Label-Conditioned Diffusion for CT-Driven Echocardiography Synthesis in TEE Geometry

This project aims to develop a two-stage, unpaired learning framework that synthesizes realistic, patient-specific transesophageal echocardiography (TEE) images directly from routine pre-procedural CT scans. By eliminating the need for large-scale paired clinical datasets and complex probe-pose modeling, this approach facilitates more accessible TEE-guided structural heart interventions.