Abstract
Background Radiographically occult hip and pelvic fractures are relatively common in older adults after low-energy trauma, and delayed or missed diagnosis is associated with worse outcomes. MRI is the preferred modality for detection, while CT remains important for fracture delineation, surgical planning, and use in patients with MRI contraindications. However, using both modalities increases cost, imaging time, and strain on imaging services. MRI-based pseudo-CT (pCT) methods derived from 3D T1-volumetric interpolated breath-hold examination (VIBE) data have shown promise in other settings, but their role in evaluating occult hip and pelvic fractures has not yet been explored. This study aimed to explore the clinical performance and usability of MR-based pCT for detecting and characterizing radiographically occult hip and pelvic fractures. Specifically, the study aimed to compare pCT and conventional CT in terms of fracture detection, fracture characterization, diagnostic confidence, and interobserver agreement. Methodology This retrospective study included 22 patients (18 men, 4 women; mean age = 75.7 ± 10.6 years) with a total of 12 radiographically occult hip and pelvic fractures. All patients underwent the same-day MRI and CT imaging, including 3D T1-VIBE sequences processed with grayscale inversion to generate pCT images. Two radiologists independently evaluated both imaging modalities in a blinded and randomized fashion for fracture presence, location, and characteristics. Diagnostic confidence was assessed using a five-point scale. The reference standard was established by two senior radiologists using all available clinical and imaging data. Statistical analyses were performed to evaluate diagnostic performance, diagnostic confidence, and interobserver agreement. Results pCT showed diagnostic performance not inferior to CT for both readers (R1 and R2) (predefined non-inferiority Δ = -10%; R1 Δ = +18.2% (95% confidence interval (CI) = -3.7% to +18.2%), p = 0.0026; R2 Δ = +27.3% (95% CI = 2.2% to +27.3%), p = 0.0010). pCT detected all CT-apparent fractures and showed perfect concordance with CT in fracture localization and characterization. There was no statistically significant difference in diagnostic performance between pCT and CT, although pCT for both readers showed a trend toward higher sensitivity (pCT vs. CT; R1 90.0% vs. 60.0%; R2 100% vs. 60.0%), lower specificity (pCT vs. CT; R1 91.7% vs. 100%; R2 83.3% vs. 100%), and higher accuracy (pCT vs. CT; R1 90.9% vs. 81.8%; R2 90.2% vs. 81.8%). There was no significant difference in diagnostic confidence between the two modalities, with pCT demonstrating a trend toward slightly higher confidence when ruling in fractures and slightly lower confidence when ruling out. Interobserver agreement was perfect for CT (κ = 1.00) and moderate for pCT (κ = 0.55). Conclusions Compared with conventional CT, 3D T1-VIBE MR-based pCT demonstrated diagnostic performance that was not inferior for detecting radiographically occult hip and pelvic fractures, with perfect concordance in fracture characterization. pCT enables comprehensive fracture assessment within a single MRI examination, and our findings suggest that pCT has the potential to serve as a viable alternative to CT, enabling a "one-stop-shop" imaging approach in appropriate clinical scenarios.