Radiology AI Fracture & Study Scanner Benchmarker

Study Radiograph Slice Occult Femoral Neck Fracture

SLICE: 01/01 [AP PELVIS]
KV: 80 | mAs: 28
W: 350 L: 120
ZOOM: 1.0x

AI Heatmap Overlay

Ground Truth Target

Human Review Window

Window Level (L): 120

Window Width (W): 350

AI Heatmap Alpha: 75%

Real-Time Diagnostic Telemetry ai_superior_detection

AI Accuracy
94.8%
Foundation Model Ensemble

Human Accuracy

64.2%

Attending Radiologist Baseline

False Negative Reduction

82.5%

Occult Finding Recovery Rate

Model Confidence

0.94

Target Region Sigmoid Score

AI Confidence Decision Threshold 0.72

High Recall (0.50) Optimal F1 (0.72) High Specificity (0.98)

Clinical Benchmark Verdict

AI successfully identified occult micro-fracture missed by standard review.

Diagnostic Parameter	Foundation AI	Human Baseline	Delta
Sensitivity (Recall)	96.2%	61.5%	+34.7%
Specificity	93.4%	88.1%	+5.3%
Mean Triage Time	1.2s	82.0s	68x Faster
Occult Fracture Recovery	Detected	Missed	Saved Finding