Abstract

Diagnoses performed on the basis of histopathological evaluation depend on the premise that information derived from a small number of samples is valid for the entire tissue volume. By insufficiently sampling a biopsy volume the ability of pathologists to draw meaningful inferences from the sample is impeded. This work attempts to apply an information theoretic approach to biopsy sampling rates informed by variation in tissue morphology identified by persistent homology. By quantifying the diagnostic information present in a sample may be possible to prevent under sampling by the clinician by creating a “Nyquist limit” for histopathological sampling given the frequency of morphologically distinct regions in a single biopsy.

© 2019 SPIE/OSA