Chest Radiograph And Pathologic Phenotype For Biological Age Prediction Using Deep Learning – Training and Validation in Chinese Patients
Principal Investigator
Name
Varut Vardhanabhuti
Degrees
M.D., Ph.D
Institution
The University of Hong Kong
Position Title
Clinical Assistant Professor
Email
About this CDAS Project
Study
PLCO
(Learn more about this study)
Project ID
PLCOI-820
Initial CDAS Request Approval
Jan 5, 2022
Title
Chest Radiograph And Pathologic Phenotype For Biological Age Prediction Using Deep Learning – Training and Validation in Chinese Patients
Summary
A deep learning approach as applied to medical imaging (e.g. radiological images and histopathology slides) has huge potential for diagnostic and prognostic prediction of diseases. In this study, we will use the PLCO datasets for training a deep learning model using chest radiograph to predict biological age, and validate on a local dataset of Chinese patients in Hong Kong. In patients with histopathology slides, we will train a new ‘CancerAge’ model which is a new concept for predicting cancer severity based on histopathology slides. This proof of concept will be applied to 2 cancers, namely lung and breast cancer patients.
Aims
1. To develop a deep learning model using chest radiograph for the prediction of biological age.
2. To develop a deep learning model using pathology slides for the prediction of cancer age.
3. To combine and chest radiograph and pathology information i.e. “rad-path Age model’ to create a combined model for prognostication.
4. To explore generalizability in Chinese patients using local validation datasets in Hong Kong.
Collaborators
N/A