Advanced Left Ventricle Segmentation from Cardiac MRI Using U-Net with MobileNetV3 Encoder

Abstract

Assessing cardiac function and diagnosing different heart illnesses rely on accurate left ventricle (LV) identification using cardiac magnetic resonance imaging (MRI). To efficiently and accurately segment the left ventricle from 2D cardiac MRI data, this study introduces a novel method that combines a U-Net model with a MobileNetV3 encoder. The ACDC dataset, which includes MRI images and associated ground truth masks, underwent rigorous preprocessing and hyperparameters were adjusted to improve model performance. The evaluation resulted in an average dice score of 92.13%, with the LV segment receiving a dice score of 96.16%, displaying greater performance compared to previous studies. The combination of MobileNetV3 and U-Net has been proven to be effective for medical image segmentation, thereby enhancing diagnostic procedures and ultimately improving patient outcomes.