Boosted consensus classifier for large images using fields of view of various sizes

摘要

A system and method for predicting disease outcome by analyzing a large, heterogeneous image by a boosted, multi-field-of-view (FOV) framework, based on image-based features from multi-parametric heterogeneous images, comprises (a) inputting the heterogeneous image; (b) generating a plurality of FOVs at a plurality of fixed FOV sizes, the method for generating the plurality of FOVs at a plurality of fixed FOV sizes comprising dividing simultaneously, via the computing device, the large, heterogeneous image into (i) a plurality of FOVs at a first fixed FOV size from among the plurality of fixed FOV sizes; and (ii) a plurality of FOVs at a second fixed FOV size from among the plurality of fixed FOV sizes; (c) producing simultaneously for the heterogeneous image a combined class decision for: (i) the plurality of FOVs at the first fixed FOV size, and (ii) the plurality of FOV s at the second fixed FOV size.

主权项

1. A method for analyzing a large, heterogeneous image, using a boosted multi-field-of-view (FOV) framework, comprising:
(a) inputting, via a computing device, the large, heterogeneous image; (b) generating, via the computing device, a plurality of field-of-views (FOVs) at a plurality of fixed FOV sizes, the method for generating the plurality of FOVs at a plurality of fixed FOV sizes comprising dividing simultaneously, via the computing device, the large, heterogeneous image into (i) a plurality of field-of-views (FOVs) at a first fixed FOV size from among the plurality of fixed FOV sizes; and (ii) a plurality of field-of-views (FOVs) at a second fixed FOV size from among the plurality of fixed FOV sizes; (c) producing simultaneously for the large, heterogeneous image, via the computing device, a combined class decision for: (i) the plurality of FOVs at the first fixed FOV size, and (ii) the plurality of FOVs at the second fixed FOV size, by:
(1) detecting simultaneously, via the computing device, (i) at least one object to yield at least one detected object from each FOV at the first fixed FOV size, and (ii), at least one object to yield at least one detected object from each FOV at the second fixed FOV size;(2) extracting simultaneously, via the computing device, (i) at least one image feature of the at least one detected object from each FOV at the first fixed FOV size, and (ii) at least one image feature of the at least one detected object from each FOV at the second fixed FOV size;(3) training simultaneously, via the computing device, a classifier from: (i) the at least one image feature from each FOV of the plurality of FOVs at the first fixed FOV size, and (ii) the at least one image feature from each FOV for the plurality of FOVs at the second fixed FOV size;(4) making simultaneously, via the computing device, using the classifier, (i) a first class prediction for each FOV of the plurality of FOVs at the first fixed FOV size based on the at least one image feature from each FOV of the plurality of FOVs at the first fixed FOV size, and (ii), a second class prediction for each FOV of the plurality of FOVs at the second fixed FOV size based on the at least one image feature from each FOV for the plurality of FOVs at the second fixed FOV size; and(5) producing simultaneously, via the computing device, (i) the combined class decision for the plurality of FOVs at the first fixed FOV size, and (ii) the combined class decision for the plurality of FOVs at the second fixed FOV size; (d) repeating simultaneously, via the computing device, for each plurality of FOVs at the plurality of fixed FOV sizes to generate a plurality of combined class decisions, detecting step (1), extracting step (2), training step (3), making step (4), and producing step (5); (e) aggregating, via the computing device, using a boosted multi-FOV classifier, the plurality of combined class decisions at each fixed FOV size of the plurality of fixed FOV sizes to yield an aggregated boosted multi-FOV decision, wherein the boosted, multi-field-of-view (multi-FOV) classifier finds optimal FOV sizes for aggregating the plurality of combined class decisions that are outputs of the ensembles at each of the plurality of fixed FOV sizes; and (f) producing, via the computing device, a consensus classification for the large, heterogeneous image, based on the aggregated multi-FOV decision.