Deep Learning for Medical Image Analysis

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering

Final Year Thesis Oral Defense

Title: "Deep Learning for Medical Image Analysis"

by

HE Zhou

Abstract:

Recent advancements in medical image segmentation techniques have achieved 
compelling results. However, most of the widely used approaches do not 
take into account any prior knowledge about the shape of the biomedical 
structures being segmented. More recently, some works have presented 
approaches to incorporate shape information. However, most of them cannot 
adapt well to small structures like brain structures due to the extreme 
imbalance between the number of positive and negative voxels, while this 
scenario is very common in medical image analysis. In this paper, we 
present a novel algorithm that seamlessly integrates the shape information 
into the segmentation network, while being robust enough on very small 
biomedical structures. Experiments on human caudate nucleus demonstrate 
that our approach can achieve a lower Hausdorff distance and higher Dice 
Coefficient than the state-of-the-art approaches.


Date            : 24 April 2018 (Tuesday)

Time            : 16:10 - 16:50

Venue           : Room 1505 (near lifts 25/26), HKUST

Advisor         : Prof. CHUNG Albert Chi-Shing

2nd Reader      : Prof. TANG Chi-Keung