Automated counting of cerebral penetrating vessels using optical coherence tomography images of a mouse brain in vivo. Issue 8 (8th June 2022)

• Record Type: Journal Article
• Authors: Choi, Woo June; Li, Yuandong; Wang, Ruikang K.; Kim, Jun Ki
• Is Part Of: Medical physics. Volume 49:Issue 8(2022)
• Page Start: 5225
• Page End: 5235
• Publication Date: 2022-06-08
• Subjects: 14.11 Segmentation -- 14.3 Optical computed tomography -- 17.1 Image processing -- 17.7 Classification methods -- 17.8 Image segmentation -- automated vessel counting -- cerebral penetrating vessels -- Doppler frequency variance -- ischemic stroke -- light attenuation -- optical coherence tomography; Medical physics -- Periodicals; Medical physics; Geneeskunde; Natuurkunde; Toepassingen; Biophysics; Periodicals; Periodicals; Electronic journals; 610.153
• DOI: 10.1002/mp.15775
• Languages: English
• ISSNs: 0094-2405

Abstract

Abstract: Rationale and objectives: Penetrating blood vessels emanating from cortical surface vasculature and lying deep in the cortex are essential vascular conduits for the shuttling of blood from superficial pial vessels to the capillary beds in parenchyma for the nourishment of neuronal brain tissues. Locating and counting the penetrating vessels is beneficial for the quantification of a course of ischemia in blood occlusive events such as stroke. This paper seeks to demonstrate and validate a method for automated penetrating vessel counting that uses optical coherence tomography (OCT). Materials and methods: This paper proposes an OCT method that effectively identifies and grades the cortical penetrating vessels in perfusion. The key to the proposed method is the harnessing of vascular features found in the penetrating vessels, which are distinctive from those of other vessels. In particular, with an increase in the light attenuation and flow turbulence, the contrast in the mean projection of the OCT datacube decreases, whereas that in the maximum projection of the Doppler frequency variance datacube increases. By multiplying the inversion of the former with the latter, its binary thresholding is sufficient to highlight the penetrating vessels and allows for their counting over the projection image. Results: A computational method that leverages the decrease in mean OCT projection intensity and the increase in Doppler frequency variance at the penetrating vessel is developed. It successfully identifies and counts penetrating vessels with a high accuracy of over 87%. The penetrating vessel density is observed to be significantly reduced in the mouse model of focal ischemic stroke. Conclusion: The OCT analysis is effective for counting penetrating blood vessels in mice brains and may be applied to the rapid diagnosis and treatment of stroke in stroke models of small animals.