US 10,380,491 B2Granted

System for predicting the recurrence of cancer in a cancer patient

AssigneeDefiniens AG

InventorsNathalie Harder, Maria Athelogou

Life of this patent

Estimating the term…

Loading patent…

Abstract

A system predicts the recurrence of cancer. A first slice of a prostate tissue sample is stained so that luminal epithelial cells and basal epithelial cells are stained different colors. A first digital image is taken of the first slice. The second slice of the sample is stained so that M1 type macrophages and M2 type macrophages are differentially stained. A second digital image is taken of the second slice. The system analyzes the first digital image and defines regions of non-intact glands. Intact gland regions are then determined, and regions of stroma are identified. The system defines influence zones between non-intact regions and stroma regions. Using information from the second image, macrophages in the tissue corresponding to the influence zones are identified and counted. Based at least in part on this count, the system determines a score. The score is indicative of whether the patient will experience PSA recurrence.

Claims18

Dependent claims

A method comprising:

generating first objects from a first digital image of a first tissue slice of a cancer patient, wherein the first tissue slice has been stained with a first stain that stains basal epithelial cells and a second stain that stains luminal epithelial cells, and wherein the first objects have been stained with the first stain;
generating second objects from the first digital image of the first tissue slice, wherein the second objects have been stained with the second stain;
defining third objects that include only those second objects that have more than a minimum separation from any of the first objects;
generating a scoring region that includes the third objects;
generating fourth objects from a second digital image of a second tissue slice of the cancer patient, wherein the second tissue slice has been stained with a third stain that stains tumoricidal M1 macrophages and a fourth stain that stains tumorigenic M2 macrophages, and wherein the fourth objects have been stained with the third stain;
generating fifth objects from the second digital image of the second tissue slice, wherein the fifth objects have been stained with the fourth stain;
co-registering the first digital image with the second digital image and transferring the scoring region to the second digital image;
defining an influence zone around the scoring region on the second digital image;
counting a first number of the fourth objects and a second number of the fifth objects that are located within the influence zone;
and
determining a score based on the first number and the second number, wherein the score indicates whether the cancer patient will likely have a recurrence of cancer.

8 more elements

A method involving a Prostate-Specific Antigen (PSA) recurrence prediction system, the method comprising:

(a) duplex staining a first sample of glandular tissue from a cancer patient with a first pair of biomarker stains thereby obtaining a first stained tissue sample;
(b) using a digital camera to generate a first digital image from the first stained tissue sample;
(c) duplex staining a second sample of glandular tissue from the prostate cancer patient with a second pair of biomarker stains thereby obtaining a second stained tissue sample;
(d) generating a second digital image from the second stained tissue sample;
(e) generating first objects from the first digital image, wherein each first object corresponds to a non-intact gland in the glandular tissue of the patient;
(f) defining a non-intact region mask from the first objects;
(g) generating second objects from the first digital image, wherein each second object corresponds to an intact gland in the glandular tissue of the patient;
(h) defining an intact region mask from the second objects;
(i) applying the non-intact region mask and the intact region mask to the second digital image and thereby defining non-intact regions of the second digital image and intact regions of the second digital image;
(j) defining stroma regions of the second digital image;
(k) defining influence zones in the second digital image, wherein each influence zone is disposed on a border of a non-intact region;
(l) generating third objects from the second digital image, wherein each third object corresponds to an M1 macrophage, wherein some of the third objects correspond to M1 macrophages disposed in influence zones;
(m) generating fourth objects from the second digital image, wherein each fourth object corresponds to an M2 macrophage, wherein some of the fourth objects correspond to M2 macrophages disposed in influence zones;
(n) counting the third objects that correspond to M1 macrophages disposed in influence zones and thereby obtaining a first number;
(o) counting the fourth objects that correspond to M2 macrophages disposed in influence zones and thereby obtaining a second number;
(p) determining a score based at least in part on the first number and the second number, wherein the score is indicative of whether the cancer patient will likely have a recurrence of cancer;
and
(q) displaying the score on a display device of the PSA recurrence prediction system, wherein (e) through (q) are performed by the PSA recurrence prediction system.

15 more elements

A method involving a Prostate-Specific Antigen (PSA) recurrence prediction system, the method comprising:

(a) generating first objects from a first digital image of a first tissue slice of a sample from a cancer patient, wherein the first tissue slice has been stained with a first stain that stains basal epithelial cells and a second stain that stains luminal epithelial cells, and wherein each first object corresponds to a part of the tissue that has been stained with the first stain;
(b) generating second objects from the first digital image of the first tissue slice, wherein each second object corresponds to a part of the tissue that has been stained with the second stain;
(c) based at least in part on the first and second objects defining a non-intact region mask, wherein the non-intact region mask defines a plurality of non-intact regions, wherein each non-intact region corresponds to an amount of the tissue of the first tissue slice that includes non-intact prostate glands;
(d) using the non-intact region mask to define an influence zone mask, wherein the influence zone mask defines a plurality of influence zones, wherein each influence zone corresponds to an amount of tissue of the first tissue slice that is disposed along an outer boundary of the tissue corresponding to a non-intact region;
(e) applying the influence zone mask to a second digital image and thereby defining an influence zone in the second digital image, wherein the second digital image is of a second tissue slice of the sample from the cancer patient, wherein the second tissue slice has been stained with a third stain that stains macrophages;
(f) generating third objects from the second digital image, wherein each third object corresponds to a macrophage in the second tissue slice that has been stained with the third stain;
(g) counting third objects that correspond to macrophages disposed in the influence zone defined in (e) thereby obtaining a number;
(h) determining a score based at least in part on the number, wherein the score is indicative of whether the cancer patient will likely have a recurrence of cancer;
and
(i) displaying the score on a display device of the PSA recurrence prediction system, wherein (a) through (i) are performed by the PSA recurrence prediction system.

7 more elements

A method comprising:

(a) generating first objects from a first digital image of a first tissue slice from a prostate cancer patient, wherein the first tissue slice has been stained with a first stain that stains basal epithelial cells and a second stain that stains luminal epithelial cells, and wherein each first object corresponds to a part of the tissue that has been stained with the first stain;
(b) generating second objects from the first digital image of the first tissue slice, wherein each second object corresponds to a part of the tissue that has been stained with the second stain;
(c) defining a non-intact region mask surrounding a plurality of non-intact regions based on the first and second objects, wherein each non-intact region corresponds to an amount of the first tissue slice that includes non-intact prostate glands;
(d) using the non-intact region mask to define an influence zone mask surrounding a plurality of influence zones, wherein each influence zone corresponds to an amount of the first tissue slice that is disposed along an outer boundary of one of the non-intact regions;
(e) applying the influence zone mask to a second digital image and thereby defining an influence zone in the second digital image, wherein the second digital image is of a second tissue slice from the prostate cancer patient, wherein the second tissue slice has been stained with a third stain that stains macrophages;
(f) generating third objects from the second digital image, wherein each third object corresponds to a macrophage in the second tissue slice that has been stained with the third stain;
(g) counting a number of third objects that correspond to macrophages disposed in the influence zone defined in (e);
and
(h) determining a score based on the number, wherein the score is indicative of whether the prostate cancer patient will likely have a recurrence of cancer.

6 more elements

Dependencies are inferred from the claim text.

Facts

Application: 15/075,180Filed 20 Mar 2016
Published: US 2017/0270420 A121 Sep 2017
Granted: 13 Aug 2019
Term ends: Not estimatedThe source records don’t support an estimate.

Classification10 codes

G06T 7/0012
Inspection of images, e.g. flaw detection: biomedical image inspection
G06F 18/24
Analysing: classification techniques
G06N 7/01
Computing arrangements based on specific mathematical models: probabilistic graphical models, e.g. probabilistic networks
G06T 2207/30004
Subject of image; Context of image processing: biomedical image processing