Single-sequence low-field 3D T1-weighted MRI for scaphoid fracture detection compared to radiography: prospective diagnostic accuracy pilot study

PURPOSE: To determine the diagnostic accuracy of a novel high-resolution 3D T1-weighted gradient-echo sequence in low-field MRI for detecting scaphoid fractures compared with radiography. METHODS: In this prospective single-center study, patients with suspected scaphoid injury underwent wrist radiography and 0.31 T extremity MRI including a 4-min 3D gradient-echo sequence (3D SST1). Three musculoskeletal radiologists independently evaluated radiographs and the isolated 3D sequence in separate sessions, rating fracture presence, displacement/stability, diagnostic confidence (five-point Likert scale), and need for additional imaging. The reference standard was based on CT and standard MRI sequences, evaluated by two experienced musculoskeletal radiologists. RESULTS: Forty patients with 11 scaphoid fractures were included. Across readers, AUC for scaphoid fracture detection was 1.000 for 3D MRI and 0.823 for radiography (AUC-difference 0.177, 95% CI: -0.004-0.359, p = 0.056). For 3D MRI, sensitivity and NPV were 1.000 for all readers. Specificity (0.931-0.966 vs 0.655-0.862) and PPV (0.846-0.917 vs 0.444-0.667) were higher than for radiography. Diagnostic confidence was excellent (median score 1) with near-perfect inter-rater agreement (κ = 0.961). Additional imaging was not recommended in 40/36/37 cases (reader 1/2/3). CONCLUSION: A single 4-min high-resolution 3D low-field MRI sequence enabled highly accurate detection of scaphoid fractures with perfect sensitivity and NPV, a high reader confidence, and an improved performance compared to radiography in this small pilot cohort. This approach may facilitate rapid, radiation-free screening in acute wrist trauma.