Whole-slide image analysis outperforms micrograph acquisition for adipocyte size quantification. (1st January 2020)

• Record Type: Journal Article
• Authors: Maguire, Anne S; Woodie, Lauren N; Judd, Robert L; Martin, Douglas R; Greene, Michael W; Graff, Emily C
• Is Part Of: Adipocyte. Volume 9:Number 1(2020)
• Page Start: 567
• Page End: 575
• Publication Date: 2020-01-01
• Subjects: Adipose tissue -- obesity -- diabetes -- hypertrophy -- hyperplasia; Fat cells -- Periodicals; Adipocytes -- Periodicals; 611.018276
• DOI: 10.1080/21623945.2020.1823139
• Languages: English
• ISSNs: 2162-3945

Abstract

ABSTRACT: The distinction between biological processes of adipose tissue expansion is crucial to understanding metabolic derangements, but a robust method for quantifying adipocyte size has yet to be standardized. Here, we compared three methods for histological analysis in situ : one conventional approach using individual micrographs acquired by digital camera, and two with whole-slide image analysis pipelines involving proprietary (Visiopharm) and open-source software (QuPath with a novel ImageJ plugin). We found that micrograph analysis identified 10–40 times fewer adipocytes than whole-slide methods, and this small sample size resulted in high variances that could lead to statistical errors. The agreement of the micrograph method to measure adipocyte area with each of the two whole-slide methods was substantially less (R 2 of 0.6644 and 0.7125) than between the two whole-slide methods (R 2 of 0.9402). These inconsistencies were more pronounced in samples from high-fat diet fed mice. While the use of proprietary software resulted in the highest adipocyte count, the lower cost, ease of use, and minimal variances of the open-source software provided a distinct advantage for measuring the number and size of adipocytes. In conclusion, we recommend whole-slide image analysis methods to consistently measure adipocyte area and avoid unintentional errors due to small sample sizes.