Quantitative Analysis Of Photoacoustic Blood Vessel Images

The occurrence and growth of many diseases(such as cerebrovascular diseases and tumor)are accompanied by microvascular changes.Optical-resolution photoacoustic microscopy(OR-PAM)has been shown to be an excellent imaging modality for monitoring and study of tumor microvasculature.However,due to the large number of neovascular in tumor tissues,and the photoacoustic signals in small-diameter neovascular and some regions with dense microvasculature distribution are weak,OR-PAM imaging of tumor is prone to unclear vascular veins or vascular rupture,which makes the early diagnose of vascular diseases difficult.Therefore,the use of image processing technology to perform a series of enhancement operations on OR-PAM tumor vascular images can provide high-quality OR-PAM images for subsequent vascular quantitative analysis,and improve the accuracy of quantitative evaluation.Traditional vascular segmentation and skeleton extraction methods are difficult to detect tumor microvascular with weak signals effectively,and previous studies focused mainly on the normal tissues,and there is no tracking quantification of tumor microvessels that continue to grow.Therefore,a set of methods for vascular segmentation,skeleton extraction and tumor microvascular quantification for photoacoustic tumor microvascular images are proposed in this paper.(1)A vascular segmentation method based on Hessian matrix and intensity transformation is proposed.The Hessian matrix can identify the microvascular veins in the photoacoustic image by constructing vascular similarity response function,and the intensity transformation further enhances the contrast of the entire microvascular network.Based on the enhanced image,the seed points of the region growth are automatically selected according to the overall gray distribution,and the segmented vascular binary image is obtained.The experimental results show that the proposed method can segment the complete tumor microvessels with weak signals and several microns in diameter.(2)A skeleton extraction method based on multistencils fast marching method is proposed.In the skeleton iterative extraction process,the multistencils fast marching method(MSFM)is used to calculate the distance field,and the distance information in the horizontal,vertical and diagonal directions can be obtained in all directions,and the obtained skeleton is smoother and more accurate.In addition,when updating the source point set,adding the farthest point can extend the skeleton to the end of the blood vessel,resulting in a more complete vascular skeleton morphology.The experimental results show that the method can extract the skeleton of smallbranches accurately,obtain the complete skeleton and ensure the correctness of the skeleton extraction,and retain the original topology of tumor microvascular network to the greatest extent.(3)In vivo experiments on mouse melanoma and hepatoma were performed to verify the correctness and effectiveness of the proposed method.Quantitative analysis of tumor microvascular morphology by calculating vessel diameter,microvascular density,tortuosity and fractal dimension revealed the growing process of tumor microvessels.The experimental results show that the vascular quantitative parameters obtained in this paper can quantitatively analyze the tumor structure and reveal the vascular changes during tumor growth,and the experimental data are consistent with the tumor growth mechanism reported in the related literatures.It can be seen that the method proposed in this paper has a good reference value for the diagnosis of tumor diseases.At the same time,it has also proved that OR-PAM is a good imaging tool for studying tumor growth,which will further promote the wide application of this technology in tumorigenesis,metastasis and therapeutic research.