Implementation of Collaborative Filtering-based Recommendation System for Grading Diabetic Foot Thermograms

In recent years, foot thermography has shown an extensive potential for the diagnosis of diabetic foot ulcers. However, how effectively the thermograms are interpreted determines how accurate the detection is. The patient will require less severe treatment, such as amputation, the earlier the ulcer has been identified. Foot thermography still has difficulties in becoming a clinically useful diagnostic tool because there is still no exact imaging methodology or grading system in practice. In this paper, Diabetic Mellitus (DM) foot thermograms from the IEEE data port repository were considered for statistical analysis and development of a grading system. The Thermal Change Index (TCI) data were ranked using a collaborative filtering-based recommender system, and regression analysis was used to examine the relationship between the right and left foot. Angiosomes played an important role in assessing the temperature distribution of the two feet, as evidenced by the results, which show a difference in the correlation coefficient when angiosomes are considered for the analysis. The linear fit for the CG group is (r = 0.89) and for the DM group is (r = 0.83). The paper concludes with suggestions for future work based on the study's findings.