The size distribution of adipocytes in fat tissue provides important information about metabolic status and overall health of patients. Histological measurements of biopsied adipose tissue can reveal cardiovascular and/or cancer risks, to complement typical prognosis parameters such as body mass index, hypertension or diabetes. Yet, current methods for adipocyte quantification are problematic and insufficient. Methods such as hand-tracing are tedious and time-consuming, ellipse approximation lacks precision, and fully automated methods have not proven reliable. A semi-automated method fills the gap in goal-directed computational algorithms, specifically for high-throughput adipocyte quantification. Here, we design and develop a tool, AdipoCyze, which incorporates a novel semi-automated tracing algorithm, along with benchmark methods, and use breast histological images from the Komen for the Cure Foundation to assess utility. Speed and precision of the new approach are superior to conventional methods and accuracy is comparable, suggesting a viable option to quantify adipocytes, while increasing user flexibility. This platform is the first to provide multiple methods of quantification in a single tool. Widespread laboratory and clinical use of this program may enhance productivity and performance, and yield insight into patient metabolism, which may help evaluate risks for breast cancer progression in patients with comorbidities of obesity.
Abbreviations: BMI: body mass index.
Keywords: MATLAB; algorithm; cancer risk; image analysis; metabolism.