US 11,748,877 B2
System and method associated with predicting segmentation quality of objects in analysis of copious image data
Joel Haskin Saltz, Manhasset, NY (US); Tahsin M. Kurc, Coram, NY (US); Yi Gao, Stony Brook, NY (US); Wei Zhu, Setauket, NY (US); Si Wen, East Setauket, NY (US); Tianhao Zhao, Coram, NY (US); and Sampurna Shrestha, Stony Brook, NY (US)
Assigned to The Research Foundation for The State University of New York, Albany, NY (US)
Appl. No. 16/605,404
Filed by The Research Foundation for The State University of New York, Albany, NY (US)
PCT Filed May 10, 2018, PCT No. PCT/US2018/032026
§ 371(c)(1), (2) Date Oct. 15, 2019,
PCT Pub. No. WO2018/209057, PCT Pub. Date Nov. 15, 2018.
Claims priority of provisional application 62/504,819, filed on May 11, 2017.
Prior Publication US 2020/0126207 A1, Apr. 23, 2020
Int. Cl. G06T 7/11 (2017.01); G06T 7/00 (2017.01); G06T 7/136 (2017.01); G06T 7/41 (2017.01); G06F 18/214 (2023.01); G06F 18/21 (2023.01); G06F 18/2431 (2023.01); G06V 10/774 (2022.01); G06V 20/69 (2022.01)
CPC G06T 7/0012 (2013.01) [G06F 18/214 (2023.01); G06F 18/217 (2023.01); G06F 18/2431 (2023.01); G06T 7/0002 (2013.01); G06T 7/11 (2017.01); G06T 7/136 (2017.01); G06T 7/41 (2017.01); G06V 10/774 (2022.01); G06V 20/695 (2022.01); G06T 2207/20021 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/30024 (2013.01); G06T 2207/30168 (2013.01); G06T 2207/30181 (2013.01)] 14 Claims
OG exemplary drawing
 
1. A system for predicting segmentation quality of segmented objects implemented in the analysis of copious image data, the system comprising:
a quality assessment engine including a computing device configured to train the system during a training phase by performing operations comprising:
receiving a collection of image data related to a particular type of data;
partitioning the image data into segmented data portions based on an object associated with the collection of image data;
determining regions of interest associated with the segmented data portions;
determining quality of segmentation of the segmented data portions for respective classification of the regions of interest, the quality of segmentation being determined based on patch-level intensity and texture features associated with the image data;
assigning a classification label to the regions of interest, the classification label describing a quality of segmentation of the objects associated with the segmented data portions;
partitioning regions of interest into sub-regions;
computing a set of intensity and texture features for each of the sub-regions of the segmented data portions; and
generating a training dataset based on the computed intensity and texture features for the sub-regions in order to train a classification model based on a predetermined threshold value; and
the computing device in the quality assessment engine being configured to predict segmentation quality during a prediction phase by performing operations comprising:
receiving test images to iteratively classify segmented data portions based on an object associated with the test images, using the trained classification model; and
predicting the segmentation quality of segmented objects in the test images based on the trained classification model.