An Ad-Hoc Image Segmentation of Subcutaneous and Visceral Adipose Tissue from Abdomino-Pelvic Magnetic Resonance Images

Overweighted people and children with obesity has reached the epidemic ranges in developed and near-development countries. They are a serious public health issue, with big economic impact. Looking for remedies, studies are being carried out to test the impact of several treatments on overweight children health. Such studies require the longitudinal measurement of fat deposits in the abdomino-pelvic regions along the study treatment. Because such studies involve large populations, these measurements requires automated procedures for reliable unbiased estimations of fat volume. This paper describes an ad-hoc image segmentation process developed in the framework of a study on the impact of physical exercise on the visceral and subcutaneous adipose tissue. Several Magnetic Resonance Imaging (MRI) modalities focused in fat visualization have been used in this study, which produce hiperintense values for fat tissues, and are combined by the proposed algorithm. Validation in this paper is qualitative, because we do not have ground truth manual segmentations for quantitative validations.