Diagnostic accuracy of inverted grayscale radiography for detection of tibial plateau fractures

Objective: This study aimed to quantitatively evaluate the diagnostic performance of inverted grayscale radiography for detecting tibial plateau fractures. Methods: In this retrospective diagnostic accuracy study, two emergency physicians independently assessed one-view (anteroposterior) knee radiographs in both conventional and inverted grayscale modes in separate, blinded sessions. Spiral knee computed tomography (CT) scans served as the reference standard for fracture confirmation. Sensitivity, specificity, and area under the receiver operating characteristic curve values were calculated. Inter-observer agreement was also assessed using Cohen's kappa. Results: Of 112 included cases, 31 (27.67%) had tibial plateau fractures confirmed on CT. For the first physician, sensitivity and specificity were 54.84% (36.00%-72.70%, CI 95%) and 85.19% (75.60%-92.10%, CI 95%) with conventional images, compared with 35.48% (19.20%-54.60%, CI 95%) and 83.95% (74.10%-91.20%, CI 95%) with inverted grayscale images. For the second physician, sensitivity was identical for both modalities (38.71% (21.90%-57.80%, CI 95%)) while specificity was higher with conventional radiography (100.00% (95.60%-100.00%, CI 95%)) than inverted grayscale radiography (90.12% (81.50%-95.60%, CI 95%)). The area under the operating characteristic curve values were consistently higher for conventional imaging: 0.70 (0.60–0.78, CI 95%) versus 0.59 (0.50–0.68, CI 95%) for the first physician (P-value=0.03) and 0.69 (0.57–0.81, CI 95%) versus 0.64 (0.52–0.76, CI 95%) for the second physician (P-value=0.18). Inter-observer agreement was moderate for conventional radiographs (κ = 0.51) and lower for inverted grayscale images (κ = 0.38). Conclusion: Inverted grayscale radiography did not improve the detection of tibial plateau fractures and was associated with reduced inter‑observer reliability.