Mujgan Orman, Deniz Celiker, Bera Koca, Deniz Alis, Altug Yucekul, Ahmet Alanay, Ercan Karaarslan
3 T zero echo time MRI showed high inter-rater reproducibility for pars-defect detection. In a small CT-linked subset, specificity was high and sensitivity moderate to high, with a further increase under exploratory combined assessment; these CT-linked estimates are preliminary given the limited number of CT-positive pars. These findings suggest that zero echo time may serve as a reproducible, radiation-sparing adjunct to lumbar MRI rather than a replacement for CT, and require confirmation in larger studies.
OBJECTIVE: To evaluate 3 T zero echo time bone magnetic resonance imaging (MRI) for pars-defect assessment in children, adolescents, and young-adults with low back pain, emphasizing inter-rater reproducibility and computed tomography (CT)-linked feasibility.
MATERIALS AND METHODS: In this single-center retrospective pilot reader study, 40 lumbar MRI examinations (median age, 15.0 years; interquartile range, 13.0-18.3) with zero echo time imaging and sagittal short tau inversion recovery imaging were independently reviewed by two radiologists. Across 480 pars units observations, readers scored pars defects and marrow edema as present or absent. Agreement was summarized using observed agreement, Gwet coefficient 1, and Cohen κ. A CT-linked subset of 11 CT-MRI examinations within 14 days (132 pars units, 14 CT-positive) was analyzed using consensus CT annotations as the reference standard.
RESULTS: For zero echo time pars-defect detection, observed agreement was 0.983, Gwet coefficient 1 was 0.981 (95% confidence interval, 0.967-0.993), and Cohen κ was 0.853 (95% confidence interval, 0.728-0.949). For short tau inversion recovery edema, values were 0.973, 0.970 (95% confidence interval, 0.949-0.987), and 0.731 (95% confidence interval, 0.558-0.878). In the CT-linked subset, sensitivity/specificity for zero echo time were 71.4%/100.0% for Reader 1 and 85.7%/99.2% for Reader 2; combined assessment increased sensitivity to 85.7% and 92.9% while maintaining high specificity.
CONCLUSION: 3 T zero echo time MRI showed high inter-rater reproducibility for pars-defect detection. In a small CT-linked subset, specificity was high and sensitivity moderate to high, with a further increase under exploratory combined assessment; these CT-linked estimates are preliminary given the limited number of CT-positive pars. These findings suggest that zero echo time may serve as a reproducible, radiation-sparing adjunct to lumbar MRI rather than a replacement for CT, and require confirmation in larger studies.