Classification of Diabetic Foot Ulcers from Images Using Machine Learning Approach

Abstract

Diabetic foot ulcers (DFUs) represent a significant and serious challenge associated with diabetes. It is estimated that approximately one third of individuals with diabetes will develop DFUs at some point in their lives. This common complication can lead to serious health issues if not properly managed. The early diagnosis and treatment of DFUs are crucial to prevent severe complications, including lower limb amputation. DFUs can be categorized into two states: ischemia and infection. Accurate classification is required to avoid misdiagnosis due to the similarities between these two states. Several convolutional neural network (CNN) models have been used and pre-trained through transfer learning. These models underwent evaluation with hyperparameter tuning for the binary classification of different states of DFUs, such as ischemia and infection. This study aimed to develop an effective classification system for DFUs using CNN models and machine learning classifiers utilizing various CNN models, such as EfficientNetB0, DenseNet121, ResNet101, VGG16, InceptionV3, MobileNetV2, and InceptionResNetV2, due to their excellent performance in diverse computer vision tasks. Additionally, the head model functions as the ultimate component for making decisions in the model, utilizing data collected from preceding layers to make precise predictions or classifications. The results of the CNN models with the suggested head model have been used in different machine learning classifiers to determine which ones are most effective for enhancing the performance of each CNN model. The most optimal outcome in categorizing ischemia is a 97% accuracy rate. This was accomplished by integrating the suggested head model with the EfficientNetB0 model and inputting the outcomes into the logistic regression classifier. The EfficientNetB0 model, with the proposed modifications and by feeding the outcomes to the AdaBoost classifier, attains an accuracy of 93% in classifying infections.

Keywords: convolutional neural networks; deep learning; diabetic foot ulcers; diabetics; diagnosis; image classifier; transfer learning.

Figure 1

Framework of the proposed approach for infection classification and ischemia classification.

Figure 2

A sample of ischemia images from DFU-Part (B) dataset: (a) before augmentation and (b–e) newly generated ischemia images after augmentation.

Figure 3

A sample of infection images from DFU-Part (B) dataset: (a) before augmentation, and (b–e) newly generated infection images after augmentation.

Figure 4

Proposed head model structure.

Figure 5

ROC curves of the CNN pre-trained models for (a) ischemia and (b) infection classification.

Figure 6

ROC curves of the CNN pre-trained models with the proposed head model for (a) ischemia and (b) infection classification.

Figure 7

Results of the accuracy of EfficientNetB0 in infection classification (top row) and ischemia classification (lower row) on the DFU-Part (B) dataset: (a,c) before augmentation and (b,d) after augmentation.

Figure 8

ROC curves of the CNN pre-trained models with the proposed head model after feeding the results to ML classifier for (a) ischemia and (b) infection classification.

Figure 9

Confusion matrix of EfficientNetB0 with the proposed head model after feeding the results of them to the LogisticRegression classifier for ischemia classification.

Figure 10

Confusion matrix of EfficientNetB0 with the proposed head model after feeding the results of them to the AdaBoostClassifier classifier for infection classification.

Figure 11

Sample of correctly classified images from DFU-Part (B) dataset of ischemia for ischemia classification.

Figure 12

Sample of correctly classified images from DFU-Part (B) dataset of infection for infection classification.

Figure 13

Training and validation loss curves of each pre-trained model before and after adding the proposed head model in ischemia classification (a) ResNet101, (b) Modified ResNet101, (c) DenseNet121, (d) Modified DenseNet121, (e) VGG16, (f) Modified VGG16, (g) InceptionV3, (h) Modified InceptionV3, (i) MobileNetV2, (j) Modified MobileNetV2, (k) InceptionResNetV2, (l) Modified InceptionResNetV2, (m) EfficientNetB0, and (n) Modified EfficientNetB0.

Figure 14

Training and validation loss curves of each pre-trained model before and after adding the proposed head model in infection classification(a) ResNet101, (b) Modified ResNet101, (c) DenseNet121, (d) Modified DenseNet121, (e) VGG16, (f) Modified VGG16, (g) InceptionV3, (h) Modified InceptionV3, (i) MobileNetV2, (j) Modified MobileNetV2, (k) InceptionResNetV2, (l) Modified InceptionResNetV2, (m) EfficientNetB0, and (n) Modified EfficientNetB0.

Figure 15

Training and validation accuracy curves of each pre-trained model before and after adding the proposed head model in ischemia classification (a) ResNet101, (b) Modified ResNet101, (c) DenseNet121, (d) Modified DenseNet121, (e) VGG16, (f) Modified VGG16, (g) InceptionV3, (h) Modified InceptionV3, (i) MobileNetV2, (j) Modified MobileNetV2, (k) InceptionResNetV2, (l) Modified InceptionResNetV2, (m) EfficientNetB0, and (n) Modified EfficientNetB0.

Figure 16

Training and validation accuracy curves of each pre-trained model before and after adding the proposed head model in infection classification (a) ResNet101, (b) Modified ResNet101, (c) DenseNet121, (d) Modified DenseNet121, (e) VGG16, (f) Modified VGG16, (g) InceptionV3, (h) Modified InceptionV3, (i) MobileNetV2, (j) Modified MobileNetV2, (k) InceptionResNetV2, (l) Modified InceptionResNetV2, (m) EfficientNetB0, and (n) Modified EfficientNetB0.