CPC C12M 23/16 (2013.01) [G01N 15/1056 (2013.01); G06F 18/2431 (2023.01); G06T 5/30 (2013.01); G06T 7/0012 (2013.01); G06T 7/11 (2017.01); G06T 7/136 (2017.01); G06T 7/74 (2017.01); G06V 10/34 (2022.01); G06V 10/809 (2022.01); G06V 20/695 (2022.01); G06V 20/698 (2022.01); H04N 23/55 (2023.01); H04N 23/56 (2023.01); G01N 2015/1006 (2013.01); G06T 2207/10056 (2013.01); G06T 2207/20076 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01)] | 15 Claims |
8. A system comprising:
a camera that produces an image of a semen sample;
an image processor that segments the image to provide a plurality of image tiles, the image processor locating individual sperm within a representation of the image generated via an template matching process in which a cross-correlation between a template representing at least a portion of a sperm and each of a plurality of locations on the image is computed and locations having a cross-correlation output above a threshold value are selected and generating a plurality of tiles from the image according to the locations of the individual sperm; and
a convolutional neural network that evaluates a subset of the plurality of image tiles and assigns to each of the image tiles to one of a plurality of output classes respective likelihoods that sperm within the image tile belongs to each of a set of at least one of a plurality of output classes representing the morphology of sperm within the image tile.
|