Author

Xinliang Zhu

Graduation Semester and Year

2020

Language

English

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Computer Science

Department

Computer Science and Engineering

First Advisor

Junzhou Huang

Abstract

I present my work towards solving the fundamental, challenging and valuable problem for automatically processing the giga-pixel level whole slide pathology images (WSIs): the representation of them. Specifically, I target on solving the combinations of three critical aspects of the problem: (1) it's not engineering feasible to directly fit them into existing convolutional neural networks because they are too large; (2) pre-trained parameters from other domains may not be effectively transferred to pathology images, and (3) both the image samples and annotations for those images are rarely available. To evaluate the effectiveness of the developed methods, I mainly focus on the primary and important applications in medicine: the survival prediction and clinical outcome prediction. I approach to the solution step by step. Firstly, I solve the major problem of effectively making survival prediction from images by proposing a DeepConvSurv network, which integrates both the strengths of convolutional neural network and Cox proportional hazard model. To better utilize the clinical information from the patients and partially solve the problem of effectively training models on the small size medical datasets, I propose a DeepMutliSurv based on DeepConvSurv, which trains multiple tasks on multi-modality data. Due to the scarce of available annotations on the WSIs, I further create an innovative framework named WSISA to do survival prediction based on all the WSIs provided by the patients without any annotations from the pathologists. Last but not least, a powerful end-to-end WSI representation learning method WSINet is developed, which solves the three major challenges efficiently and effectively. WSINet can be adopted in various WSI based applications like survival prediction, tumor subtype classification, and biochemical indicator prediction, etc. due to its compelling end-to-end learning and representation nature.

Keywords

Deep Learning, Computational pathology, CNN, Whole slide image, Representation learning, Survival analysis

Disciplines

Computer Sciences | Physical Sciences and Mathematics

Comments

Degree granted by The University of Texas at Arlington

30698-2.zip (7712 kB)

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.