Artificial Intelligence for Diabetic Foot Screening Based on Digital Image Analysis: A Systematic Review
Clinical Snapshot
PICO Framework
| P — Population | Individuals with diabetes mellitus at risk of or diagnosed with diabetic foot complications |
| I — Intervention | Artificial intelligence (AI) models utilising digital image analysis (including thermal imagery, thermograms, and plantar temperature distribution) for diabetic foot screening |
| C — Comparator | Not explicitly defined; implied comparison to conventional clinical screening or no AI-assisted screening |
| O — Outcomes | Diagnostic accuracy of AI models (accuracy, sensitivity, specificity); classification performance of deep learning architectures (ANN, CNN) for diabetic foot lesion detection and characterisation |
Bottom Line
This systematic review examines AI-based digital image analysis for diabetic foot screening, identifying nine studies published between 2018 and 2023. While the topic is clinically important — diabetic foot complications remain a leading cause of preventable amputation — the methodological quality of this review is substantially limited. The search strategy covers only three databases, omitting major engineering and clinical repositories. No PROSPERO registration is reported, no meta-analysis or pooled diagnostic accuracy statistics are presented, and no GRADE certainty of evidence assessment is performed. The headline finding of 97.5% classification accuracy from a single ANN model should be interpreted with considerable caution: it derives from a model development study without independent clinical validation, and confidence intervals are absent. Thermal imaging and deep learning architectures show genuine promise as adjuncts to clinical diabetic foot assessment, but the current evidence base — characterised by small, heterogeneous, non-validated studies — does not support changes to clinical practice. Senior clinicians and health services should monitor this field closely but await prospective, multi-centre clinical validation studies before considering implementation. In the Australian context, TGA regulatory approval and MBS pathway development would be prerequisites for any clinical adoption.
Key Findings
P Value: Not reported
Effect Size: Best-performing model: ANN using MATLAB's Image Processing Toolbox achieved 97.5% classification accuracy for macula type identification; CNN-based models also reported high accuracy across included studies (specific figures not pooled)
Primary Outcome: Diagnostic accuracy of AI models for diabetic foot screening using digital image analysis, with thermal imagery and plantar temperature distribution as the predominant data source
Nnt Or Sensitivity: Sensitivity and specificity not pooled or reported; AUC values not synthesised; individual study performance metrics not systematically tabulated in the abstract
Confidence Interval: Not reported — no confidence intervals provided for any accuracy estimates across included studies
Clinical Application
AI-assisted thermal imaging for diabetic foot screening is technically feasible in research settings but remains at an early developmental stage for routine clinical deployment. Barriers include cost of thermal imaging equipment, lack of standardised image acquisition protocols, absence of prospectively validated clinical decision thresholds, and limited integration with electronic health record systems. Workforce training requirements and medicolegal considerations for AI-assisted diagnosis are unresolved. In Australia, diabetic foot disease represents a significant burden, with approximately 4,400 lower limb amputations performed annually in people with diabetes (Australian Institute of Health and Welfare data). The RACGP Standards for General Practice and the Australian Diabetes Society guidelines recommend annual foot examinations for all people with diabetes, creating a potential implementation pathway for AI-assisted screening tools. However, no AI-based diabetic foot imaging system is currently listed on the Australian Register of Therapeutic Goods (ARTG) as a Class IIb or III medical device, and no Medicare Benefits Schedule (MBS) item exists for AI-assisted diabetic foot thermography. The TGA's Software as a Medical Device (SaMD) regulatory framework would apply to any clinical deployment. PBS-listed wound care products and podiatry services under Enhanced Primary Care plans remain the current standard of care. This review does not provide sufficient evidence to support TGA submission or MBS listing applications. Adults with type 1 or type 2 diabetes mellitus undergoing routine diabetic foot surveillance, particularly those at intermediate to high risk of foot ulceration (Wagner grade 0–1 or IWGDF risk categories 1–3)
Abstract
INTRODUCTION: Early detection of diabetic foot complications is essential for effective management and prevention of complications. Artificial intelligence (AI) technology based on digital image analysis offers a promising noninvasive method for diabetic foot screening. This systematic review aims to identify a study on the development of an AI model for diabetic foot screening using digital image analysis. METHODOLOGY: The review scrutinized articles published between 2018 and 2023, sourced from PubMed, ProQuest, and ScienceDirect. The keyword-based search resulted in 2214 relevant articles and nine articles that met the inclusion criteria. The article quality assessment was done through Quality Assessment of Diagnostic Accuracy Studies (QUADAS). Data were extracted and analyzed using NVivo. RESULTS: Thermal imagery or foot thermogram was the main data source, with plantar temperature distribution patterns as an important indicator. Deep learning methods, specifically artificial neural networks (ANNs) and convolutional neural networks (CNNs), are the most commonly used methods. The highest performance is demonstrated by the ANN model with MATLAB's Image Processing Toolbox that is able to classify each type of macula with 97.5% accuracy. The findings show the great potential of AI in improving the accuracy and efficiency of diabetic foot screening. CONCLUSION: This research provides important insights into the development of AI in digital image-based diabetic foot screening. Future studies need to focus on evaluating clinical applicability, including ethical aspects and patient data security, as well as developing more comprehensive data sets.
References
- 1.Anggreni, N. K. I. S., Kristianto, H., Handayani, D., Yueniwati, Y., Irawan, P. L. T., Rosandi, R., Kapti, R. E., & Purnama, A. D. (2026). Artificial intelligence for diabetic foot screening based on digital image analysis: A systematic review. Journal of Diabetes Science and Technology. Advance online publication. https://doi.org/10.1177/19322968251317521
Related Research
Analytical chemistry
Biodegradable Self-Powered Electrotherapy Patch for Integrated Smart Wound Management
15 July 2026
Lasers in medical science
Clinical dosimetry and efficacy of LED photobiomodulation for chronic lower-limb wound healing: a systematic review of randomized trials
20 June 2026
This content is for educational purposes for healthcare professionals only and does not constitute clinical advice. Clinical decisions should be based on individual patient assessment, current guidelines, and appropriate specialist consultation. Editorial Standards · Privacy Policy · Terms of Service