Deep Learning-Driven 68GaFAPI-04 PET/CT Imaging for Cardiac Fibrosis Assessment (FIBRAI)

• Grant number: 25/11759-5
• Support Opportunities: Regular Research Grants
• Start date: June 01, 2026
• End date: May 31, 2029
• Field of knowledge: Engineering - Biomedical Engineering
• Principal Investigator: Marco Antonio Gutierrez
• Grantee: Marco Antonio Gutierrez
• Principal researcher abroad: Seng Peng Mok
• Institution abroad: Universidade de Macau (UM) , China
• Host Institution: Instituto do Coração Professor Euryclides de Jesus Zerbini (INCOR). Hospital das Clínicas da Faculdade de Medicina da USP (HCFMUSP). Secretaria da Saúde (São Paulo - Estado). São Paulo , SP, Brazil
• City of the host institution: São Paulo
• Associated researchers: Estela Ribeiro ; José Eduardo Krieger ; Marina de Fátima de Sá Rebelo
• Associated scholarship(s): 26/13159-8 - Multimodal Artificial Intellligence Approach for Myocardial Fibrosis Segmentation and Clinical Decision Support, BP.PD; 26/13236-2 - Support for Artificial Intelligence Infrastructure and Data Management, BP.TT

Abstract

This research proposal aims to develop and validate a deep learning-based framework for the automated segmentation, quantification, and analysis of cardiac fibrosis using 68Ga-FAPI-04 PET/CT imaging in clinical populations. Fibroblast activation protein (FAP) is overexpressed in activated fibroblasts, which play a central role in the pathogenesis of cardiac fibrosis. PET imaging with ¿¿Ga-FAPI-04 enables noninvasive, molecular-level assessment of fibroblast activity, offering greater sensitivity than conventional imaging modalities.Leveraging recent advances in convolutional neural networks (CNNs), particularly attention U-Net and Transformer-based architectures, we propose to automate the segmentation of fibroblast activity and extract quantitative biomarkers such as SUV_max, SUV_mean, and FAPI-positive volumes. These PET-derived biomarkers will be validated against quantitative parameters obtained from cardiac magnetic resonance imaging (LGE), echocardiography, and clinical outcomes, with emphasis on diagnostic accuracy, prognostic value, and therapy monitoring.The successful implementation of this AI-driven approach has the potential to significantly enhance the clinical utility of FAPI PET imaging in the management of cardiovascular diseases, particularly for the early detection of fibrosis and the monitoring of therapeutic interventions. (AU)