Towards 3D Atlas of Human Body via the Foundation Model: A Survey

PhD Qualifying Examination


Title: "Towards 3D Atlas of Human Body via the Foundation Model: A Survey"

by

Mr. Jiaxin ZHUANG


Abstract:

3D medical images are pivotal for understanding the human body's atlas, 
allowing clinicians to visualize and analyze complex anatomical and 
pathological conditions with high precision. Deep learning has shown superior 
performance in 3D medical imaging tasks, leading to more accurate diagnostics, 
personalized treatment plans, and a more efficient healthcare system. However, 
annotating 3D medical images is time-consuming and challenging, with 
radiologists spending around 30-60 minutes per organ in a three-dimensional CT 
volume. The demand for accurate diagnostics outpaces the growth of 
radiologists, highlighting the need for improved AI-assisted systems. The 
complexity of the anatomy and subtle variations in lesion areas further 
complicates the analysis for radiologists. Therefore, foundation models for 3D 
medical images have gained increasing attention for their potential to 
alleviate the burden on radiologists. In this survey, we present a 
comprehensive review of recent research on foundation models for the 3D atlas 
of the human body. We categorize existing methods using a developed taxonomy 
and summarize strategies for improving model performance into five categories. 
We also discuss the importance of pretraining datasets and network architecture 
in current works. Finally, we pointed out the primary challenges in this field 
and highlighted promising future directions.


Date:                   Tuesday, 30 July 2024

Time:                   10:00am - 12:00noon

Zoom Meeting ID:        333 202 7090

Committee Members:      Dr. Hao Chen (Supervisor)
                        Dr. Dan Xu (Chairperson)
                        Dr. Qifeng Chen
                        Dr. Junxian He