Automatic Quantitative Detection Algorithm of Abdominal Fat Based on Improved K-Means Algorithm

As many important diseases of human body are closely related to the abdominal obesity, such as diabetes and cardiovascular disease, detecting distribution and quantity of abdominal fat in human body is of great significance to determine the type of abdominal obesity in the medical research and clinical applications. By analyzing the abdominal Magnetic Resonance(MR) images of human body, we propose an unsupervised automatic detection algorithm of abdominal fat. First, the abdominal MR image was preprocessed to be represented by super-pixels using SLIC algorithm; Next, the flood fill algorithm was used to remove background; Then, an improved K-means algorithm was used to realize the automatic segmentation of fat and non-fat zone area, and automatically segment subcutaneous fat and intra-abdominal fat; Finally, the segmentation results were used to achieve the qualitative and quantitative analysis of abdominal fat. Experiments show that the abdominal fat detection algorithm can accurately detect distribution and quantity of abdominal fat. Compared to the previous semi-automatic or fully automatic algorithm, its accuracy has been significantly improved.