Resource-frugal classification and analysis of pathology slides using image entropy

Pathology slides of lung malignancies are classified using resource-frugal convolution neural networks (CNNs) that may be deployed on mobile devices. In particular, the challenging task distinguishing adenocarcinoma (LUAD) and squamous-cell carcinoma (LUSC) cancer subtypes is approached in two stages. First, whole-slide histopathology images downsampled to a size too large for CNN analysis but enough retain key anatomic detail. The decomposed into smaller square tiles, which sifted based their image entropies. A lightweight produces tile-level classifications aggregated classify slide. resulting accuracies comparable those obtained with much more complex CNNs larger training sets. To allow clinicians visually assess basis classification -- is, see regions underlie it color-coded probability maps created by overlapping tiles averaging probabilities at pixel level.