Using artificial intelligence to improve wound image quality: a feasibility study.
Background: Artificial intelligence or AI has been used successfully in wound measurement and diagnosis. We undertook a feasibility study to improve image quality for wounds healing by primary intention. Method: Within our AI Lab, we used a three-step approach. Firstly, we collected a test set (and...
| Publicado en: | Wounds UK Vol. 16; no. 4; pp. 54 - 60 |
|---|---|
| Autores principales: | , , , |
| Formato: | glossary pictorial research tables/charts Journal Article |
| Publicado: |
SB Communications Group, A Schofield Media Company
2020
|
| Acceso en línea: | Ver este registro en EBSCOhost |
| fields | @attributes: recordID: 1 pdfLink: plink: https://search.ebscohost.com/login.aspx?direct=true&db=ccm&AN=146962858&site=ehost-live header: @attributes: shortDbName: ccm uiTerm: 146962858 longDbName: CINAHL Complete uiTag: AN controlInfo: bkinfo: dissinfo: jinfo: jid: 17466814 5DJ2 jtl: Wounds UK issn: 17466814 maglogo: N pubinfo: dt: 2020 vid: 16 iid: 4 pid: 13783 pub: SB Communications Group, A Schofield Media Company artinfo: ui: 146962858 146962858 146962858 146962858 ppf: 54 ppct: 6 formats: fmt: @attributes: type: P tig: atl: Using artificial intelligence to improve wound image quality: a feasibility study. aug: au: ROCHON, MELISSA JURKIEWICZ, JAMES MORAIS, CARLOS GONDO, TARIRO affil: Quality & Safety Lead for Surveillance, Royal Brompton & Harefield NHS Foundation Trust (RBHT) sug: subj: Artificial Intelligence Wound Healing Evaluation Photography Digital Imaging Quality Improvement Human Pilot Studies Surgical Wound Infection Comparative Studies Algorithms Suture Techniques Descriptive Statistics Wound Assessment Wound Care ab: Background: Artificial intelligence or AI has been used successfully in wound measurement and diagnosis. We undertook a feasibility study to improve image quality for wounds healing by primary intention. Method: Within our AI Lab, we used a three-step approach. Firstly, we collected a test set (and training set) of clear and blurred images and reviewed existing solutions, models and documentation. In the next stage, a model was trained to be able to automatically classify between the two groups. We used a non-wound publicly available data set of blurred and clear images, looking specifically at contours in the images to focus into blur. Finally, the model was tested on a set of wound images, which had been pre-classified by surveillance colleagues (266 images total). Result: We were able set a threshold value on the resulting algorithm, which resulted in a 5% false rejection of good images, while correctly flagging 62% of the images classified as blurred by surveillance colleagues. This operating point was selected for further investigation and user acceptability testing. Conclusion: Clinical wound care and quality improvement using visual data requires high-quality images. pubtype: Academic Journal doctype: glossary pictorial research tables/charts Journal Article ougenre: Article language: English refInfo: holdings: @attributes: islocal: N |
|---|