Computational Feature Profiling for Prostate Cancer

 Prostate cancer is a leading cause of cancer death in American men, but current risk stratification methods are often ineffective, leading to significant overdiagnosis and overtreatment. Up to 60% of patients diagnosed with prostate cancer receive unnecessary treatment, which can result in long-term reductions in functional outcomes and high costs. This research aims to investigate new methods for more accurately detecting aggressive cancer and assessing its aggression, ultimately reducing overdiagnosis and overtreatment. The project's objective is to develop novel markers and models to forecast the arrival of aggressive cancer. Key components of the research include:

• Identifying novel pathomic and germline features that indicate aggressive cancer or its precursors.
• Implementing an integrative graph convolutional network (GCN) combined with a convolutional neural network (CNN) to generate new multi-modal representations of the underlying cancer state within the entire prostate. This framework will combine multiparametric magnetic resonance imaging (mpMRI), digital histology images, germline features, biomarkers, and other predictors.
• Developing new nomogram models that incorporate the newly identified features for improved risk stratification.

The ultimate goal is to prevent patients from receiving unnecessary interventions and incurring associated negative functional outcomes.

Contact PI: Corey Arnold

Funding Source: NIH NCI