Computer-aided pathological diagnosis system
AssigneeUniversity of South Florida
InventorsWei Qian, Dansheng Song, Tatyana A. Zhukov
Life of this patent
Estimating the term…
Abstract
The present invention is a computer-aided pathological diagnosis method for the classification of cancer cells in a tissue specimen based on a digital cellular image of the tissue specimen. The method of the present invention includes the steps of, extracting the histological characteristic features of the cellular image using preprocessing algorithms having adaptive strategies to enhance the cellular image, declustering the extracted histological characteristic features of the cellular image to isolate the individual cells and the nuclei inside the cells, segmenting the declustered cellular image, labeling the segmented cellular image and classifying the cells in the labeled cellular image as cancer cells or non-cancer cells.
Claims28
A computer-aided pathological diagnosis method for the classification of cancer cells in a tissue specimen based on a digital cellular image of the tissue specimen, the method comprising the steps of:
- extracting the histological characteristic features of the cellular image using preprocessing algorithms having adaptive strategies to enhance the cellular image;
- declustering the cellular image to isolate the individual cells and the nuclei inside the cells using a distance-transform that yields the minimal distance to the boundary of an object for every pixel in the cellular image, wherein the distance-transform is given by
- Y
-
- (
- x
- ,
- y
- )
- =
- {
- min
-
-
-
- (
- x
- ,
- y
- )
- -
- η
-
- (
- x
- k
- ,
- y
- k
- )
-
-
- Vk
- }
- ,
- where the coordinate of every pixel x=(x;y) in a uniform area (x;y)ε A and η(xk;yk) is the surrounding curve of the area;
- segmenting the declustered cellular image;
- labeling the segmented cellular image;
- and
- classifying the cells in the labeled cellular image as cancer cells or non-cancer cells.
A computer-aided pathological diagnosis system for the classification of cancer cells in a tissue specimen based on a digital cellular image of the tissue specimen, the system comprising:
- a computer-readable storage device containing a set of instructions that causes a computer to perform the steps of:
- extracting the histological characteristic features of the cellular image using preprocessing algorithms having adaptive strategies to enhance the cellular image;
- declustering the cellular image to isolate the individual cells and the nuclei inside the cells using a distance-transform that yields the minimal distance to the boundary of an object for every pixel in the cellular image, wherein the distance-transform is given by
- Y
-
- (
- x
- ,
- y
- )
- =
- {
- min
-
-
-
- (
- x
- ,
- y
- )
- -
- η
-
- (
- x
- k
- ,
- y
- k
- )
-
-
- Vk
- }
- ,
- where the coordinate of every pixel x=(x;y) in a uniform area (x;y)ε A and η(xk;yk) is the surrounding curve of the area;
- segmenting the declustered cellular image;
- labeling the segmented cellular image;
- and
- classifying the cells in the labeled cellular image as cancer cells or non-cancer cells.
Dependencies are inferred from the claim text.