Massive training artificial neural network (mtann) for detecting abnormalities in medical images

Abstract

A method of training an artificial neural network (ANN) involves receiving a likelihood distribution map as a teacher image, receiving a training image, moving a local window across sub-regions of the training image to obtain respective sub-region pixel sets, inputting the sub-region pixel sets to the ANN so that it provides output pixel values that are compared to output pixel values of corresponding teacher image pixel values to determine an error, and training the ANN to reduce the error. A method of detecting a target structure in an image involves scanning a local window across sub-regions of the image by moving the local window for each sub-region so as to obtain respective sub-region pixel sets, inputting the sub-region pixel sets to an ANN so that it provides respective output pixel values that represent likelihoods that respective image pixels are part of a target structure, the output pixel values collectively constituting a likelihood distribution map. Another method for detecting a target structure involves training N parallel ANNs on either (A) a same target structure and N mutually different non-target structures, or (B) a same non-target structure and N mutually different target structures, the ANNs outputting N respective indications of whether the image includes a target structure or a non-target structure, and combining the N indications to form a combined indication of whether the image includes a target structure or a non-target structure. The invention provides related apparatus and computer program products storing executable instructions to perform the methods.