Quantification of cerebral vascular perfusion density via optical coherence tomography based on locally adaptive...

Optical coherence tomography (OCT) angiography is a noninvasive imaging modality that produces volumetric views of blood flow perfusion in vivo with resolution at capillary level, which has been widely adopted to monitor cerebral perfusion status after stroke in experimental settings. Accurate quantification of cerebral perfusion from OCT angiograms is important for understanding the cerebral vascular pathophysiology and assessing the treatment of ischemic stroke. Quantification of blood vessels from OCT angiography faces some problems; one is uneven backscatter (which causes some blood vessels to be very bright, some very dark), and the other is that the brightness in the same blood vessel also changes due to the difference in diameter or depth. In this paper, we proposed a locally adaptive region growing algorithm to solve this problem. The algorithm, which confines the region growing process to a local region, is used to segment blood vessels in different images to cope well with the intensity changes in blood vessels. During segmentation, the initial seed pixels were selected with the aid of the Otsu algorithm, the growth criterion considered both global and local information, and the thresholds were also adjusted adaptively as local regions varied. After these processes are completed, we can calculate the percentage of segmented blood vessels across field of view of the images, named cerebral vascular perfusion density, and use it as an indicator to evaluate the cerebral blood perfusion of middle cerebral artery occlusion in mice. This paper demonstrates that the algorithm can produce satisfactory vascular segmentation results, and CVPD can be used as an effective indicator for evaluating post-ischemic injury.