Three dimensional volume measurement of mouse abdominal fat in magnetic resonance images

Obesity adversely affects the health of people. Therapeutics against obesity could be validated by measuring the change of body fat tissues repeatedly in magnetic resonance (MR) images. In this study, an accurate method for three dimensional (3D) volume measurement of abdominal fat tissue has been developed for mouse MR images. The MR image was acquired by gradient echo technique and preprocessed by low pass filtering. 3D images were segmented by three-level adaptive thresholding based on the intensity histogram. Small objects were removed by erosion followed by binary reconstruction. Fat tissues were separated by ultimate erosion and individual labeling in 3D, followed by conditional dilation. Abdominal subcutaneous and visceral fat tissues were interactively classified and compared to manually obtained ground truth images. Measurement accuracy was greater than 83.4% for total body fat and 81.9% for abdominal visceral fat, showing the feasibility of routinely measuring the specific component of body fats in mice.