A Look into the Added Value of Artificial Intelligence in Detecting Cardiac Ischemia with Combined Myocardial SPECT and PET Perfusion Imaging and Anatomic Multimodality Imaging

Saturday, April 13, 2024

10:00 AM - 10:00 PM ET

Location: Foyer Kiosk

First Author(s)

JY

Johnny Yang, BS, BA

Medical Student
University of Mississippi Medical Center
University of Mississippi Medical Center, United States

Co-Author(s)

ML

Marjorie Lam, BA

Medical Student
University of Mississippi Medical Center, United States
ES

Easton Sheffield, BS

Medical Student
University of Mississippi Medical Center
University of Mississippi Medical Center
Flowood, Mississippi, United States
CC

Chanukya Cherukuri, BS

Medical Student
University of Mississippi Medical Center
University of Mississippi Medical Center, United States
BB

Bryson Brister, BS

Medical Student
University of Mississippi Medical Center
University of Mississippi Medical Center
Brandon, Mississippi, United States
JT

John Hollis Tackett, BS

Student
University of Mississippi Medical Center
University of Mississippi Medical Center
Flowood, Mississippi, United States
VV

Vani Vijayakumar, MD

Professor
University of Mississippi Medical Center, United States

Purpose: To review role of artificial intelligence (AI) on Myocardial Perfusion imaging (MPI) in ischemic heart disease with anatomic imaging.

Methods/Materials: Databases Medical Literature Analysis and Retrieval System Online (MEDLINE)/PubMed and Google Scholar were used for the literature search between 2013 and 2023.

Results: AI will play a pivotal role in echocardiography, coronary computed tomography angiography, magnetic resonance, and MPI, improving efficiency and reliability for clinicians and sustainability in healthcare. Nuclear cardiology (i.e. single-photon emission computed tomography (SPECT) and positron emission tomography (PET)) with multimodality imaging plays a fundamental role in the diagnosis and risk stratification of acute and chronic coronary artery disease (CAD).

Excluding image pre-processing and segmentation, the major AI applications to myocardial perfusion SPECT focused on two principal tasks: identify and prognosticate patients with obstructive CAD. Fully-automated quantification of myocardial perfusion SPECT was equivalent on a per-patient level and superior on a per-vessel level in detecting significant stenosis when compared with expert visual analysis. Learning models have also been applied to myocardial perfusion SPECT to improve prognostic value. Machine learning algorithm performed per-vessel prediction of early coronary revascularization after SPECT myocardial perfusion imaging and showed comparable or better performance with respect to expert readers in predicting early revascularization.

Conclusions: AI with MPI and multimodality imaging will continue to play a prominent role in the diagnostic, risk stratification and follow up of patients with CAD. Clinical validation before adoption to clinical practice is essential. Due to the intrinsically rich information available in flow, perfusion, and CT images, cardiac PET/CT enhanced with AI methods has a true potential to become a frontline modality for CAD.