Abstract #1271

Software for Fully Automatic Quantification of Abdominal Fat with Manual Correction Option: Evaluation in Morbidly Obese Patients

Gregor Thrmer¹, Henriette Helene Bertram¹, Veronika Peter², Tatjana Schtz², Edward Shang², Matthias Blher³, Thomas Kahn¹, Harald Busse¹

¹Dept. of Diagnostic and Interventional Radiology, Leipzig University Hospital, Leipzig, Saxony, Germany; ²Dept. of Visceral, Transplantation, Thoracic and Vascular Surgery, Section Bariatric Surgery, Leipzig University Hospital, Leipzig, Saxony, Germany; ³Dept. of Endocrinology and Nephrology, Leipzig University Hospital, Leipzig, Saxony, Germany

The risk of obesity-related diseases can be assessed by MRI estimates of the abdominal fat distribution. Computer-assisted techniques can be used for rapid image segmentation, but any unsupervised analysis is prone to errors caused by anatomical variations and artifacts, in particular for advanced stages of obesity. A custom-made software with a user interface for the rapid inspection of automatically segmented images was evaluated on the example of 20 extremely obese patients (mean BMI=44 kg/m2) and compared to manual processing by two observers. The presented software appears to be promising for accurate and time-effective fat quantification and larger segmentation errors can be easily corrected in the graphical user environment.