Fig. 3: Characterizing underperformance of saliency method pipeline. | Nature Machine Intelligence

Fig. 3: Characterizing underperformance of saliency method pipeline.

From: Benchmarking saliency methods for chest X-ray interpretation

Fig. 3: Characterizing underperformance of saliency method pipeline.The alternative text for this image may have been generated using AI.

a, Example CXRs that highlight the three pathological characteristics identified by our qualitative analysis: (1) left, number of instances; (2) middle, size; (3) right, shape complexity. b, Example CXRs with the four geometric features used in our quantitative analysis: (1) top left, number of instances; (2) top right, size = area of segmentation/area of CXR; (3) bottom left, elongation; (4) bottom right, irrectangularity. Elongation and irrectangularity were calculated by fitting a rectangle of minimum area enclosing the binary mask (as indicated by the yellow rectangles). Elongation = maxAxis/minAxis. Irrectangularity = 1 − (area of segmentation/area of enclosing rectangle).

Back to article page