Pittsburgh Liver Research Center/Director’s Innovative Award

University of Pittsburgh

 

The University of Pittsburgh & UPMC Pittsburgh Liver Research Center (PLRC) is pleased to announce the Director’s Award in New Direction and Innovation (DANDI). In March of 2021, the PLRC/University of Pittsburgh awarded a grant aimed at developing and evaluating artificial intelligence models for the non-invasive detection of beta-catenin in HCC patients. These models will utilize MRI and CT images and be able to non-invasively detect and identify beta-catenin mutations to optimize treatment planning and provide prognostic assessment.

 

We are delighted to announce PLRC member and junior investigator Dr. Dooman Arefan, in collaboration with PLRC member Dr. Alessandro Furlan, has been selected as recipient of the PLRC–DANDI! The award period is March 15, 2021 – March 14, 2022. The awardee will work closely with PLRC members and lab mentor, Dr. Shandong Wu, and Dr. Paul Monga to help assess the development of AI in clinical use of beta-catenin detection/identification. PLRC members Dr. Alessandro Furlan and Dr. Aatur Singhi will lend their clinical expertise as clinical co-mentors.

“A radio-genomics study for predicting beta-catenin mutation in hepatocellular carcinoma patients using multi-modal imaging data and artificial intelligence”

Clinical Co-Mentors – Dr. Alessandro Furlan, Dr. Aatur Singhi

Research Co-Mentors – Dr. Shandong Wu, Dr. Paul Monga

Alessandro Furlan, MD
Aatur D. Singhi, MD, PhD
dooman_arefan
Shandong Wu, PhD
Paul Monga, MD

A radio-genomics study for predicting beta-catenin mutation in hepatocellular carcinoma patients using multi-modal imaging data and artificial intelligence

Abstract
Hepatocellular carcinoma (HCC) is a major health-care issue causing about 700,000 deaths each year worldwide and being associated with high cost of care. Current clinical guidelines allow a non-invasive diagnosis of HCC without the need for a confirmatory biopsy when the lesion shows typical imaging features on liver magnetic resonance imaging (MRI) or computed tomography (CT). MRI and CT are typical diagnostic imaging for HCC management; they can provide a volumetric characterization on HCC tumor’s heterogeneity and may carry more information to inform prognosis and treatment of HCC patients. CTNNB1, one of the primary oncogenes involved in HCC development, encodes beta-catenin, a protein that serves multiple important cellular functions. Beta-catenin mutation may affect prognosis of HCC and offer a target for established and novel therapeutic agents. The unmet clinical need is a non-invasive assessment of the beta-catenin status of HCC in order to facilitate the decision-making process and inform prognosis in the context of a “precision medicine” approach. Artificial intelligence (AI) has been revolutionizing medical imaging analysis where radiomics and machine/deep learning are representative computational modeling techniques. AI provides game-changing ways to quantitatively interpret radiological images in-depth (especially volumetric images), going beyond the qualitative assessment provided by the interpreting radiologists. The goal of this study is to develop and evaluate AI models (i.e., a radiomic model and a deep learning model) for the non-invasive detection of beta-catenin status in HCC patients, using both liver MRI and CT images. The AI models will be able to detect/identify HCC patients with beta-catenin mutation in a non-invasive way and using standard-of-care clinical imaging data, which has great potential to 1) avoid unnecessary invasive procedures, 2) optimize treatment planning and provide prognostic assessment, 3) improve efficiency of clinical procedures and reduce costs, and 4) improve overall patient experience.

Relevance
We propose to develop and evaluate artificial intelligence (AI) models (i.e., a radiomic model and a deep learning model) for the non-invasive detection of beta-catenin status in hepatocellular carcinoma (HCC) patients, using both liver magnetic resonance imaging (MRI) and computed tomography (CT) images. The AI models will be able to detect/identify HCC patients with beta-catenin mutation in a non-invasive way to avoid unnecessary invasive procedures and optimize treatment planning and provide prognostic assessment.