Breast Cancer Research

Lesion Detection

Characterization

Cancer Risk Prediction

Tomosynthesis Reconstruction

RADIOGENOMICS

We study the associations among radiomics, genomics, and clinical characteristics of breast cancer

Breast MRI radiomics biomarkers
Genomics biomarkers
Molecular classification of breast cancer types

Radiogenomics

Convolutional autoencoder to visualize latent space to study the effects of imaging center dependency.

Latent space visualization with and without homogenization of the MRI

Spectral embedding of the latent spaces for MRI volumes before (left) and after (right) normalization. Each color represents MRI slices from the same imaging center. Each marker is 3D embedding of the high dimensional latent space for each MRI slice.

Radiomic Features

Radiogenomic Associations

Pilot study of the key associations between genomic features (outer circle in red color) and radiomic features (inner circle) and age (inner circle in black). Only robust associations are shown using bright colored curves while the rest are shown with increased transparency. The mapping between radiomic features and genomic aberrations may reveal underlying connections between biological processes and the structure and functional changes observed through imaging.

Mapping

Page updated

Report abuse