TY - JOUR

T1 - Right-Ventricular Myocardial Perfusion Gated SPECT Data Processing Using a Deep Learning Model

AU - Ларочкин, Петр Викторович

AU - Котина, Елена Дмитриевна

PY - 2026/6/30

Y1 - 2026/6/30

N2 - A sequence-based deep learning pipeline is proposed for automated processing of gated SPECT studies of the right ventricle. The model processes a temporal sequence of 3D tomographic volumes and predicts right-ventricular myocardial masks for each cardiac phase. Based on the predicted segmentation masks, clinical parameters are computed and perfusion polar maps of the “bull's eye” type, as well as magnitude and phase planar maps, are constructed. In experiments, the recurrent model achieved a Dice score of up to 0.8119.

AB - A sequence-based deep learning pipeline is proposed for automated processing of gated SPECT studies of the right ventricle. The model processes a temporal sequence of 3D tomographic volumes and predicts right-ventricular myocardial masks for each cardiac phase. Based on the predicted segmentation masks, clinical parameters are computed and perfusion polar maps of the “bull's eye” type, as well as magnitude and phase planar maps, are constructed. In experiments, the recurrent model achieved a Dice score of up to 0.8119.

KW - deep learning

KW - image segmentation

KW - myocardial perfusion imaging

KW - gSPECT

KW - functional images

KW - Gated single photon emission computed tomography (gSPECT)

KW - deep learning

KW - functional images

KW - image segmentation

KW - myocardial perfusion imaging

UR - https://www.mendeley.com/catalogue/4c8a7888-33af-305d-8776-d17baa9cd214/

U2 - 10.35470/2226-4116-2026-15-1-64-70

DO - 10.35470/2226-4116-2026-15-1-64-70

M3 - Article

VL - 15

SP - 64

EP - 70

JO - Cybernetics and Physics

JF - Cybernetics and Physics

SN - 2223-7038

IS - 1

ER -