Deteksi Struktur Jantung pada anak menggunakan CNN Arsitektur YOLO versi 5

Abstract

 A major challenge in the medical field is detecting heart structures in children, which requires a high level of time and accuracy. To address this issue, the You Only Look Once version 5 (YOLO v5) method is employed to identify children's heart structures using a convolutional neural network (CNN). YOLO v5s, YOLO v5n, and YOLO v5x are three versions tested to identify children's heart structures. Standard evaluation metrics such as precision, recall, F1 score, mean average precision, and IoU threshold 0.5 (mAP_0.5) are used to assess the model's performance. Experimental results indicate that YOLO v5s demonstrates the best performance in detecting children's heart structures with high detection rates and accuracy. This model can effectively detect heart structures in various image positions and conditions, suggesting potential for more accurate and effective diagnostic use in identifying heart diseases in children. The development of heart structure detection models is highly relevant in the medical field. The deep learning model using YOLO v5s offers remarkable capabilities in various visual applications. This model can be an efficient and reliable solution in various fields, providing reliable and accurate performance to streamline data analysis processes and enhance work efficiency.

 Keywords—Detection, Pediatric Cardiac Structures, Convolutional Neural Network, YOLO v5