Real-time whole slide pathology image cell counting

A method of calculating a whole cell count for a whole slide pathology image, comprising:

• segmenting the whole slide pathology image using a global thresholding filter, wherein the global thresholding filter determines if one or more features of the whole slide pathology image satisfy a global threshold value, and wherein the one or more features that satisfy the global threshold value define a tissue area;
• selecting a plurality of patches comprising the tissue area;
• generating a stain intensity image, wherein the generating comprises determining stain intensity vectors within the plurality of patches;
• generating a cell mask defining a plurality of mask blobs by iteratively segmenting the stain intensity image using the global thresholding filter, wherein the global thresholding filter determines if one or more features of the stain intensity image satisfy a local threshold value that is initially less than the global threshold value and gradually reduced after each iteration until a size of each mask blob within the seed mask is related to a selected cell size in a final iteration;
• applying a chamfer distance transform to the cell mask to generate a distance map;
• determining an h_maxima marker on the distance map;
• determining cell seeds in the distance map using the h_maxima marker;
• determining cell segments within the plurality of patches using a watershed transformation, wherein inputs to the watershed transformation comprise the distance map with the h_maxima marker and the cell seeds;
• calculating a whole cell count for the plurality of patches based on the cell segments;
• and
• configuring a client device for real-time cell counting based on the whole cell count.

A non-transitory computer-readable medium having computer instructions stored thereon for calculating a whole cell count for a whole slide pathology image, which, when executed by a processor, cause the processor to perform one or more steps comprising:

• segmenting the whole slide pathology image using a global thresholding filter, wherein the global thresholding filter determines if one or more features of the whole slide pathology image satisfy a global threshold value, and wherein the one or more features that satisfy the global threshold value define a tissue area;
• selecting a plurality of patches comprising the tissue area;
• generating a stain intensity image, wherein the generating comprises determining stain intensity vectors within the plurality of patches;
• generating a cell mask defining a plurality of mask blobs by iteratively segmenting the stain intensity image using the global thresholding filter, wherein the global thresholding filter determines if one or more features of the stain intensity image satisfy a local threshold value that is initially less than the global threshold value and gradually reduced after each iteration until a size of each mask blob within the seed mask is related to a selected cell size in a final iteration;
• applying a chamfer distance transform to the cell mask to generate a distance map;
• determining an h_maxima marker on the distance map;
• determining cell seeds in the distance map using the h_maxima marker;
• determining cell segments within the plurality of patches using a watershed transformation, wherein inputs to the watershed transformation comprise the distance map with the h_maxima marker and the cell seeds;
• calculating a whole cell count for the plurality of patches based on the cell segments;
• and
• configuring a client device for real-time cell counting based on the whole cell count.

An apparatus for determining a cell count for a region of interest within a whole slide pathology image, the apparatus comprising:

• a processor;
• a memory device storing software instructions for calculating a cell count;
• a client device executable on the processor according to software instructions stored in the memory device and configured to:
• receive a region of interest indication comprising a closed curve drawn within a patch grid overlay of the whole slide pathology image,
• wherein the patch grid comprises a plurality of patches;
• determine, for each patch on a boundary of the closed curve, a percentage of the patch enclosed by the closed curve;
• and
• determine a cell count within the closed curve based on a whole cell count for each patch inside of the closed curve, and the determined percentage for each patch on the boundary.