Convolutional Neural Networks (CNNs) are an outstanding branch of deep learning applications to visual purposes and it has earned major attention in the last years due to its breakthrough performances in varied computer vision application. However, CNNs often suffer from overfitting and sensitivity to noise in complex medical images despite their exceptional capabilities. While metaheuristic optimization techniques like Genetic Algorithms (GAs) can mitigate these limitations, but struggle with slow convergence to achieve global optima solution. Over the years, improvement over Genetic Algorithm has produced Elitist Genetic Algorithm (EGA) which is known as an efficient algorithm to solve complex optimization problems. Hence, this research developed an optimized CNN using EGA to fine tune the hyper parameters of the CNN to solve the limitation of overfitting and sensitivity to noise in medical images. Cardiac Magnetic Resonance Images (CMRI was acquired from Kaggle which is an online repository. The images were preprocessed and segmented using contrast enhancement and fuzzy c-means respectively. The features were classified using the optimized CNN (EGA+CNN). The performance evaluation was performed based on False Positive Rate which is 1.2%, Specificity rate of 95%, Precision value of 95%, Accuracy of 98% and computational time of 73.3seconds. The developed EGA+CNN outperforms the conventional CNN in detecting heart attack in MRI, this makes it a reliable diagnostic tool to reduce human error and contribute to earlier and more accurate heart attack diagnoses, improving patient survival rates. 

Keywords: Artificial Intelligence, Machine learning, Convolutional Neural Network, Genetic Algorithm, Elitist Genetic Algorithm, Cardiac Magnetic Resonance Images, Heart attack.