CPC G16H 40/67 (2018.01) [G06F 18/2148 (2023.01); G06N 3/08 (2013.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G16H 30/20 (2018.01)] | 19 Claims |
1. A method for training a neural network for medical image analysis using mammalian transfer learning, the method comprising:
(a) receiving, by a processor, one or more comparative datasets, wherein each of the one or more comparative datasets comprises labeled image data associated with a species;
(b) creating, by the processor, a mixed domain dataset based on the one or more comparative datasets;
(c) for each image of a plurality of images of the mixed domain dataset:
(i) defining, by the processor, a plurality of chunks within that image, wherein the size of each of the plurality of chunks is selected to obfuscate a species of the source of the image;
(ii) adding, by the processor, the plurality of chunks of that image and any associated labels to a mixed domain training dataset;
(d) by the processor, and with the neural network:
(i) training the neural network to identify a medical characteristic of a case study from a target species based on the mixed domain training set and the associated labels, wherein the one or more species of the one or more comparative datasets includes at least one species other than the target species; and
(ii) validating the trained neural network based upon a validation dataset selected from the mixed domain dataset; and
(e) creating, by the processor, a checkpoint based on the trained and validated neural network.
|