Slip ring technology replaces the multitude of cables from the X-ray tube and detectors to the scan gantry with contacts similarly to an electric trolly. Consequently with the elimination of these cables the need to reverse gantry rotation direction to avoid cable tangling is obviated. With slip ring scanners, continuous X-ray output and continuous table translation during
’To whom correspondence
Three-dimensional,
Methods During the period 6/l/9 1 to 4130193 three patients with facial trauma referred for CT at our institution were scanned with SCT. During the period 1 l/1/92 to 1 l/27/92 three patients referred for facial CT at another site were scanned with CCT. All cases were overseen and reconstructed by the same radiologist (RT). In the SCT the patient is placed on the CT table (Somatom Plus, Siemens Medical Systems, Erlangen, Germany) in the supine position with the arms by the side and a safety strap is secured around the mid thorax to support the arms. A lateral topogram is taken and the gantry is angled 0” to the Infraorbital-meatal line. Spiral CT (120 kVp, 165 mA) is performed with a 3 mm per s table feed for 32 s with a mean beamwidth of 2 mm
should be addressed.
423
424
November-December/l994,
Computerized Medical Imaging and Graphics
Table 1. Study scores and time to perform
Volume 18, Number 6
each study.
Case
Total time for CT
Axial score
Coronal score coronal CCT & MPR SCT
3D score
BT (spiral) HL (spiral) CK (spiral) TF (conventional) CR (conventional) JS (conventional)
25 min 32 min 21 min 55 min 61 min 73 min
I,2 2,1 I,1 2,2 131 ],I
2,2 2,2 2,2 2,2 18 1,1
O,l 2,2 2,2 2,2 0 I,1
in order to cover the complete facial region in a single spiral acquisition. Axial images are reconstructed at 1 mm intervals with the high resolution algorithm for multiplanar reformation (MPR) analysis of the bone data and with the medium (soft) resolution algorithm for three dimensional image generation. Coronal MPR was performed using the vendor supplied software and 1 mm thick slices reconstructed at 1.5 mm intervals. Paraoblique MPR were also generated in cases to evaluate the orbital floor as 1 mm thick slices reconstructed at 1.5 mm intervals. CCT was performed on a GE Quick CT 9800 (General Electric, Milwaukee, Wis) with the patient in the supine position with the arms by the side. A lateral topogram is taken and two consecutive studies with the gantry angled 0” and 90” (with repositioning and extension of the head for the coronal images) to the infraorbital-meatal line are obtained. Contiguous scans of the face were obtained using 1.5 mm collimation (140 kVp, 170 mA) and a 2 s scan duration with 8 s interscan delay. Images were reconstructed using a high resolution algorithm. Multiplanar images were not generated since they were not considered part of the standard facial trauma CT protocol. CCT and SCT studies were photographed on both an extended bone (window width 3000, window length +350 Hounsfield Units [HI) and soft tissue (window width 400, window length +50H) windows. The (3D) reconstruction was done after an appropriate threshold (usually 150 to 300 H) was selected by the radiologist after reviewing all images. Quality of studies was rated using a three point scale; 0 = not satisfactory for diagnosis, 1 = adequate image quality for diagnosis, 2 = excellent image quality, where the difference between a rating of 1 and 2 was determined based on background noise suppression and contiguity of bony structures. Three dimensional studies, coronal images (direct coronal on CCT and MPR coronal on SCT), and axial images were all scored independently. MPR’s in paraoblique planes were not formally scored nor was a comparison between SCT MPR and CCT MPR images performed due to inability to remove observer bias in choosing MPR planes. The time for all the ac-
quisitions was measured as the total time that the patients were on the CT table and that the scanner was unavailable to perform another patient study. All patients had surgical confirmation of their diagnosis. Quality of the studies was evaluated by two reviewers who were blind to the patients surgical results and each others reading with kappa statistic measured (4). RESULTS The CT studies results are summarized in Table 1. SCT studies took a mean of 26 min while CCT studies took a mean of 63 min, of this time set up took 5 min on each scanner and the remaining time was the total time that the scanner was dedicated to complete processing of each individual case. The quality of the axial studies was rated similarly on both modalities, and the multiplanar reconstruction images were rated slightly better on the spiral studies due to lack of stairstepping artifact. The quality of the (3D) studies was rated slightly better on the SCT reconstruction images. Case BT is a 2 1-yr-old male with a left orbital floor fracture. The SCT 3D reconstruction did not readily demonstrate the fracture due to partial volume artifact and the use of a high enough threshold to eliminate soft tissues of the face. However the multiplanar reformations readily demonstrate the fracture as shown in Fig. 1. Case HL is a 40-yr-old male with a right mandibular ramus fracture. Note that the SCT 3D reconstruction shown in Fig. 2 easily demonstrates the fracture and with the assistance of the reformatted images it is possible to delineate the dislocation and medial angulation of the condylar fragment. Note that dental amalgam beam hardening artifact has been edited out. Case CK is a 22-yr-old female with a left orbital floor fracture. The SCT 3D reconstruction targeted to the orbit is able to demonstrate the fragments. Oblique parasagittal reformation was able to demonstrate these fragments in great detail without stair-stepping artifact or in-plane partial volume artifact. Case TF is a 40-yr-old male with tripod fracture. The CCT (3D) reconstruction readily demonstrated this fracture shown in Fig. 3.
3D visualization of facial trauma
Fig. 1. Oblique multiplanar
R. TELLO etal.
??
425
reformation of left orbital floor fracture in 2 1-yr-old male with spiral CT.
Case RR is a 35yr-old male with a right orbital floor fracture with fracture of the zygomatic arch. The CCT (3D) reconstruction was suboptimal in demonstrating the orbital fracture diagnosed oh plain film, and the axial and coronal images were necessary to evaluate the zygomatic arch.
Fig. 2. Three-dimensional reconstruction in a 40-yr-old male with a right mandibular ramus fracture with dental amalgam beam hardening artifact edited out from a Spiral CT.
Case JS is a 44-yr-old male with diastasis of the left zygomatic-frontal suture which CCT (3D) reconstruction was able to demonstrate. The diagnostic quality of the reconstruction and the axial images were comparable. DISCUSSION The availability of state-of-the-art CT scanners has altered our approach to the analysis of complex craniofacial trauma. It has become routine in a trauma environment to obtain a set of equally-spaced, nonoverlapping, high resolution CT scans of the face with narrow collimation. Reformatting in the sag&al, coronal and oblique planar orientations can be used to demonstrate aberrant anatomy and facilitate surgical planning and evaluation (5). In particular it has been reported that not only is CT better than pluridirectional tomography at demonstrating fractures of the face (6), but that the coronal sections are most helpful in accurately depicting the structures of the face that are most likely to be injured in trauma (7). In particular the recent work of Levy (8) was able to determine that 3D images add significantly to the two-dimensional 2D CT evaluation of severe facial trauma in 29% of trials, consequently expediting 3D generation in facial trauma would be critical in maintaining diagnostic quality while improving patient throughput. In comparing CCT and SCT it is important that the CCT examination requires two separate examinations with the re-positioning of the patient and angulation of the gantry, thereby more than doubling the
426
Computerized Medical Imaging and Graphics
Fig. 3. Three-dimensional
reconstruction
November-December/
1994, Volume 18, Number 6
in a 40-yr-old male from coronal images of a con-
of a tripod fracture
ventional CT.
examination time. In addition if an acute multiple trauma work-up is to be performed it is not unusual to perform an abdominal CT followed by clearing of the C-spine prior to proceeding with facial CCT so that coronal images may be obtained. Thus, a rapid SCT for facial trauma can be readily added on during the multiple trauma work-up without requiring that the patient be taken off the CT table for clearance of the C-spine prior to repositioning for coronal images. Another reason that SCT is more rapid than CCT in facial trauma evaluation is that the (3D) reconstruction can be carried out faster and more efficiently by maintenance of all images in dynamic memory. The maintenance of the data in dynamic memory thereby obviates the need to reload the data into memory after acquisition for three dimensional reconstruction. CCT did not suffer in (3D) reconstruction using the high resolution algorithm. Though the ability to rapidly reconstruct the complete SCT data set using the soft resolution algorithm undoubtedly contributed to the improved visual appearance of SCT three dimensional images this did not affect diagnostic accuracy which had good to excellent interobserver agreement (kappa 0.47-0.89, Table 2). Note that there is a limitation in orbital floor (i.e., in plane axial) fractures, in that both CCT and SCT (3D) reconstruction’s suffer due to partial volume artifact. However, the SCT coronal reformations can give results comparable to directly acquired coronal CCT. It is also noteworthy that not only is scan time decreased with SCT, potentially trebling patient throughput, but the radiation exposure is also reduced. Ac-
cording to the respective vendors specifications the dosimetry is reduced from two exams at approximately 3000 mr ESE each for CCT (GE Quick CT 9800) to a single exam with 4650 mr ESE for SCT (Siemens Somatom Plus-S) using the described protocols. Though ultimately a phantom study would be beneficial in determining the theoretical and practical limits of quality to be expected from SCT the purpose of this study was to compare SCT with the community standard conventional CT study used for facial trauma evaluation. In addition a more comprehensive study in a blinded randomized fashion would be a next step, however this will have to await the opportunity of having state of the art CCT and SCT at the same institutional site and at the same time, which is not readily possible with and emerging technology such as spiral CT. CONCLUSION These cases highlight (3D) visualization of SCT data to facilitate the solution of complex problems that could require multiple examinations and increased scanning time. Though the ability to make the diag-
Table 2. Inter-observer
agreement
(kappa)
Score
0
1
2
k
0
l/18 l/18 0 .47
l/18 5/18 0 .64
0 l/18 9/18 .I8
.47 56 .89
1 2 k
3D visualization
of facial trauma
noses described in some cases was possible solely on the axial images, the ability to generate three dimensional images without motion induced mis-registration allowed for reduced ambiguity. In addition in concordance with the work of Levy 33% of the cases required 3D and multiplanar reforlmation for diagnosis. The results from our small series suggest that SCT will play an important role in evaluating craniofacial trauma in the future.
Acknowledgments-Portions
of,;his work were presented at the 1993 RSNA. Portions of this work were funded by PHS grant RR 0559 1 and portions of this work were funded by a 1992 RSNA Research Resident Award.
??
R. TELLO et al.
427
7. Gentry, L.R.; Manor, W.F.; Turski, P.A.; Stother, C.M. High resolution CT analysis of facial struts in trauma: 1. Normal anatomy. AJR 140(3):523-32; 1983. 8. Levy, R.A.; Edwards, W.T.; Meyer, J.R.; Rosenbaum, A.E. Facial trauma and 3D reconstructive imaging: Insufficiencies and correctives. AJNR 13:885-892: 1992.
About the Author-RICHARD
TELLO graduated from MIT with BS in mathematics, and BSE in Mechanical Engineering 1982, MSME in mechanical engineering 1983, and is pursuing the Ph.D. He received the MD from Stanford in 1989, and Interned at St. Mary’s Hospital in San Francisco. He has completed a residency in Radiology at New England Deaconess Hospital-Harvard Medical school and is an RSNA research resident award recipient. Research interests include cardiovascular and three dimensional imaging.
About the Author-JAMES
REFEIRENCES 1. Mayer, J.S.; Wainwright, D.J.; Yeakley, J.W.; Lee, K.F.; Harris, J.H.; Kulkarni, M. The role of three dimensional computed tomography in the management of maxillofacial trauma. J. Trauma 28(7):1043-53; 1988. 2. Vannier, M.W.; Marsh, J.L.; Warren, J.O. Three dimensional CT reconstruction Images for craniofacial surgical planning and Evaluation. Radiology 150: 179-184; 1984. 3. Costello, P.; Dupuy, D.E.; .Ecker, C.P.; Tello, R. Spiral CT of the thorax with small volumes of contrast material: A comparative study. Radiology 183663-666; 1992. 4. Fleiss, J.L. The measurement of interrater agreement. In: Statistical methods for rates and proportions. 2nd ed. New York: Wiley 1981: 212. 5. Gentry, L.R.; Manor, W.F.; Turski, P.A.; Strother, C.M. Highresolution CT analysis of facial Struts in Trauma: 2. Osseous and Soft-tissue complications. AJR 140(3):533-541; 1983. 6. Kreipke, D.L.; Moss, J.J.; France, J.M.; Maves, M.D.; Smith, D.J. Computed tomography and thin-section tomography in facial trauma. AJR 142(5): 1041-5; 1984.
SUOJANEN graduated from Harvard University with a BA in 1975 and University of Rochester School of medicine and dentistry in 1979. Since 1990 he has been chief of Neuroradiology at New England Deaconess Hospital and is currently an Instructor of Radiology at Harvard Medical School. He has published extensively in areas related to cocaine induced effects on cerebral circulation, computed tomography, and magnetic resonance imaging.
About the Author-PHILIP
COSTELLO graduated from the University of London Medical School. He pursued his radiology and postgraduate work at the University of Toronto, Canada. Since 1976 he has been a staff radiologist at the New England Deaconess Hospital and is currently an Associate Professor of Radiology at Harvard Medical School. He has published extensively in areas related to thoracic radiology and computed tomography.
About the Author-ANNE MCGINNESgraduated
from Bunker Hill Community College with an associates degree in radiologic technology. She is currently pursuing a Baccalaureate at Emmanuel College in health care administration. She has been at the New England Deaconess Hospital since 1985 and a CT technologist since 199 1. She is interested in new applications of computed tomography.