Optical-resolution photoacoustic microscopy (OR-PAM) has been shown to be an excellent tool for high-resolution imaging of microvasculature, and quantitative analysis of the microvasculature can provide valuable information for the early diagnosis and treatment of various vascular-related diseases. In order to address the characteristics of weak signals, discontinuity and small diameters in photoacoustic microvascular images, we propose a method adaptive to the microvascular segmentation in photoacoustic images, including Hessian matrix enhancement and the morphological connection operators. The accuracy of our vascular segmentation method is quantitatively evaluated by the multiple criteria. To obtain more precise and continuous microvascular skeletons, an improved skeleton extraction framework based on the multistencil fast marching (MSFM) method is developed. We carried out in vivo OR-PAM microvascular imaging in mouse ears and subcutaneous hepatoma tumor model to verify the correctness and superiority of our proposed method. Compared with the previous methods, our proposed method can extract the microvascular network more completely, continuously and accurately, and provide an effective solution for the quantitative analysis of photoacoustic microvascular images with many small branches.

中文翻译：

---

一种光学分辨率光声显微镜的微血管图像分析方法

光学分辨率光声显微镜（OR-PAM）已被证明是微血管高分辨率成像的绝佳工具，对微血管的定量分析可以为各种血管相关疾病的早期诊断和治疗提供有价值的信息。针对光声微血管图像信号弱、不连续、直径小等特点，提出了一种适应光声图像微血管分割的方法，包括Hessian矩阵增强和形态连接算子。我们的血管分割方法的准确性通过多个标准进行定量评估。为了获得更精确和连续的微血管骨架，开发了一种基于多模板快速行进（MSFM）方法的改进骨架提取框架。我们在小鼠耳朵和皮下肝癌模型中进行了体内OR-PAM微血管成像，以验证我们提出的方法的正确性和优越性。与以往的方法相比，我们提出的方法可以更完整、连续、准确地提取微血管网络，为具有许多小分支的光声微血管图像的定量分析提供有效的解决方案。