🤖 AI Summary
This study addresses evaluation bias arising from data leakage in musculoskeletal abnormality detection on elbow X-rays by introducing a leakage-aware, patient-level evaluation protocol. Using the MURA dataset with DenseNet121 as the backbone, the work systematically assesses the impact of both learning-based (e.g., DnCNN) and conventional (e.g., CLAHE) preprocessing strategies on model discriminative performance, calibration, and clinical utility. Comprehensive evaluation at high-specificity operating points employs multiple metrics, including AUROC, PR-AUC, Expected Calibration Error (ECE), and Brier score. Results indicate that preprocessing yields only marginal improvements highly dependent on specific configurations, and the baseline model using raw inputs remains competitive across most metrics, with no single preprocessing method consistently outperforming the baseline.
📝 Abstract
In this study, we examine learned preprocessing pipelines in the context of triage-oriented orthopedic abnormality detection task using elbow radiographs from MURA dataset. The evaluation focuses on patient-level detection of musculoskeletal abnormalities under a leakage-aware protocol. We compare multiple preprocessing pipelines, with and without a lightweight DnCNN module as a learned preprocessing component, to assess their impact on discrimination and calibration. Performance is assessed using discrimination metrics (AUROC, PR-AUC), calibration measures (ECE, Brier score), and validation-selected operating point analysis targeting high specificity. Results show that differences across preprocessing strategies are modest and configuration-dependent, with no consistent discrimination advantage over the raw-input DenseNet121 baseline. The raw and diverse inputs combined with the DnCNN front-end showed reduced ECE and Brier score, while CLAHE combined with DnCNN did not improve calibration. Overall, the results suggest that under patient-level evaluation, preprocessing gains are modest and configuration-dependent; the raw-input DenseNet121 baseline remains competitive throughout, and no tested preprocessing strategy produced a consistent discrimination advantage across all metrics.