A semi-automated method for non-invasive internal organ weight estimation by post-mortem magnetic resonance imaging in fetuses, newborns and children

Eur J Radiol. 2009 Nov;72(2):321-6. doi: 10.1016/j.ejrad.2008.07.013. Epub 2008 Sep 2.

Abstract

Magnetic resonance (MR) imaging allows minimally invasive autopsy, especially when consent is declined for traditional autopsy. Estimation of individual visceral organ weights is an important component of traditional autopsy.

Objective: To examine whether a semi-automated can be used for non-invasive internal organ weight measurement using post-mortem MR imaging in fetuses, newborns and children.

Methods: Phase 1: In vitro scanning of 36 animal organs (heart, liver, kidneys) was performed to check the accuracy of volume reconstruction methodology. Real volumes were measured by water displacement method. Phase 2: Sixty-five whole body post-mortem MR scans were performed in fetuses (n=30), newborns (n=5) and children (n=30) at 1.5T using a 3D TSE T2-weighted sequence. These data were analysed offline using the image processing software Mimics 11.0.

Results: Phase 1: Mean difference (S.D.) between estimated and actual volumes were -0.3 (1.5)ml for kidney, -0.7 (1.3)ml for heart, -1.7 (3.6)ml for liver in animal experiments. Phase 2: In fetuses, newborns and children mean differences between estimated and actual weights (S.D.) were -0.6 (4.9)g for liver, -5.1 (1.2)g for spleen, -0.3 (0.6)g for adrenals, 0.4 (1.6)g for thymus, 0.9 (2.5)g for heart, -0.7 (2.4)g for kidneys and 2.7 (14)g for lungs. Excellent co-correlation was noted for estimated and actual weights (r(2)=0.99, p<0.001). Accuracy was lower when fetuses were less than 20 weeks or less than 300g.

Conclusion: Rapid, accurate and reproducible estimation of solid internal organ weights is feasible using the semi-automated 3D volume reconstruction method.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Algorithms*
  • Female
  • Humans
  • Image Enhancement / methods
  • Image Interpretation, Computer-Assisted / methods*
  • Infant, Newborn
  • Magnetic Resonance Imaging / methods*
  • Male
  • Organ Size*
  • Reproducibility of Results
  • Sensitivity and Specificity