Summary
The TCGA Breast Phenotype Research Group is part of the CIP TCGA Radiology Initiative focused on analyzing images from the TCGA-BRCA collection. The group began with a research project aimed at creating an ad hoc multi-institutional research team dedicated to discovering the value of applying controlled terminology to the MR imaging features of patients with breast cancer. These images correlate to the Breast Invasive Carcinoma (BRCA) data in the TCGA Data Portal. |
Starting or Joining a Research Project
We are currently hosting calls in support of TCGA-BRCA research projects on Wednesdays at 4pm Eastern. Please contact us at cancerimagingarchive@mail.nih.gov if you would like to inquire about setting up a new research project, discuss potential collaborations with existing groups or be otherwise kept in the loop as this effort moves forward.
Group Projects
This is a listing of ongoing projects. If you are working with the TCGA-BRCA data hosted on TCIA please let us know and we would be happy to add a section describing your project here.
- Breast Image Feature Scoring Project - This project is in the preliminary stages and is being collaborated on by group members from MSKCC, MDACC, Roswell Park, and UPMC. Multiple readers are evaluating each subject's imaging features using a BIRADS inspired Breast Feature Key and then investigating potential correlations with the TCGA genomic and clinical data. This project is being led by Liz Morris of MSKCC.
- Mapping of Multi-modality Breast Image-based Phenotypes to Histopathology and Genomics – This project seeks to advance and relate quantitative, computer-extracted tumor and parenchymal characteristics from multi-modality breast images (e.g, mammography, ultrasound, and MRI) to clinical outcomes (diagnosis, staging, and response to therapy), histopathology, and genomics. This project being led by Maryellen Giger, Nancy Cox, and Robert Grossman at the University of Chicago. The quantitative output from this project also feeds other projects such as the Breast Image Feature Scoring Project.
- Correlating computer-extracted MRI features with clinical and genomic data - In this project, features are extracted from breast MR images using computer vision algorithms. These features are then correlated with clinical and genomic information. This project is being led by Maciej Mazurowski from Duke University.
- Clustering (supervised & unsupervised) of BRCA Data - Cases are clustered into semantically-distinct categories using image-derived features, followed by examination of genomic correlates from the obtained clusters. This project is being led by Arvind Rao and Gary Whitman of MDACC.
Publications
There are currently no imaging based publications for TCGA-BRCA. However, the following links contain publications from the main TCGA project, as well as their posted publication guidelines.