A new imaging era aided by artificial intelligence to enhance cardio-oncology care. Tun-AI Enhance Study

3D echocardiography is recommended as the preferred echocardiographic modality to measure left ventricular ejection fraction (LVEF). This technique needs an experimented operators and expensive machines. Artificial Intelligence (AI) based systems were reported to be of clinical interest. This study aims to compare performances and variability of echocardiographic measures between a 3D-equipped ultrasound system and an AI equipped one for cancer patients while assessing inter-oberver variability between different expertise level operators. This was a prospective study done between May 2023 and February 2024 in the cardiology departement of Rabta hospital. It included patients > 18 years with a malignant tumor and candidates for cardiotoxic chemotherapy. We excluded patients who had started chemotherapy and who had an irregular rythm. TTE were performed by a senior operator, using a Philips EPIC7 ultrasound system to measure the 3D LVEF. An Echonous Kosmos system equipped with an US2-AI software was used for automatic LVEF measurement by a senior operator and a junior one. This study included 47 patients (mean age 50 ± 12 years, 68% female). There was no statistically significant difference in LVEF measurments when performed by the senior operator using the 3D echocradiography or the AI systems ( P = 0.16). When using the AI system, there was no statiscally significant difference between the results found by the senior and the junior doctor ( P = 0.918). There was a correlation between 3D measured LVEF and AI-measured LVEF when done by the senior operator ( r = 0.43, P = 0.003). A correlation between AI-measured LVEF by the senior operator and the junior operator was found ( r = 0.571, P < 10 −3 ). The AI echocardiography system demonstrates potential to reduce inter-observer variability, minimize expertise-related errors and improve test performance in the assessment of left ventricular function.