Systems and methods for automated detection of an indication of malignancy in a mammographic image

A method of computing an indication of likelihood of malignancy in a two dimensional (2D) x-ray based single channel mammographic image by a trained statistical classifier, comprising:

• receiving a plurality of single channel 2D mammographic images of a plurality of patients;
• for each one of the plurality of single channel 2D mammographic images:
• receiving a single channel 2D mammographic image of at least a portion of a breast, wherein the single channel 2D mammographic image includes a single pixel intensity value for each pixel of a plurality of pixels thereof;
• converting the single channel 2D mammographic image into a multi channel 2D mammographic image including a plurality of pixel intensity value channels for each pixel of a plurality of pixels thereof;
• computing by a first sub-classifier of the trained statistical classifier according to the whole multi channel 2D mammographic image, a first score indicative of likelihood of malignancy within the whole multi channel 2D mammographic image;
• computing by a second sub-classifier of the trained statistical classifier according to each respective patch of a plurality of patches extracted from the multi channel 2D mammographic image, a respective second score of a plurality of second scores indicative of likelihood of malignancy within each respective patch of the plurality of patches;
• computing by a gating sub-classifier of the trained statistical classifier according to the first score and the plurality of second scores, an indication of likelihood of malignancy and a location of the malignancy;
• and
• providing the indication of likelihood of malignancy and the location of the malignancy,
• wherein the single channel 2D mammographic image comprises a black and white image, and the multi channel 2D mammographic image comprises a two, three, or four channel false color image,
• wherein single channel 2D mammographic images associated with an indication of relatively low likelihood of malignancy are further categorized as benign, indicative of breast tissue with at least one benign abnormality, or categorized as normal, indicative of normal breast tissue.

A method of computing an indication of likelihood of malignancy in a two dimensional (2D) x-ray based single channel mammographic image by a trained statistical classifier, comprising:

• receiving a single channel 2D mammographic image of at least a portion of a breast, wherein the single channel 2D mammographic image includes a single pixel intensity value for each pixel of a plurality of pixels thereof;
• converting the single channel 2D mammographic image into a multi channel 2D mammographic image including a plurality of pixel intensity value channels for each pixel of a plurality of pixels thereof;
• computing by a first sub-classifier of the trained statistical classifier according to the whole multi channel 2D mammographic image, a first score indicative of likelihood of malignancy within the whole multi channel 2D mammographic image;
• computing by a second sub-classifier of the trained statistical classifier according to each respective patch of a plurality of patches extracted from the multi channel 2D mammographic image, a respective second score of a plurality of second scores indicative of likelihood of malignancy within each respective patch of the plurality of patches;
• computing by a gating sub-classifier of the trained statistical classifier according to the first score and the plurality of second scores, an indication of likelihood of malignancy and a location of the malignancy;
• and
• providing the indication of likelihood of malignancy and the location of the malignancy,
• wherein the indication of likelihood of malignancy and the location of the malignancy is provided as a second reader to an interpreting radiologist,
• wherein the single channel 2D mammographic image comprises a black and white image, and the multi channel 2D mammographic image comprises a two, three, or four channel false color image.

A method of computing an indication of likelihood of malignancy in a two dimensional (2D) x-ray based single channel mammographic image by a trained statistical classifier, comprising:

• receiving a single channel 2D mammographic image of at least a portion of a breast, wherein the single channel 2D mammographic image includes a single pixel intensity value for each pixel of a plurality of pixels thereof;
• converting the single channel 2D mammographic image into a multi channel 2D mammographic image including a plurality of pixel intensity value channels for each pixel of a plurality of pixels thereof;
• computing by a first sub-classifier of the trained statistical classifier according to the whole multi channel 2D mammographic image, a first score indicative of likelihood of malignancy within the whole multi channel 2D mammographic image;
• computing by a second sub-classifier of the trained statistical classifier according to each respective patch of a plurality of patches extracted from the multi channel 2D mammographic image, a respective second score of a plurality of second scores indicative of likelihood of malignancy within each respective patch of the plurality of patches;
• computing by a gating sub-classifier of the trained statistical classifier according to the first score and the plurality of second scores, an indication of likelihood of malignancy and a location of the malignancy;
• providing the indication of likelihood of malignancy and the location of the malignancy;
• and
• presenting on a display, at least on of:
• the computed multi channel mammographic image, and a visual marking indicative of the location of malignancy,
• wherein the single channel 2D mammographic image comprises a black and white image, and the multi channel 2D mammographic image comprises a two, three, or four channel false color image.