Optimal, user-friendly, object backround separation

A method comprising:

• receiving a plurality of digital images, each indicative of one or more objects of interest;
• determining, for each of the plurality of digital images, an intensity distribution for the one or more objects of interest;
• based on the one or more objects of interest in each of the plurality of digital images, generating an object scale indicative of one or more of a size or a shape of a representative object of interest;
• sub-dividing the plurality of digital images into one or more subsets;
• based on the object scale and the intensity distribution of the one or more objects of interest in each of the plurality of digital images, generating a set of object filters configured to differentiate a first type of object of interest from a second type of object of interest;
• based on one or more of the first type of object of interest or the second type of object of interest, selecting one or more object filters from the set of object filters, wherein each of the selected one or more object filters are assigned a weight;
• generating, with the selected one or more object filters, a feature space comprising one or more reference points associated with at least one of the first type of object of interest or the second type of object of interest;
• and
• based on the feature space and the selected one or more object filters, configuring an object detector to classify an object of interest in a sample image.

A method of identifying an object of interest in a plurality of digital images configured as an image group, the method comprising:

• a) obtaining first samples of an intensity distribution of one or more objects of interest in the image group;
• b) obtaining second samples of the intensity distribution of confounder objects in the image group, at a frequency high enough to affect a pre-defined performance metric, wherein the frequency is set based upon an area under a ROC curve;
• c) transforming the first and second samples into an appropriate first space;
• d) performing dimension reduction on the transformed first and second samples, whereby the dimension reduction of the transformed first and second samples generates an object detector;
• e) transforming the image group into the first space;
• f) projecting the transformed image group into the reduced dimension space obtained in step d) to generate a reduced image group;
• g) classifying each pixel in the reduced image group with an appropriate neighborhood based on a comparison with the object detector;
• and
• h) automatically identifying one or more objects of interest in the reduced image group from abnormal pixels based upon the comparison with the object detector;
• i) generating, for an image in the image group, a spatial distribution of features and one or more scalar features, wherein the spatial distribution of features comprises a point-wise output from feature detection or a heat map, and wherein the one or more scalar features comprise one or more histograms;
• and
• j) based on the spatial distribution of features, determining, using a neural network trained according to steps (a)-(f), a disease or a condition.

A system comprising:

• a memory for storing digital images;
• and
• a processor in communication with the memory, the processor configured to:
• receive a plurality of digital images, each indicative of one or more objects of interest;
• determine, for each of the plurality of digital images, an intensity distribution for the one or more objects of interest;
• based on the one or more objects of interest in each of the plurality of digital images, generate an object scale indicative of one or more of a size or a shape of a representative object of interest;
• sub-divide the plurality of digital images into one or more subsets;
• based on the object scale and the intensity distribution of the one or more objects of interest in each of the plurality of digital images, generate a set of object filters configured to differentiate a first type of object of interest from a second type of object of interest;
• based on one or more of the first type of object of interest or the second type of object of interest, select one or more object filters from the set of object filters, wherein each of the selected one or more object filters are assigned a weight;
• generate, with the selected one or more object filters, a feature space comprising one or more reference points associated with at least one of the first type of object of interest or the second type of object of interest;
• and
• based on the feature space and the selected one or more object filters, configure an object detector to classify an object of interest in a sample image.