Optimizing the initialization and convergence of active contours for segmentation of cell nuclei in histological sections

A computer-implemented segmentation method for processing a gray-scale image made up of a plurality of pixels on a display, the image representing a plurality of nuclei, the method comprising:

• providing a set of instructions to a physical memory of the computer;
• configuring a processor of the computer by executing the set of instructions in the physical memory so as to automatically segment the image by:
• (a) setting at least one parameter in the segmentation algorithm,
• (b) thresholding the grey-scale image to create a black and white image, wherein the thresholding step calculates for each pixel (i,j) a threshold t based on t=m(i,j)+k.s(i,j) where m and s are the mean and standard deviation of a neighborhood of pixels around the pixel (ij) and k is a constant in the range 0.1, to 0.4;
• (c) post-processing to identify objects in the black and white image and to remove objects failing to meet predetermined criteria based on the black and white image and the gray-scale image;
• (d) segmenting to extract objects corresponding to the objects remaining after the post-processing step;
• (e) applying an edge detector on the segmented image to identify the edges of nuclei;
• (f) creating a mask for each nucleus to create a plurality of masks;
• (g) where masks overlap by a predetermined range of overlaps, splitting the overlapping masks to replace the overlapping masks in the plurality of masks;
• and
• (h) providing as an output a transformed image made up of mask regions representing the plurality of masks representing respective nuclei.

A computer-implemented segmentation method for processing a ray-scale image made of a plurality of pixels on a display, the image representing a plurality of nuclei, the method comprising:

• providing a set of instructions to a physical memory of the computer;
• configuring a processor of the computer by executing the set of instructions in the physical memory so as to automatically segment the image by:
• (a) setting at least one parameter in the segmentation algorithm,
• (b) thresholding the grey-scale image to create a black and white image;
• (c) post-processing to identify objects in the black and white image and to remove objects failing to meet predetermined criteria based on the black and white image and the gray-scale image, wherein the post-processing step includes:
• labeling all objects and holes in the black and white image;
• calculating an average gradient magnitude of the grey scale image around the boundary of each labeled object and hole;
• removing objects and holes for which the average gradient magnitude does not exceed a predetermined threshold;
• filling holes;
• and
• removing objects with an area less than a predetermined number of pixels;
• (d) segmenting to extract objects corresponding to the objects remaining after the post-processing step;
• (e) applying an edge detector on the segmented image to identify the edges of nuclei;
• (f) creating a mask for each nucleus to create a plurality of masks;
• (g) where by a predetermined range of overlaps, splitting the overlapping masks to replace the overlapping masks in the plurality of masks;
• and
• (h) providing as an output a transformed image made up of mask regions representing the plurality of masks representing respective nuclei.

A computer program product comprising a non-transitory computer-readable storage medium having code stored thereon, the code being executable on a processor of a computer to cause the processor to perform a segmentation method on a grey scale image using the steps of:

• (a) setting, in the processor, at least one parameter in the segmentation algorithm,
• (b) thresholding the grey-scale image, in the processor, to create a black and white image;
• (c) post processing, in the processor, to identify objects in the black and white image and remove objects failing to meet predetermined criteria based on the black and white image and the gray-scale image;
• (d) segmenting, in the processor, to extract objects corresponding to the objects remaining after post-processing from the gray-scale image;
• (e) applying an edge detector on the segmented image, in the processor, to identify the edges of nuclei;
• (f) creating a mask, in the processor, for each segmented nucleus;
• and
• (g) where masks overlap by a predetermined range of overlaps, splitting, in the processor, the overlapping masks, wherein the predetermined range of overlaps is from a lower bound to an upper bound of overlap area as a percentage of the combined area of the masks, the lower bound is in the range 1% to 10%, preferably 3% to 7%, and the upper bound is in the range 15% to 40%, preferably 20% to 30%.