Automated Left Ventricle Segmentation in MRI Using UNet Architecture in MATLAB

Automated segmentation of the left ventricle (LV) in cardiac magnetic resonance imaging (MRI) is crucial for accurate diagnosis and treatment planning of cardiovascular diseases. This study presents an efficient method for segmenting the left ventricle in MRI images using the UNet architecture implemented in MATLAB. The process begins with a curated dataset of MRI left ventricle images and their corresponding ground truth segmentations. Input images are pre-processed, including resizing to a uniform dimension, to ensure consistency and compatibility with the UNet model. The UNet, a state-of-the-art semantic segmentation network, is trained using this dataset under optimized training conditions, including hyperparameter tuning and data augmentation, to enhance model generalization. The final segmented images are evaluated using Peak Signal-to-Noise Ratio (PSNR) to quantify the quality and fidelity of segmentation compared to the ground truth. This approach not only provides precise LV delineation but also demonstrates high computational efficiency, making it suitable for real-world clinical applications. By leveraging the robust features of UNet and MATLAB's versatile image processing tools, the proposed method addresses challenges in LV segmentation, such as variability in image quality and anatomical structure, achieving accurate and reproducible results. This work highlights the potential of deep learning-based methods in automating complex medical imaging tasks, contributing to advancements in non-invasive cardiac assessment. 

Keywords: Dataset, Image Processing Techniques, UNet, Sematic segmentation and PSNR.