Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


PhD_thesis_YChen.pdf (2.51 MB)

ETD Abstract Container

Abstract Header

DEVELOPMENT OF COMPUTATIONAL APPROACHES FOR MEDICAL IMAGE RETRIEVAL, DISEASE GENE PREDICTION, AND DRUG DISCOVERY

Chen, Yang
http://rave.ohiolink.edu/etdc/view?acc_num=case1435601642

Abstract Details

, Doctor of Philosophy, Case Western Reserve University, EECS - Computer Engineering.
With the deluge of biomedical data, developing computational approaches for data analysis and interrogation has become a key step in translational biomedical research. It is critical to leverage existing data to ask the right question and design algorithms for specific biomedical applications. This dissertation proposes a domain knowledge guided strategy for data gathering, data fusion and algorithm design in solving specific biomedical problems. This strategy is demonstrated in three distinct application contexts. The first application is retrieving disease manifestation images from the web for supporting patients' self-education and decision making. The challenge is three-fold: heterogeneous irrelevant web images need to be filtered; the positive examples of disease images contain diverse objects and complex backgrounds; and large amounts of manual efforts in generating training data are unaffordable. Based on a key observation that detecting disease-affected abnormal organs may greatly reduce the manual labeling efforts, our approach extracts the disease-organ semantic relationships from ontologies to guide the organ detection with pre-trained detectors. Comparing with a standard supervised method, our approach improved the average precision by 5\% while reduced the manual efforts by 85\%. The second application is developing disease-specific models to detect genetic basis for human diseases. We first develop a cross-species genetic network analysis approach to study the host-pathogen interactions in parasitic infectious diseases and predict disease associated genes. This approach was applied on malaria and demonstrated useful in guiding anti-malaria drug discovery. In the second phenome-driven approach, we explore a new disease phenotype data source in medical ontologies to construct the Disease Manifestation Network (DMN), and integrate multiple phenotype networks with genetic networks to predict genes. An application of this approach on Crohn's disease demonstrated the translational potential of the predicted genes in drug discovery. The third approach is identifying the mutual comorbidity for colorectal cancer and obesity in the comorbidity network to detect genetic basis for the link between the two diseases. The last application is drug repositioning based on combining disease genetics and mouse phenotype data. Disease associated genes have the potential to guide drug discovery. On the other hand, the mouse phenotypes provide knowledge on gene functions, which is impossible to be obtained in human. Our approach first identifies disease-specific mouse phenotypes using well-studied disease genes, and then search all FDA-approved drugs for the candidates that share similar mouse phenotype profiles with the disease.The approach was applied to predict drugs for Parkinson's disease, and achieved significant improvements comparing with a state-of-art approach based on mouse phenotype data. In summary, this dissertation demonstrates the effectiveness of computational algorithms in translational biomedical research. It also demonstrates that the computation-based work have great potential in elucidating disease genetic basis, finding innovative drugs, and improving patient health education.
Zhang Guo-qiang (Committee Chair)
Xu Rong (Advisor)
Li Jing (Committee Member)
Cavusoglu Cenk (Committee Member)
Zhang Xiang (Committee Member)
148 p.
Biomedical Research; Computer Engineering; Computer Science
computational approach; translational biomedical research; image retrieval; disease gene prediction; drug repositioning

Recommended Citations

Citations

  • Chen, Y. (2015). DEVELOPMENT OF COMPUTATIONAL APPROACHES FOR MEDICAL IMAGE RETRIEVAL, DISEASE GENE PREDICTION, AND DRUG DISCOVERY [Doctoral dissertation, Case Western Reserve University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=case1435601642

    APA Style (7th edition)

  • Chen, Yang. DEVELOPMENT OF COMPUTATIONAL APPROACHES FOR MEDICAL IMAGE RETRIEVAL, DISEASE GENE PREDICTION, AND DRUG DISCOVERY. 2015. Case Western Reserve University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=case1435601642.

    MLA Style (8th edition)

  • Chen, Yang. "DEVELOPMENT OF COMPUTATIONAL APPROACHES FOR MEDICAL IMAGE RETRIEVAL, DISEASE GENE PREDICTION, AND DRUG DISCOVERY." Doctoral dissertation, Case Western Reserve University, 2015. http://rave.ohiolink.edu/etdc/view?acc_num=case1435601642

    Chicago Manual of Style (17th edition)

case1435601642
403
© 2015, all rights reserved.
This open access ETD is published by Case Western Reserve University School of Graduate Studies and OhioLINK.