Abnormality detection of diabetic foot and fetal growth monitoring using thermography along with image processing techniques

Abstract

The correlation between temperature and medical conditions have been well established using thermography by scanning area such as breast, brain, abdomen and spine. In this work, thermal image of diabetic foot and abdominal region of pregnant women of various gestations was analyzed using image processing techniques. The present work focused on segmentation based on pixel index value using histogram and OTSU Thresholding methods and results analyzed. This was different from segmentation and evaluation using pixel temperature values. The advantage of the present work is that the acquired image which has different temperature scale that changes from person to person is indirectly normalized while processing using pixel index. A total of 9-foot images were processed from known cases of diabetic and normal foot. The threshold was found to be 79 ± 3, for detecting the abnormality and greater than 100 as normal. All the thermal images of the diabetic foot analyzed were within this range. Twenty-four women, pregnant with singleton and twin with various gestation periods were also analyzed. The histogram values varied with gestation period and with the presence of twin or singleton. Though the histogram could identify a twin from a singleton with few outliers, the histogram value of a twin in the third trimester was similar to that of a singleton in the third trimester. newline