Classifying nuclei in histology images
AssigneeVentana Medical Systems, Inc.
InventorsJoerg Bredno, Christophe Chefd'hotel, Kien Nguyen
Life of this patent
Estimating the term…
Abstract
Disclosed, among other things, is a computer device and computer-implemented method of classifying cells within an image of a tissue sample comprising providing the image of the tissue sample as input; computing nuclear feature metrics from features of nuclei within the image; computing contextual information metrics based on nuclei of interest with the image; classifying the cells within the image using a combination of the nuclear feature metrics and contextual information metrics.
Claims32
A computer system for classifying cells within an image of a tissue sample stained in an IHC assay for the presence of at least one biomarker comprising one or more processors and at least one non-transitory computer-readable memory, the at least one memory storing instructions for execution by the one or more processors to cause the one or more processors to:
- compute nuclear feature metrics, wherein the nuclear feature metrics are derived from features within cell nuclei in the image of the tissue sample;
- compute contextual information metrics of nuclei of interest within the image of the tissue sample;
- and
- classify cells within the image of the tissue sample based on the nuclear feature metrics and contextual information metrics.
A computer-implemented method of classifying cells within an image of a tissue sample stained in an IHC assay for the presence of at least one biomarker comprising computing nuclear feature metrics from features of nuclei within the image of the tissue sample;
- computing contextual information metrics based on nuclei of interest within the image of the tissue sample;
- and classifying the cells within the image of the tissue sample using a combination of the nuclear feature metrics and contextual information metrics.
A computer system for classifying cells within an image of a tissue sample stained in an IHC assay for the presence of a PD-L1 biomarker comprising one or more processors and at least one memory, the at least one memory storing non-transitory computer-readable instructions for execution by the one or more processors to cause the one or more processors to:
- run a feature extraction module (101) to derive nuclear feature metrics (111) and contextual information metrics (112) from the image of the tissue sample;
- and run a classification module (102) to classify the cells within the image of the tissue sample, wherein the cells are classified as at least one of positive immune cells, positive tumor cells, negative immune cells, and negative tumor cells, or other cells.
The computer system of 28, wherein the data describing neighboring nuclei are computed through application of one of a context conditional random field method or a context bag of words method.
Dependencies are inferred from the claim text.