Purpose: Development of a feasible and reliable method for determining abdominal fat using breath-hold T1-weighted magnetic resonance imaging.
Materials and methods: The high image contrast of T1-weighted gradient echo MR sequences makes it possible to differentiate between abdominal fat and non-fat tissue. To obtain a high signal-to-noise ratio, the measurements are usually performed using phased array surface coils. Inhomogeneity of the coil sensitivity leads to inhomogeneity of the image intensities. Therefore, to examine the volume of abdominal fat, an automatic algorithm for intensity correction must be implemented. The analysis of the image histogram results in a threshold to separate fat from other tissue. Automatic segmentation using this threshold results directly in the fat volumes. The separation of intraabdominal and subcutaneous fat is performed by interactive selection in a last step.
Results: The described correction of inhomogeneity allows for the segmentation of the images using a global threshold. The use of semiautomatic interactive volumetry makes the analysis more subjective. The variance of volumetry between observers was 4.6 %. The mean time for image analysis of a T1-weighted investigation lasted less than 6 minutes.
Conclusion: The described method facilitates reliable determination of abdominal fat within a reasonable period of time. Using breath-hold MR sequences, the time of examination is less than 5 minutes per patient.