WounDAR: LiDAR and Machine Vision Based Wound Assessment

Wound assessment is one of the most important aspects of wound care as this provides data needed for treatment. Due to mobility restrictions during the pandemic, medical practitioners have been limited to telemedicine, making wound assessment difficult. There is a lack of a consistent and effective wound assessment instrument for Diabetic Foot Ulcers (DFUs). This study addresses this through the development of a proof-of-concept LiDAR and machine vision-based wound assessment tool. Given this prospect, this study uses the HELIOS LiDAR framework to simulate LiDAR measurements and the Open3D library to visualize the DFU wound models. Additionally, this system gathers and visualizes progressing wound dimensions (length, width, depth, and area) which can aid medical practitioners in remote assessment. Possible wound infections are also detected using a densely connected convolutional network (DenseNet), a variation of the convolutional neural networks (CNN), trained using the DFUC2020 and DFUC2021 datasets. Given this simulated setup, the LiDAR and machine-vision-based measurement system was able to gather a result of a maximum of 5% measurement error, while the infection detection model resulted in a test F1-score of 0.6937, both of which are within this research’s acceptable range. This wound assessment framework was created not only to assist telemedicine during the pandemic, but it also hopes to go beyond in future medical applications through the development of LiDAR technology. Overall, this research shows a promising future in sensory wound care and telemedicine, aiding medical professionals in wound assessment.