Deep Learning Algorithm For Wound Assessment After Knee Arthroplasty
Research type
Research Study
Full title
The development of a deep learning-based algorithm for automating the assessment of a post-total knee arthroplasty wound.
IRAS ID
340642
Contact name
Pauline Swift
Contact email
Sponsor organisation
Epsom and St Helier University Hospitals NHS Trust
Duration of Study in the UK
1 years, 0 months, 1 days
Research summary
A total knee replacement (TKR) is common intervention for the treatment of end-stage arthritis of the knee. It is an open surgical procedure, leading to a large wound over the front of the knee.
The healing of this wound is an important part of recovery. Healthcare professionals (HCPs) assess healing with a clinical examination (e.g. at SWLEOC, TKR wounds are reviewed in the first and second weeks post procedure). Wound healing is a complex process, and its evaluation requires an HCP’s assessment. Within the patient forum at SWLEOC, patients informed us the healing of their wound was a key concern after hospital discharge which can even affect their overall recovery. These concerns and requests for reviews create a significant demand on healthcare professionals and resources. Wounds are often reviewed by healthcare professionals in the community setting who may lack surgical wound expertise. Additionally, as the number of TKRs being performed increases, so will the demand on HCPs to conduct these reviews.
Our project aims to train a computer program to interpret photographs of knee wounds and identify any wound concerns and whether a review is needed by an HCP. The computer program will allow for immediate assessment of wounds and alert an HCP of any concerns to enable an earlier review. Most wounds heal well and do not require additional visits to the hospital, thus the computer program helps to give patients the reassurance they need. We will take photographs of knee wounds and survey the symptoms of up to 300 patients after a TKR, within one month after their operation. This information will be entered into the computer program and train it to recognise which wounds are not healing well and require a review by an HCP. We will then assess its accuracy, in comparison to an HCP.REC name
West Midlands - Solihull Research Ethics Committee
REC reference
25/WM/0045
Date of REC Opinion
14 Apr 2025
REC opinion
Further Information Favourable Opinion