PERBANDINGAN POTENSI APLIKASI DAN PROGRAM BERBASIS CNN-ML (CONVOLUTIONAL NEURAL NETWORKS-MACHINE LEARNING) SEBAGAI DIAGNOSIS DAN SEGMENTASI ULKUS DEKUBITUS

Abstract

Introduction: Decubitus ulcers are damage to the skin and tissue due to pressure. This condition is quite a burden, especially for health workers because it is difficult to predict and classify the level of lesions that occur. However, decubitus ulcers can be prevented in several ways, one of the most effective ways is by using examination or evaluation tools that can predict, diagnose and assess wound segmentation in decubitus ulcers. 

Discussion: As technology develops, Machine Learning is now available which functions to increase the accuracy of clinical data analysis from electronic medical records on a large scale, namely Convolutional Neural Network (CNN). CNN is a method that applies computer vision technology to monitor the wound healing process accurately. In addition, CNN can classify different types of tissue, granulation, necrotic tissue, and peeling surfaces in decubitus ulcers so that it can classify the severity of the wound. Various applications have been developed using a CNN basis, including the AlexNet, VGG, ResNet, Inception-ResNet-v2 algorithms in the Pressure Ulcer Assessment System (PUAS), U-Net, DeeplabV3, PsPnet, FPN, and Mask R-CNN in Wound Segmentation Detection and Tissue Ulcers. 

Conclusion: Among various CNN-ML models, DeeplabV3 and Inception-ResNet-v2 have the highest F1-score values. This indicates that the two CNN-ML models have good precision and accuracy in diagnosing and segmenting lesions in decubitus ulcers.