Error Evaluation for Automated Diameter Measurements of Cerebral Capillaries Captured with Two-Photon Laser Scanning Fluorescence Microscopy

Abstract

Cerebral capillaries respond to changes in neural activity to maintain regional balances between energy demand and supply. However, the quantitative aspects of the capillary diameter responses and their contribution to oxygen supply to tissue remain incompletely understood. The purpose of the present study is to check if the diameters measured from large-scale angiographic image data of two-photon laser scanning fluorescent microscopy (2PLSM) are correctly determined with a custom-written MATLAB software and to investigate how the measurement errors can be reduced, such as at the junction areas of capillaries. As a result, nearly 17% of the measured locations appeared to be outliers of the automated diameter measurements, in particular arising from the junction areas where three capillary segments merged. We observed that about two-thirds of the outliers originated from the measured locations within 6 μm from the branching point. The results indicate that the capillary locations in the junction areas cause non-negligible errors in the automated diameter measurements. Considering the common site of the outliers, the present study identified that the areas within 6 μm from the branch point could be separately measured from the diameter analysis, and careful manual inspection with reference to the original images for these transition areas around the branch point is further recommended.