Software for automated MRI-based quantification of abdominal fat and preliminary evaluation in morbidly obese patients

Authors

  • Gregor Thörmer MSc,

    1. Department of Diagnostic and Interventional Radiology, Leipzig University Hospital, Leipzig, Germany
    Search for more papers by this author
  • Henriette Helene Bertram MD,

    1. Department of Diagnostic and Interventional Radiology, Leipzig University Hospital, Leipzig, Germany
    2. Integrated Research and Treatment Center (IFB), Adiposity Diseases, Leipzig University Hospital, Leipzig, Germany
    Search for more papers by this author
  • Nikita Garnov PhD,

    1. Department of Diagnostic and Interventional Radiology, Leipzig University Hospital, Leipzig, Germany
    2. Integrated Research and Treatment Center (IFB), Adiposity Diseases, Leipzig University Hospital, Leipzig, Germany
    Search for more papers by this author
  • Veronika Peter MD cand.,

    1. Integrated Research and Treatment Center (IFB), Adiposity Diseases, Leipzig University Hospital, Leipzig, Germany
    2. Department of Operative Medicine, Clinic of Visceral, Transplantation, Thoracic and Vascular Surgery, Section of Bariatric Surgery, Leipzig University Hospital, Leipzig, Germany
    Search for more papers by this author
  • Tatjana Schütz PhD,

    1. Integrated Research and Treatment Center (IFB), Adiposity Diseases, Leipzig University Hospital, Leipzig, Germany
    2. Department of Operative Medicine, Clinic of Visceral, Transplantation, Thoracic and Vascular Surgery, Section of Bariatric Surgery, Leipzig University Hospital, Leipzig, Germany
    Search for more papers by this author
  • Edward Shang MD,

    1. Integrated Research and Treatment Center (IFB), Adiposity Diseases, Leipzig University Hospital, Leipzig, Germany
    2. Department of Operative Medicine, Clinic of Visceral, Transplantation, Thoracic and Vascular Surgery, Section of Bariatric Surgery, Leipzig University Hospital, Leipzig, Germany
    Search for more papers by this author
  • Matthias Blüher MD,

    1. Integrated Research and Treatment Center (IFB), Adiposity Diseases, Leipzig University Hospital, Leipzig, Germany
    2. Department of Endocrinology and Nephrology, Leipzig University Hospital, Leipzig, Germany
    Search for more papers by this author
  • Thomas Kahn MD,

    1. Department of Diagnostic and Interventional Radiology, Leipzig University Hospital, Leipzig, Germany
    Search for more papers by this author
  • Harald Busse PhD

    Corresponding author
    1. Department of Diagnostic and Interventional Radiology, Leipzig University Hospital, Leipzig, Germany
    • Department of Diagnostic and Interventional Radiology, Leipzig University Hospital, Liebigstrasse 20, 04103 Leipzig, Germany
    Search for more papers by this author

Abstract

Purpose:

To present software for supervised automatic quantification of visceral and subcutaneous adipose tissue (VAT, SAT) and evaluates its performance in terms of reliability, interobserver variation, and processing time, since fully automatic segmentation of fat-fraction magnetic resonance imaging (MRI) is fast but susceptible to anatomical variations and artifacts, particularly for advanced stages of obesity.

Materials and Methods:

Twenty morbidly obese patients (average BMI 44 kg/m2) underwent 1.5-T MRI using a double-echo gradient-echo sequence. Fully automatic analysis (FAA) required no user interaction, while supervised automatic analysis (SAA) involved review and manual correction of the FAA results by two observers. Standard of reference was provided by manual segmentation analysis (MSA).

Results:

Average processing times per patient were 6, 6+4, and 21 minutes for FAA, SAA, and MSA (P < 0.001), respectively. For VAT/SAT assessment, Pearson correlation coefficients, mean (bias), and standard deviations of the differences were R = 0.950, +0.003, and 0.043 between FAA and MSA and R = 0.981, +0.009, and 0.027 between SAA and MSA. Interobserver variation and intraclass correlation were 3.1% and 0.996 for SAA, and 6.6% and 0.986 for MSA, respectively.

Conclusion:

The presented supervised automatic approach provides a reliable option for MRI-based fat quantification in morbidly obese patients and was much faster than manual analysis. J. Magn. Reson. Imaging 2013;37:1144–1150. © 2012 Wiley Periodicals, Inc.

Ancillary