Artificial Intelligence in Medical Imaging: Segmentations and Classification of Cardiac Diseases

Authors

Mosa Alassiri, Physics Department, Faculty of Sciences, King Abdulaziz University, Jeddah, 21589, Saudi ArabiaFollow
Samia Faisal Aboushoushah, King Abdulaziz University, Physics Department, Faculty of Sciences, King Abdulaziz University, Jeddah, 21589, Saudi ArabiaFollow
Wassim Zouch, King Abdulaziz University, Jeddah, 21589, Saudi ArabiaFollow
Ahmad M Alqahtani, King Saud Medical City, Riyadh, 11196, Saudi ArabiaFollow
Salha Mohammed Al Makwash, King Abdulaziz University, Physics Department, Faculty of Sciences, King Abdulaziz University, Jeddah, 21589, Saudi ArabiaFollow
Hatem Besbes, Physics Department, Faculty of Sciences, King Abdulaziz University, Jeddah, 21589, Saudi Arabia AND Center for Artificial Intelligence in Precision Medicine, King Abdulaziz University, Jeddah, 21589, Saudi ArabiaFollow

Abstract

Background: Medical imaging refers to the technique of creating visual representations of the structures and functions of various tissues and organs within the human body. AI’s data processing competencies are applied in areas such as predictive analytics, disease modeling, and the automation of routine tasks.

Aim: Making cardio-radiologists’ task easier by applying AI for segmentation and classifications of Short Axis Cardiac MRI (SACMR).

Methods: Datasets were collected from GitHub - Roboflow. Systems used to analyse and learn the data by deep learning is Python 3.9 (64-bit)) and libraries such as Yolo, TensorFlow, and PyTorch. Short-Axis Cardiac MRI (SACMR). Convolutional Neural Networks CNNs have been deemed the ideal method of employment for their acute efficacy in addressing matters related to image scrutiny. Data analysis implies specialized techniques that identify anatomical structures and conduct measurements, which are subsequently reviewed by a cardiologist and radiologist. Machine learning methods have now advanced to aid not only in registration and segmentation but also in carrying out the measurements typically performed by humans.

Results: segmentation of three classes categories, left ventricle, myocardium, and right ventricle. The diagonal elements (0.91 for the left ventricle, 1.00 for myocardium, and 0.64 for the right ventricle) represent the pixel classification accuracy achieved by the proposed model for each class.

Conclusion: AI techniques have revealed superior performance in segmenting and classifying vital cardiac structures, particularly in studies involving deep learning and machine learning models.

First Page

12

Last Page

23

Recommended Citation

Alassiri, Mosa; Aboushoushah, Samia Faisal; Zouch, Wassim; Alqahtani, Ahmad M; Makwash, Salha Mohammed Al; and Besbes, Hatem (2025) "Artificial Intelligence in Medical Imaging: Segmentations and Classification of Cardiac Diseases," The Journal of King Abdulaziz University: Science: Vol. 1: Iss. 1, Article 2.
DOI: https://doi.org/10.64064/1658-4252.1001

Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 International License.