eduzhai > Applied Sciences > Engineering >

Fully Automated and Standardized Segmentation of Adipose Tissue Compartments by Deep Learning in Three-dimensional Whole-body MRI of Epidemiological Cohort Studies

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 30 pages

Abstract: Purpose: To enable fast and reliable assessment of subcutaneous and visceraladipose tissue compartments derived from whole-body MRI. Methods:Quantification and localization of different adipose tissue compartments fromwhole-body MR images is of high interest to examine metabolic conditions. Forcorrect identification and phenotyping of individuals at increased risk formetabolic diseases, a reliable automatic segmentation of adipose tissue intosubcutaneous and visceral adipose tissue is required. In this work we propose a3D convolutional neural network (DCNet) to provide a robust and objectivesegmentation. In this retrospective study, we collected 1000 cases (66$ pm$ 13years; 523 women) from the Tuebingen Family Study and from the German Centerfor Diabetes research (TUEF DZD), as well as 300 cases (53$ pm$ 11 years; 152women) from the German National Cohort (NAKO) database for model training,validation, and testing with a transfer learning between the cohorts. Thesedatasets had variable imaging sequences, imaging contrasts, receiver coilarrangements, scanners and imaging field strengths. The proposed DCNet wascompared against a comparable 3D UNet segmentation in terms of sensitivity,specificity, precision, accuracy, and Dice overlap. Results: Fast (5-7seconds)and reliable adipose tissue segmentation can be obtained with high Dice overlap(0.94), sensitivity (96.6 ), specificity (95.1 ), precision (92.1 ) andaccuracy (98.4 ) from 3D whole-body MR datasets (field of view coverage450x450x2000mm${}^3$). Segmentation masks and adipose tissue profiles areautomatically reported back to the referring physician. Conclusion: Automaticadipose tissue segmentation is feasible in 3D whole-body MR data sets and isgeneralizable to different epidemiological cohort studies with the proposedDCNet.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...