James Shin MD, Presenter: Nothing to Disclose

Matthew A. Barish MD, Abstract Co-Author: Stockholder, Blackford Analysis Ltd

Jung Hwoon Edward Yoon MD, Abstract Co-Author: Nothing to Disclose

Splenic size can be reliably evaluated through imaging. Quantitative methods have been proposed, including the Rezai method (0.36 x (L x W x H)+28) and prolate ellipse method (0.524 x (L x W x H)). This study examines the robustness of these methods, and evaluates the normal variation of splenic volume over time.

34 adults without known splenic pathology and 82 abdominal CT scans were included retrospectively. Image analysis was performed in Alice (Parexel Informatics; Waltham, MA). Manual tracings were performed to provide ground truth splenic volume. Volume estimations were calculated by human observers and automated VOI analysis. True volumes ranged from 73.46-704.9 mL (mean 297.27 ± 185.18 mL; 95% CI 232.65-361.88 mL). Intra-subject variance ranged from 0.001 to 0.367, for a maximum of 3 time-points. There was excellent concordance between observers (r=0.979, 95% CI -7.00-10.42 mm), and between computed and observed maximum axial length (r=0.984, 95% CI -8.64-2.74 mm; r=0.982, 95% CI -9.73-7.23 mm). There was intermediate concordance in computed and observed maximum orthogonal width (r=0.639, 95% CI -20.87-32.84 mm; r=0.691, 95% CI -23.07-29.49 mm). Overall, the Rezai method underestimated volume by 53.95 mL (r=0.892, 95% CI -169.94-62.04 mL), while the prolate ellipse method overestimated volume by 42.82 mL (r=0.879, 95% CI -141.74-227.38 mL). 

Potentially sizable changes in splenic volume may be subclinical or incidental. Estimations were more concordant with true volumes using the Rezai method. While correlation between both methods and true volume was not weak, occasionally changes in morphology significantly altered the maximum length and/or orthogonal width (observed and computed), resulting in falsely large calculated changes in splenic volume.

Quantitative index estimates of splenic volume are limited due to highly variable morphology between subjects and normal variation over time, which is further pronounced in the setting of pathology. Increasingly sophisticated and automated segmentation methodologies can obviate such estimates, making robust ground truth analysis of splenic volume change readily available.

http://abstract.rsna.org/uploads/2014/14005859/14005859_vtkc.jpg

Shin, J, Barish, M, Yoon, J, Variability of Spleen Morphology and Volume Over Time Reveal Limitations of Index Estimates of Splenic Volume.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14005859.html