OrthoScan AI: A Unified, Interpretable Deep-Learning Pipeline for Scoliosis Severity and Knee Osteoarthritis Grading from Plain Radiographs on Commodity CPUs

Scoliosis and knee osteoarthritis (OA) are among the most frequently radiographed musculoskeletal conditions, yet both are still graded by eye: the Cobb angle for spinal curvature and the Kellgren–Lawrence (KL) scale for knee OA carry well-documented inter-observer variability, and specialist readers are scarce outside large hospitals. We present OrthoScan AI, a single screening pipeline that accepts an unlabeled radiograph, decides automatically whether it shows a spine or a knee, and returns a clinically anchored severity output—a Cobb-angle band with an angle estimate, or a KL grade—together with a Grad-CAM explanation, on a CPU-only server (0.3–0.4 s per image without, and 0.7–1.2 s with, the explanation on a 12-core laptop). The two diagnostic branches share one preprocessing chain (border cropping, CLAHE, edge-preserving denoising and letterboxing) and one transfer-learning recipe (frozen-trunk warm-up, discriminative learning rates with cosine decay, class-balanced sampling, label smoothing and early stopping on balanced accuracy). We evaluate on three public datasets. On the Osteoarthritis-Initiative-derived KL-graded knee set of Chen et al. (8,260 images, fixed split) a single DenseNet-121 reaches 62.3 ± 0.6% five-class accuracy, 67.0 ± 0.9% balanced accuracy and a quadratic-weighted κ of 0.79 over three seeds, with 94% of predictions within one grade and KL-4 recall of 0.88; a pre-declared soft-vote ensemble of the nine trained knee networks raises this to 67.6% accuracy, 71.0% balanced accuracy and κ = 0.84 (McNemar p < 10−4 ), and an ablation with paired tests shows that balanced sampling alone recovers minority-grade recall while stacking it with loss re-weighting over-corrects. On 338 real clinical spine radiographs the system detects scoliosis with 96.8% sensitivity (pooled five-fold accuracy 95.9%, AUC 0.986). On the 20,000-image Spinal-AI2024 synthetic set the learned angle head estimates the major Cobb angle with a mean absolute error of 2.6◦ (89% within 5◦ ), whereas the weights-free geometric tracer the system uses as a confidence-gated fallback reaches only 11.6◦ , so we characterise it as a coarse screening signal rather than a measurement. Robustness to compression, noise, blur and exposure, Grad-CAM plausibility and per-stage latency are reported. Eval