B Sudha Devi1,D.S Misbha2
1,2, Nesamony Memorial Christian College, Marthandam, India
In recent years, obesity is highly prevalent and is related with increased risk of many diseases. The distribution of abdominal adipose tissue plays a major role to assess central obesity. The basic objective of this study is to develop a novel method for automatic segmentation of visceral adipose tissue(VAT) and subcutaneous adipose tissue(SAT) from abdominal Magnetic resonance imaging(MRI) slices which is implemented in two steps. First, clustering of image is done to classify MR image into adipose tissue and non-adipose tissue. Second, after clustering the image, segmentation is done to separate VAT and SAT by a convolutional deep neural network. Sixty five MR images have been used in this study where deep learning techniquet have been adopted for the segmentation of VAT and SAT. The proposed and the manual measurements produced the Dice scores of 0.97 and 0.96 for SAT and VAT respectively. The experimental results show that the deep learning method produces better segmentation results with high accuracy.
Obesity, Visceral adipose tissue (VAT), subcutaneous adipose tissue (SAT), Magnetic resonance imaging (MRI), deep learning
