Classification of Diabietic Foot Thermograms using Deep Learning Matlab
Abstract
According to the World Health Organization (WHO), Diabetes Mellitus (DM) is one of the most prevalent diseases in the world. It is also associated with a high mortality index. Diabetic foot is one of its main complications, and it comprises the development of plantar ulcers that could result in an amputation. Several works report that thermography is useful to detect changes in the plantar temperature, which could give rise to a higher risk of ulceration. However, the plantar temperature distribution does not follow a particular pattern in diabetic patients, thereby making it difficult to measure the changes. Thus, there is an interest in improving the success of the analysis and classification methods that help to detect abnormal changes in the plantar temperature. All this leads to the use of computer-aided systems, such as those involved in artificial intelligence (AI), which operate with highly complex data structures. This paper compares machine learning-based techniques with Deep Learning (DL) structures. We tested common structures in the mode of transfer learning, including AlexNet and GoogleNet. Moreover, we designed a new DL-structure, which is trained from scratch and is able to reach higher values in terms of accuracy and other quality measures. The main goal of this work is to analyze the use of AI and DL for the classification of diabetic foot thermograms, highlighting their advantages and limitations. To the best of our knowledge, this is the first proposal of DL networks applied to the classification of diabetic foot thermograms. The experiments are conducted over thermograms of DM and control groups. After that, a multi-level classification is performed based on a previously reported thermal change index. The high accuracy obtained shows the usefulness of AI and DL as auxiliary tools to aid during the medical diagnosis.
Introduction
Diabetes Mellitus (DM) is one of the leading worldwide causes of death and degrading elements in the quality of life of those affected by the disease. There are several complications associated with DM, including heart attacks, vision loss, kidney failure, and amputations of inferior limbs. Such difficulties not only affect people’s health, but they also have a significant impact on their personal and working life. Diabetic foot is one of the main complications. It has been reported that a loss of sensitivity in the diabetic foot, along with mechanical stress in the plantar region, may increase the risk of ulceration , which can lead to an amputation. It is also known that an increase in temperature in the plantar region of diabetic patients is associated with a higher risk of ulceration. Hence, the interest in monitoring the temperature frequently through different approaches has arisen.
Thermography is a technique that has been applied to the study of diabetic foot by analyzing the thermal changes that occur in the affected foot. This technique presents two main advantages. Firstly, contact is not required, and, secondly, it is non-invasive. Several works concerning its use to study diabetic foot have been reported. Two main approaches have been proposed for thermogram analysis. These involve identifying characteristic patterns and measuring thermal changes. On one hand, a control group has been demonstrated to show a particular spatial pattern, called butterfly pattern. However, there is a wide variation of spatial patterns in the DM group. On the other hand, it is possible to measure the thermal changes and to make an assessment of these with respect to a reference. Some works propose a contralateral comparison of temperature by assuming that one foot serves as a reference to the other. However, the contralateral comparison is limited when one of the feet cannot serve as reference. For example, if both feet have changes in temperature, but neither one of them has the butterfly pattern, then one cannot be a reference of the other. If changes in both feet have similar spatial distributions, asymmetry will not be detected, even when there is a significant change in temperature. An alternative to this approach is to measure the changes by computing a representative value for each foot of the DM group and taking the butterfly pattern as reference. Therefore, the measurement depends on the temperature distribution, and not on a spatial pattern. This kind of analysis can help to describe thermograms to improve their automatic classification and bring additional information to the medical expert.