Multi-Organ Lesion Analysis using Deep Learning in Chest CT
Principal Investigator
Name
Le Lu
Degrees
Ph.D.
Institution
Alibaba DAMO Academy
Position Title
Director
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-860
Initial CDAS Request Approval
Nov 30, 2021
Title
Multi-Organ Lesion Analysis using Deep Learning in Chest CT
Summary
Lesion analysis is a routine task for radiologists. However, finding all lesions in a CT is time-consuming and prone to missing findings. We hope to develop deep learning algorithms to help radiologists find lesions in multiple organs, measure them, and analyze their characteristics. For example, we hope to analyze lung nodules, coronary artery calcification, enlarged lymph nodes, and all types of lesions in the pancreas, esophagus, bone, liver, etc. We may also track the lesions in longitudinal follow-up studies and summarize the findings of a patient to provide a holistic diagnosis. NLST is a great source of data for these studies.
Aims
1. Developing lesion detection models that focus on lung, heart, lymph nodes, spine and rib, liver, pancreas, etc.
2. Tracking the lesions’ changes in multiple CT scans of the same patient for longitudinal analysis.
3. Training lesion measurement and segmetation models for all types of lesions.
4. Learning lesion classification algorithms to analyze the lesions' characteristics.
5. Summarizing the findings of a patient to provide a holistic diagnosis.
Collaborators
Ke Yan, Alibaba DAMO Academy
Minfeng Xu, Alibaba DAMO Academy
Ling Zhang, Alibaba DAMO Academy
Dakai Jin, Alibaba DAMO Academy