Skip to Main Content
 

Global Search Box

 
 
 
 

ETD Abstract Container

Abstract Header

Annotation, Enrichment and Fusion of Multiscale Data: Identifying High Risk Prostate Cancer

Singanamalli, Asha

Abstract Details

, Master of Sciences (Engineering), Case Western Reserve University, Biomedical Engineering.
Increasingly, advanced diagnostic techniques have begun to generate large volumes of multiscale, multimodal data spanning radiologic, histologic and molecular length scales. Although these data streams have been studied independently, their associations and collective potential largely remain unexplored. In this work, we introduce two strategies called radiohistomorphometry and supervised multiview canonical correlation analysis (sMVCCA) for data enrichment and fusion, respectively, both of which we demonstrate in context of predicting prostate cancer risk. While radiohistomorphometry probes cross-modality (ex vivo pathologic and in vivo radiologic) correlations to learn in vivo imaging markers of high-risk disease, sMVCCA seeks to fuse all available data streams into a unified canonical representation. On a data cohort comprising 16 prostate cancer patients with in vivo multiparametric (MP) magnetic resonance imaging (MRI) and ex vivo vascular (CD31) stained histology specimen, radiohistomorphometry identified a set of 14 dynamic contrast enhanced (DCE) MRI markers that were highly correlated with quantitative descriptors of microvessel architecture. These in vivo imaging markers showed moderate separability between intermediate Gleason grades, which served as surrogate markers of outcome. In a second data cohort comprising 40 prostate cancer patients who underwent surgery, mass-spectrometry derived proteomic features were fused with quantitative features of glandular distribution and morphology on H\&E stained histology to predict 5-year biochemical recurrence. Preliminary results from Kaplan-Meier survival curves indicate that the fused representation is better able to predict which patients undergo biochemical recurrence in comparison with the performances of histology and proteomics features alone. With these preliminary results, we demonstrate the clinical promise of quantitative data convergence.
Anant Madabhushi (Advisor)

Recommended Citations

Citations

  • Singanamalli, A. (n.d.). Annotation, Enrichment and Fusion of Multiscale Data: Identifying High Risk Prostate Cancer [Master's thesis, Case Western Reserve University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=case1386084613

    APA Style (7th edition)

  • Singanamalli, Asha. Annotation, Enrichment and Fusion of Multiscale Data: Identifying High Risk Prostate Cancer. Case Western Reserve University, Master's thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=case1386084613.

    MLA Style (8th edition)

  • Singanamalli, Asha. "Annotation, Enrichment and Fusion of Multiscale Data: Identifying High Risk Prostate Cancer." Master's thesis, Case Western Reserve University. Accessed MARCH 29, 2024. http://rave.ohiolink.edu/etdc/view?acc_num=case1386084613

    Chicago Manual of Style (17th edition)