MRI Based Fat Segmentation at 7T with Confidence Image

Small animal MRI at 7 T provides tool for adiposity research. Separation of fat from surrounding tissues is the key to quantitative analysis in this field. The application of fat segmentation in small animals is a recent extension of the field and it is not known which specific biological questions related to fat deposition will be relevant. Thus it is impossible to predict what accuracy and what spatial resolution will be required in all or even many cases. Herein we describe fat segmentation based on optimizing a variety of factors relevant to small animal imaging at 7 T. In contrast to most previously described MRI methods based on signal intensity of MR T1 weighted image, we chose to use parametric images based on T2 relaxation time and on a pixel by pixel basis initially. The innovation in our method is to separate the fat by the confidence distribution of regions on a scale dictated by practical aspects of MRI at 7 T. In this paper, we describe how we arrived at our recommended procedure (really a collection of procedures) and key aspects of the important post-processing steps in our method. A software tool was created to help fat separation even when the anatomical information makes it difficult to distinguish between fat or nonfat. Further more our software allows the operator to make adjustments to many of the key steps for comparison purposes and to quantitatively assess the difference these changes make.