Fully Automated Wound Tissue Segmentation Using Deep Learning on Mobile Devices: Cohort Study

April 22, 2022|Published Research

Published on JMIR Publications -

The composition and identification of wound tissue types can be subjective, which can result in inaccurate assessments and possible misdiagnoses of wound healing progression. A study was conducted to determine if a more objective assessment of tissue types can be achieved using deep neural networks. The study used 58 anonymized wound images of different types of chronic wounds from Swift Medical’s Wound Database. The aim was to determine inter- and intra-rater agreement. 

The study revealed that the proposed deep learning technique provides objective tissue identification and sizing, which can help to document wounds more accurately. The deep neural networks were 30% better at identifying and segmenting tissue types than wound care specialists. 

Read more: https://mhealth.jmir.org/2022/4/e36977 

Subscribe for Email Updates

hbspt.forms.create({ region: "na1", portalId: "2024649", formId: "3a7abf57-238a-43f7-b8ae-8e6bae308e60" });