A Study of Data Augmentation Techniques to Overcome Data Scarcity in Wound Classification using Deep Learning

TL;DR

It is shown that geometric data augmentation can improve classification performance, F1 scores, by up to 11% on top of state-of-the-art models, across several key classes of wounds, and it is believed it will be part of any real-world ML-based wound care system.

Abstract

Chronic wounds are a significant burden on individuals and the healthcare system, affecting millions of people and incurring high costs. Wound classification using deep learning techniques is a promising approach for faster diagnosis and treatment initiation. However, lack of high quality data to train the ML models is a major challenge to realize the potential of ML in wound care. In fact, data limitations are the biggest challenge in studies using medical or forensic imaging today. We study data augmentation techniques that can be used to overcome the data scarcity limitations and unlock the potential of deep learning based solutions. In our study we explore a range of data augmentation techniques from geometric transformations of wound images to advanced GANs, to enrich and expand datasets. Using the Keras, Tensorflow, and Pandas libraries, we implemented the data augmentation techniques that can generate realistic wound images. We show that geometric data augmentation can improve classification performance, F1 scores, by up to 11% on top of state-of-the-art models, across several key classes of wounds. Our experiments with GAN based augmentation prove the viability of using DE-GANs to generate wound images with richer variations. Our study and results show that data augmentation is a valuable privacy-preserving tool with huge potential to overcome the data scarcity limitations and we believe it will be part of any real-world ML-based wound care system.

Figures (8)

• Figure 1: A flow chart representing the overall methodology employed in the study
• Figure 2: Examples of all categories of wounds addressed in this study
• Figure 3: A Flow Chart Representing the Parts of a DE-GAN zhong2020generativeNote. The right diagram represents the structure of the variational autoencoder in the DE-GAN, and the left diagram represents the structure of the naive GAN component of the DE-GAN.
• Figure 4: The Combined Loss Function Used by the DE-GAN as described in zhong2020generative
• Figure 5: Confusion Matrix of Top Two Models - Note. The right image shows the confusion matrix of the base model VGG16 with 30 epochs and 0.0005 learning rate. The left image shows the confusion matrix of the base model ResNet50 with 37 epochs and 0.0005 learning rate.