Discriminative-Region-Aware Residual Network for Adolescent Brain Structure and Cognitive Development Analysis

Yongsheng Pan, Mingxia Liu, Li Wang, Yong Xia, Dinggang Shen

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The brains of adolescents undergo profound cognitive development, especially the development of fluid intelligence (FI) that is the ability to reason and think logically (independent of acquired knowledge). Such development may be influenced by many factors, such as changes in the brain structure caused by neurodevelopment. Unfortunately, the association between brain structure and fluid intelligence is not well understood. Cross-sectional structural MRI data released by the Adolescent Brain Cognitive Development (ABCD) study pave a way to investigate adolescents’ brain structure via MRIs, but each 3D volume may contain irrelevant or even noisy information, thus degrading the learning performance of computer-aided analysis systems. To this end, we propose a discriminative-region-aware residual network (DRNet) to jointly predict FI scores and identify discriminative regions in brain MRIs. Specifically, we first develop a feature extraction module (containing several convolutional layers and ResNet blocks) to learn MRI features in a data-driven manner. Based on the learned feature maps, we then propose a discriminative region identification module to explicitly determine the weights of different regions in the brain, followed by a regression module to predict FI scores. Experimental results on 4, 154 subjects with T1-weighted MRIs from ABCD suggest that our method can not only predict fluid intelligence scores based on structural MRIs but also explicitly specify those discriminative regions in the brain.

Original languageEnglish
Title of host publicationGraph Learning in Medical Imaging - 1st International Workshop, GLMI 2019, held in Conjunction with MICCAI 2019, Proceedings
EditorsDaoqiang Zhang, Luping Zhou, Biao Jie, Mingxia Liu
PublisherSpringer
Pages138-146
Number of pages9
ISBN (Print)9783030358167
DOIs
Publication statusPublished - 2019 Jan 1
Externally publishedYes
Event1st International Workshop on Graph Learning in Medical Imaging, GLMI 2019 held in conjunction with the 22nd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2019 - Shenzhen, China
Duration: 2019 Oct 172019 Oct 17

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11849 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference1st International Workshop on Graph Learning in Medical Imaging, GLMI 2019 held in conjunction with the 22nd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2019
CountryChina
CityShenzhen
Period19/10/1719/10/17

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'Discriminative-Region-Aware Residual Network for Adolescent Brain Structure and Cognitive Development Analysis'. Together they form a unique fingerprint.

  • Cite this

    Pan, Y., Liu, M., Wang, L., Xia, Y., & Shen, D. (2019). Discriminative-Region-Aware Residual Network for Adolescent Brain Structure and Cognitive Development Analysis. In D. Zhang, L. Zhou, B. Jie, & M. Liu (Eds.), Graph Learning in Medical Imaging - 1st International Workshop, GLMI 2019, held in Conjunction with MICCAI 2019, Proceedings (pp. 138-146). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11849 LNCS). Springer. https://doi.org/10.1007/978-3-030-35817-4_17