eduzhai > Applied Sciences > Engineering >

Smile-GANs Semi-supervised clustering via GANs for dissecting brain disease heterogeneity from medical images

  • Save

... pages left unread,continue reading

Document pages: 12 pages

Abstract: Machine learning methods applied to complex biomedical data has enabled theconstruction of disease signatures of diagnostic prognostic value. However,less attention has been given to understanding disease heterogeneity.Semi-supervised clustering methods can address this problem by estimatingmultiple transformations from a (e.g. healthy) control (CN) group to a patient(PT) group, seeking to capture the heterogeneity of underlying pathlogicprocesses. Herein, we propose a novel method, Smile-GANs (SeMi-supervIsedcLustEring via GANs), for semi-supervised clustering, and apply it to brain MRIscans. Smile-GANs first learns multiple distinct mappings by generating PT fromCN, with each mapping characterizing one relatively distinct pathologicalpattern. Moreover, a clustering model is trained interactively with mappingfunctions to assign PT into corresponding subtype memberships. Using relaxedassumptions on PT CN data distribution and imposing mapping non-linearity,Smile-GANs captures heterogeneous differences in distribution between the CNand PT domains. We first validate Smile-GANs using simulated data, subsequentlyon real data, by demonstrating its potential in characterizing heterogeneity inAlzheimer s Disease (AD) and its prodromal phases. The model was first trainedusing baseline MRIs from the ADNI2 database and then applied to longitudinaldata from ADNI1 and BLSA. Four robust subtypes with distinct neuroanatomicalpatterns were discovered: 1) normal brain, 2) diffuse atrophy atypical of AD,3) focal medial temporal lobe atrophy, 4) typical-AD. Further longitudinalanalyses discover two distinct progressive pathways from prodromal to full AD:i) subtypes 1 - 2 - 4, and ii) subtypes 1 - 3 - 4. Although demonstrated on animportant biomedical problem, Smile-GANs is general and can find application inmany biomedical and other domains.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...