The use of artificial intelligence in three-dimensional imaging modalities and diabetic foot disease: A systematic review

Background

Diabetic foot disease (DFD) is serious complication of diabetes with a multifactorial etiology and carries a significant risk of lower limb amputations. The prevalence of DFD continues to grow globally. Artificial intelligence has been proposed in aiding early detection and risk stratification for ulceration and other major complications, including sepsis, minor or major lower limb amputation, and death. We systematically reviewed the literature available on the use of artificial intelligence in three-dimensional imaging modalities in DFD.

Methods

A literature review was conducted in accordance with PRISMA guidelines. Embase and Medline (via the Ovid interface), CINAHL (via Ebsco Host), Web of Science, and Scopus databases were searched. The gray literature was also reviewed on ClinicalTrials.gov and the National Institute for Health Research journals library. The medical subject headings terms “diabetes” AND “diabetic foot disease” AND “artificial intelligence” and various permutations of three-dimensional imaging modalities, including “computed tomography,” “magnetic resonance imaging” and “positron emission tomography” were used in the primary search string. The articles were independently screened and reviewed by two reviewers.

Results

We identified 4865 studies and removed 102 duplicates. We excluded 4721 during title and abstract screening. Overall, 42 articles underwent full text review and 1 article was included in the final review, which used computed tomography scanning in patients with DFD to create a risk prediction model.

Conclusions

The use of machine learning and deep learning models is still being explored and evaluated in this context. Current methodologies focus on wound imaging classification, plantar thermography and plantar pressures. Specialized models that evaluate three-dimensional imaging are currently primitive and limited in their use; however, they have potential for the generation of suprahuman insights into existing imaging, extraction of novel metadata features, and prediction using integration of multidimensional patient characteristics.