Methods, kits, and systems for scoring the immune response to cancer

A computer-implemented method for counting a number of types of cells in a single tissue specimen that has been stained with a multiplex assay, comprising:

• imaging the single tissue specimen that has been stained with the multiplex assay that includes lymphocyte markers CD3, CD8, CD20, FoxP3, and tumor detection markers;
• unmixing the image of single tissue specimen that has been stained with a multiplex assay into separate image channels for each marker of the multiplex assay;
• identifying regions of interest in each image channel based on intensity information in each channel, wherein regions of low intensity in each channel are removed, and regions of high intensity represent cell signals;
• generating a single surrogated image, wherein the surrogated image is a combination of the image channel information of all the lymphocyte markers;
• applying a cell detection algorithm, wherein the cell detection algorithm is a membrane finding algorithm or a nucleus finding algorithm;
• identifying features of the lymphocytes and combinations of lymphocytes in each image channel or image of combined channels, or a transformed image such as grayscale or absorbance image, or a surrogated image;
• training a classification algorithm based on features of known lymphocytes and lymphocyte combinations;
• applying the trained algorithm to features of the lymphocytes and combinations of lymphocytes in each image channel or image of combined channels, or a transformed image such as grayscale or absorbance image, or a surrogated image, that were identified to classify the detected cells as at least one of false positive cells, CD3 only T-cells, CD3 and CD8 T-cells, FP3 T-cells;
• and CD20 B-cells;
• counting a number of each different type of cell classified;
• generating a score the tissue specimen, wherein the score is based on the number each type of cell counted.