Tumor plus adjacent benign signature (TABS) for quantitative histomorphometry

摘要

Methods, apparatus, and other embodiments associated with predicting prostate cancer (CaP) progression using tumor cell morphology features and benign region graph features are described. One example apparatus includes a set of logics that acquires an image of a region of tissue, detects and segments cells in the image, extracts a set of morphological features from cells in a first region in the image, constructs a graph of a localized cellular network in a second region of the image, extracts a set of graph features from the graph, generates a set of tumor plus adjacent features signature (TABS) features from the sets of graph features and the set of morphological features, and calculates the probability that the image is a progressor or non-progressor based, at least in part, on the set of TABS features. The first region may concern cancerous cells and the second region may concern benign cells.

主权项

1. A non-transitory, computer-readable storage medium storing computer executable instructions that when executed by a computer control the computer to perform a method for predicting prostate cancer (CaP) progression in a patient using field effects in automated quantitative histomorphometry (QH), the method comprising:
accessing a first digital image of a tissue slide, where the first digital image includes a cancerous section of a prostate demonstrating pathology associated with CaP in the patient; detecting a first cell in a first region of interest of the first digital image; segmenting the boundary of the first cell into a set of cell boundary points; extracting a set of morphological features from the set of cell boundary points; accessing a second digital image of a tissue slide, where the second digital image includes a benign section of the prostate; detecting a second cell in a second region of interest in the second digital image; constructing a subgraph of a localized cell network within the second region of interest by linking individual cells located proximal to each other, where the nodes of the subgraph represent individual cell nuclei centroids, and where the edges of the subgraph are defined between pairs of cells by a probabilistic decaying function; extracting a set of subgraph features from the subgraph, where the set of subgraph features characterizes a field effect of the local cell organization; producing a set of tumor plus adjacent benign signature (TABS) features from the set of morphological features and the set of subgraph features; and controlling a computer to predict CaP progression in the patient, based, at least in part, on the set of TABS features.