Feature Ranking by Variational Dropout for Classification Using Thermograms from Diabetic Foot Ulcers

Abstract

Diabetes mellitus presents a high prevalence around the world. A common and long-term derived complication is diabetic foot ulcers (DFUs), which have a global prevalence of roughly 6.3%, and a lifetime incidence of up to 34%. Infrared thermograms, covering the entire plantar aspect of both feet, can be employed to monitor the risk of developing a foot ulcer, because diabetic patients exhibit an abnormal pattern that may indicate a foot disorder. In this study, the publicly available INAOE dataset composed of thermogram images of healthy and diabetic subjects was employed to extract relevant features aiming to establish a set of state-of-the-art features that efficiently classify DFU. This database was extended and balanced by fusing it with private local thermograms from healthy volunteers and generating synthetic data via synthetic minority oversampling technique (SMOTE). State-of-the-art features were extracted using two classical approaches, LASSO and random forest, as well as two variational deep learning (DL)-based ones: concrete and variational dropout. Then, the most relevant features were detected and ranked. Subsequently, the extracted features were employed to classify subjects at risk of developing an ulcer using as reference a support vector machine (SVM) classifier with a fixed hyperparameter configuration to evaluate the robustness of the selected features. The new set of features extracted considerably differed from those currently considered state-of-the-art but provided a fair performance. Among the implemented extraction approaches, the variational DL ones, particularly the concrete dropout, performed the best, reporting an F1 score of 90% using the aforementioned SVM classifier. In comparison with features previously considered as the state-of-the-art, approximately 15% better performance was achieved for classification.

Keywords: deep learning; diabetic foot; feature extraction; infrared; thermography.

Figure 1

Dropout as a feature selection method. The variational parameter ${\varphi }_d$, where $d\in \left[1..D\right]$, controls the probability of the input feature $x_d$ to be dropped out, that is, an estimate of the noise level of $x_d$. The variational parameters are used for feature ranking.

Figure 2

Graphical illustration of the defined angiosomes. The main reference points considered and the proportional foot division are also specified [41,47]. Reference point A was located at the tip of the innermost toe, whereas B at the center of the calcaneal base. Points C and D corresponded to the wider part of the foot. Point E corresponded to the 60% height of the foot.

Figure 3

Deep learning architecture used for the feature selection based on variational dropout approaches. The first layer corresponds to variational dropout as a feature selector, illustrated in Figure 1.

Figure 4

The sparse rate obtained in the variational feature selector using $\tau >0.9$ as threshold in the different cross-validation iterations.