Skip to Main Content

An official website of the United States government

Government Funding Lapse

Because of a lapse in government funding, the information on this website may not be up to date, transactions submitted via the website may not be processed, and the agency may not be able to respond to inquiries until appropriations are enacted. The NIH Clinical Center (the research hospital of NIH) is open. For more details about its operating status, please visit  cc.nih.gov. Updates regarding government operating status and resumption of normal operations can be found at OPM.gov.

Principal Investigator
Name
Ahmet Coskun
Degrees
Ph.D.
Institution
Georgia Institute of Technology
Position Title
Bernie-Marcus Early-Career Professor/Assistant Professor
Email
About this CDAS Project
Study
PLCO (Learn more about this study)
Project ID
PLCOI-695
Initial CDAS Request Approval
Nov 17, 2020
Title
Deep learning based cancer grading for early detection from histology image biostatistics
Summary
Conventional cancer diagnostics focus on histopathology tissue image using immunohistochemistry. However, this approach is limited to only labeling of a single marker per tissue section, making it challenging to identify multiple markers in the same histology data. This project will use this PCOI data to train cancer staging and survival analyses to distinct patients. The deep learning model will then be implemented in highly multiplex tissue imaging data that we will obtain in our lab. The multiplexed imaging allows better analysis of cell composition, functional state, and cell microenvironment, providing precise molecular portrait cancer progression. Therefore, the presented study will bring the gap between histopathology imaging and multiplex imaging by studying the correlation from grade assessment and marker molecules location using the PLCO database and complementary high-dimensional data from biospecimens.
Aims

Aim 1: To accurately predict cancer grade assessment from pathology images using a deep convolutional neural network.
Aim 2: To robustly classify prostate and lung cancer pathology image regions into cancerous and healthy cells.
Aim 3: To perform efficient and automated feature extraction from pathology image regions using variational autoencoders, yielding the features for cancer grade and clinical phenotypes.
Aim 4: To integrate deep learning models of pathology images with multiplex immunofluorescence images of specific cancer biomarkers for grade assessment, feature extraction, and clinical survival analysis.

Collaborators

Thomas Hu (Georgia Institute of Technology)