TY - GEN

T1 - A Deep Learning Model for Wound Size Measurement Using Fingernails

AU - Nguyen, Duc Khanh

AU - Chang, Dun Hao

AU - Nguyen, Thi Ngoc

AU - Nguyen, Trinh Trung Duong

AU - Chan, Chien Lung

N1 - Publisher Copyright: © 2022 ACM.

PY - 2022

Y1 - 2022

N2 - Wound size is an important parameter in the evaluation of healing status of chronic wounds. Many technologies, such as software embedded digital camera or artificial intelligence assisted smart phone applications, have been applied to the automatic wound size measurement. However, these methods or devices are either expensive or inconvenient. Instead of using a ruler to measure wound size, we propose a novel method using fingernails as the reference objects with the combination of two deep learning models. The width of the nail was first detected and computed by RCNN deep learning (DL) model. After that, the width and height of the wound were inferred by those of the bounding box generated from YoloV5 DL model. The wound size can be obtained from the known nail width. The experimental results showed that the mean Pearson correlation coefficient reached 0.914 in comparing the prediction and the standard wound sizes. We believe our proposed model is a simple and effective method for wound size measurement.

AB - Wound size is an important parameter in the evaluation of healing status of chronic wounds. Many technologies, such as software embedded digital camera or artificial intelligence assisted smart phone applications, have been applied to the automatic wound size measurement. However, these methods or devices are either expensive or inconvenient. Instead of using a ruler to measure wound size, we propose a novel method using fingernails as the reference objects with the combination of two deep learning models. The width of the nail was first detected and computed by RCNN deep learning (DL) model. After that, the width and height of the wound were inferred by those of the bounding box generated from YoloV5 DL model. The wound size can be obtained from the known nail width. The experimental results showed that the mean Pearson correlation coefficient reached 0.914 in comparing the prediction and the standard wound sizes. We believe our proposed model is a simple and effective method for wound size measurement.

KW - Deep learning approach

KW - key-points detection

KW - wound detection

KW - wound size measurement

U2 - 10.1145/3545729.3545758

DO - 10.1145/3545729.3545758

M3 - Article in proceedings

AN - SCOPUS:85140001446

SP - 141

EP - 146

BT - ICMHI' 2022

PB - Association for Computing Machinery

T2 - 6th International Conference on Medical and Health Informatics, ICMHI 2022

Y2 - 12 May 2022 through 15 May 2022

ER -