June 29, 2023 — According to an accepted manuscript published in the American Journal of Radiology (AJR) of the ARRS, large-scale assessment of splenomegaly can be achieved using automated deep learning AI tools and weight-based volume thresholds. It may become possible. A CT scan done for any symptom.
Noting that historically, standard splenic linear measurements, which were used as a surrogate for spleen volume, had suboptimal performance in detecting volume-based splenomegaly, we noted that “a weight-based volume threshold is , showed the presence of splenomegaly in most patients who underwent CT before liver transplantation,” explained corresponding author Perry J. Pickhart, M.D., Department of Radiology, Department of Public Health, University of Wisconsin School of Medicine.
Pickhardt et al.’s AJR-approved manuscript included 8,901 patients who underwent CT colonoscopy (n = 7736) or renal donor CT (n = 1165) between April 2004 and April 2004. A screening sample (4,235 men, 4,666 women, mean age 56 years) was included. January 2017. A secondary cohort of 104 patients with end-stage liver disease (62 men, 42 women, mean age 56 years) underwent pre-transplant CT from January 2011 to May 2013. The deep learning algorithm of Pickhardt et al., previously developed, trained and tested at the National Institutes of Health Clinical Center, was used for spleen segmentation to determine spleen volume, and two radiologists performed the segmentation. A subset was examined independently.
Ultimately, this automated deep learning AI tool was utilized to calculate spleen volumes from CT examinations of 8,853 primary outpatients. Moreover, spleen volume was most strongly associated with body weight among various patient factors.
The AJR authors concluded, “To our knowledge, this study represents the largest reported sample of patients undergoing splenic volume fractionation.”
For more information, please visit www.arrs.org.
