The objective of this study was to compare the diagnostic performance and confidence of conventional, optimized, and ultrashort time to echo (UTE) magnetic resonance (MR) protocols for detection of simulated lumbar spondylolysis in human cadavers. In addition, we sought to demonstrate the feasibility of the UTE technique in subjects with and without spondylolysis.
Four human lumbar spine specimens with 46 individual pars interarticularis were randomly left intact (n = 26) or received experimental osteotomy (n = 20) using a microsurgical saw to simulate spondylolysis. The specimens were imaged using a computed tomography (CT) scan along with 3 “Tiers” of MR protocols at 3 T: Tier 1, conventional lumbar MR protocol; Tier 2, optimized conventional protocol consisting of a sagittal oblique spoiled gradient recall echo and axial oblique T1 and short tau inversion recovery sequences; and Tier 3, a sagittal UTE MR sequence. Two blinded readers evaluated the images using a 4-point scale (1 = spondylolysis certainly absent, 2 = probably absent, 3 = probably present, 4 = certainly present) at each individual pars. For each imaging protocol, diagnostic performance (sensitivity, specificity, and area under the receiver operating characteristic curve, using the surgical osteotomy as the reference) and confidence were assessed and compared using the McNemar test. Furthermore, 2 human subjects were imaged with the conventional and UTE MR protocols to demonstrate feasibility in vivo.
Diagnostic performance was moderate for Tiers 1 and 2, with a moderate sensitivity (0.70 to 0.75) and high (1.00) specificity. In contrast, CT and Tier 3 UTE MR imaging had both high sensitivity (1.00) and specificity (1.00). The sensitivities of CT or Tier 3 were statistically greater than Tier 1 sensitivity (P = 0.041) and neared statistical significance when compared with Tier 2 sensitivity (P = 0.074). Area under the receiver operating characteristic curve was also significantly greater for CT and Tier 3 (each area = 1.00), compared with the areas for Tier 1 (0.89, P = 0.037) or Tier 2 (0.873, P = 0.024). Diagnostic confidences of CT or Tier 3 were much greater than other Tiers: Both Tiers 1 and 2 had a large percentage of uncertain (>60%, P < 0.001) or wrong interpretations (>10%, P < 0.001), unlike CT or Tier 3 (0% uncertain or wrong interpretations). Preliminary in vivo UTE images clearly depicted intact and fractured pars.
Our study demonstrated that the detection of pars fractures using a single sagittal UTE MR sequence is superior in performance and confidence to conventional and optimized MR protocols at 3 T, whereas matching those from CT evaluation. Furthermore, we demonstrated the feasibility of in vivo application of the UTE sequence in subjects with and without spondylolysis.
From the *Department of Radiology, University of California, San Diego, La Jolla, CA;
†Institute of Diagnostic and Interventional Radiology, University Hospital Zurich, University of Zurich, Switzerland;
‡Department of Radiology, Siriraj Hospital, Bangkok, Thailand;
§Department of Family Medicine, University of California, San Diego, La Jolla;
∥General Electric Healthcare, San Diego, CA;
¶Department of Neurosurgery, University Hospital Zurich, University of Zurich, Switzerland; and
#Department of Radiology, VA San Diego Healthcare System, San Diego, CA.
Received for publication May 16, 2018; and accepted for publication, after revision, July 9, 2018.
Research reported in this publication was supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases of the National Institutes of Health under award number R01 AR066622 in support of Dr Bae, and award number R01 AR064321 in support of Dr Chung. Furthermore, this work was supported by grant funding from the Swiss National Science Foundation (P2SKP3_168412) and Swiss Society of Radiology in support of Dr Finkenstaedt. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Correspondence to: Won C. Bae, PhD, Department of Radiology, University of California, San Diego, 9427 Health Sciences Dr, San Diego, CA 92093–0997. E-mail: email@example.com.