Diagnostic Accuracy of Magnetic Resonance Imaging versus Computed Tomography for Lumbar Spondylolysis in Patients with Chronic Low Back Pain: A Single-Center Retrospective Study

Purpose The objective of this study is to compare the diagnostic efficacy of Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) in detecting lumbar spondylolysis among patients presenting with chronic low back pain, and to determine the rate of missed diagnoses when MRI is reported as "normal". Methods Radiological images of 2,104 patients who presented with chronic low back pain between December 2024 and December 2025 and underwent simultaneous lumbar CT and MRI were retrospectively reviewed. The study group comprised 67 patients diagnosed with spondylolysis based on CT findings, which served as the gold standard. Demographic characteristics, affected spinal levels, presence of degeneration, and MRI findings (classified as no findings, suspicious findings, or definitive findings) were analyzed. Results The prevalence of spondylolysis in the screened population was 3.18%. The mean age of the study cohort was 46.7 ± 11.2 years, with a female predominance (71.6%). The most frequently affected level was L5-S1 (82.1%). In the retrospective MRI evaluation of patients with a definitive CT diagnosis, 41.8% exhibited no pathological signal alterations (such as edema or fracture lines) in the pars interarticularis. Furthermore, an analysis of the hospital registry revealed that the diagnosis of spondylolysis was completely missed (100%) in all routine MRI reports. The rate of missed diagnosis on MRI was significantly higher in younger patients without degenerative changes (51%) compared to those with concomitant degeneration (31%). Conclusion Routine MRI evaluations yield a high rate of false-negative results in the diagnosis of lumbar spondylolysis. A "normal" MRI report is insufficient to rule out spondylolysis, particularly in younger patients lacking degenerative changes. In cases where clinical suspicion persists, CT imaging must be incorporated into the diagnostic algorithm.