US 11,748,879 B2
Method and system for intracerebral hemorrhage detection and segmentation based on a multi-task fully convolutional network
Feng Gao, Seattle, WA (US); Youbing Yin, Kenmore, WA (US); Danfeng Guo, Beijing (CN); Pengfei Zhao, Shenzhen (CN); Xin Wang, Seattle, WA (US); Hao-Yu Yang, Seattle, WA (US); Yue Pan, Seattle, WA (US); Yi Lu, Seattle, WA (US); Junjie Bai, Seattle, WA (US); Kunlin Cao, Kenmore, WA (US); Qi Song, Seattle, WA (US); and Xiuwen Yu, Redmond, WA (US)
Assigned to KEYAMED NA, INC., Seattle, WA (US)
Filed by KEYAMED NA, INC., Seattle, WA (US)
Filed on Sep. 21, 2021, as Appl. No. 17/480,520.
Application 17/480,520 is a continuation of application No. 16/861,114, filed on Apr. 28, 2020, granted, now 11,170,504.
Claims priority of provisional application 62/842,482, filed on May 2, 2019.
Prior Publication US 2022/0005192 A1, Jan. 6, 2022
This patent is subject to a terminal disclaimer.
Int. Cl. G06K 9/00 (2022.01); G06T 7/00 (2017.01); G06T 1/00 (2006.01); G06T 7/11 (2017.01); G06N 3/08 (2023.01); A61B 5/02 (2006.01); A61B 5/00 (2006.01); G06T 11/00 (2006.01); G06F 18/24 (2023.01); G06N 3/044 (2023.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01)
CPC G06T 7/0012 (2013.01) [A61B 5/0042 (2013.01); A61B 5/02042 (2013.01); A61B 5/7264 (2013.01); G06F 18/24 (2023.01); G06N 3/044 (2023.01); G06N 3/08 (2013.01); G06T 1/0007 (2013.01); G06T 7/11 (2017.01); G06T 11/003 (2013.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G06T 2207/10081 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30016 (2013.01); G06T 2207/30101 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A system for detecting a medical condition of a subject, comprising:
a communication interface configured to receive a sequence of images acquired from the subject by an image acquisition device and an end-to-end multi-task learning model, the end-to-end multi-task learning model comprising an encoder, a Convolutional Recurrent Neural Network (ConvRNN) and at least one of a decoder or a classifier; and
at least one processor, configured to:
extract feature maps from the images using the encoder;
capture contextual information between adjacent images in the sequence using the ConvRNN; and
detect the medical condition of the subject using the classifier based on the extracted feature maps and the contextual information or segment at least one image in the sequence using the decoder to obtain a region of interest indicative of the medical condition based on the extracted feature maps.