Significance: The practicality of optical methods detecting tissue optical contrast (absorption, elastic and inelastic scattering, fluorescence) for surgical guidance is limited by interferences from blood pooling and the resulting partial or complete inability to interrogate cortex and blood vessels. Aim: A multispectral diffuse reflectance technique was developed for intraoperative brain imaging of hemodynamic activity to automatically discriminate blood vessels, cortex, and bleeding at the brain surface. Approach: A manual segmentation of blood pooling, cortex, and vessels allowed the identification of a frequency range in hemoglobin concentration variations associated with high optical signal in blood vessels and cortex but not in bleeding. Reflectance spectra were then used to automatically segment areas with and without hemodynamic activity as well as to discriminate blood from cortical areas. Results: The frequency range associated with low-frequency hemodynamics and respiratory rate (0.03 to 0.3 Hz) exhibits the largest differences in signal amplitudes for bleeding, blood vessels, and cortex. A segmentation technique based on simulated reflectance spectra initially allowed discrimination of blood (bleeding and vessels) from cortical tissue. Then, a threshold applied to the low-frequency components from deoxyhemoglobin allowed the segmentation of bleeding from vessels. A study on the minimum acquisition time needed to discriminate all three components determined that ∼25 s was necessary to detect changes in the low-frequency range. Other frequency ranges such as heartbeat (1 to 1.7 Hz) can be used to reduce the acquisition time to few seconds but would necessitate optimizing instrumentation to ensure larger signal-to-noise ratios are achieved. Conclusions: A method based on multispectral reflectance signals and low-frequency hemoglobin concentration changes can be used to distinguish bleeding, blood vessels, and cortex. This could be integrated into fiber optic probes to enhance signal specificity by providing users an indication of whether measurements are corrupted by blood pooling, an important confounding factor in biomedical optics applied to surgery.

中文翻译：

---

多光谱漫反射可基于低频血流动力学区分神经外科手术中的血管和出血

意义：用于手术指导的光学方法检测组织光学对比度（吸收、弹性和非弹性散射、荧光）的实用性受到血液汇集的干扰以及由此导致的部分或完全无法询问皮层和血管的能力。目的：开发了一种多光谱漫反射技术，用于血流动力学活动的术中脑成像，以自动区分血管、皮层和脑表面出血。方法：血池、皮层和血管的手动分割允许识别与血管和皮层中的高光信号相关但与出血无关的血红蛋白浓度变化的频率范围。然后使用反射光谱自动分割有和没有血液动力学活动的区域，以及区分血液和皮质区域。结果：与低频血流动力学和呼吸频率（0.03 至 0.3 Hz）相关的频率范围在出血、血管和皮质的信号幅度上表现出最大的差异。基于模拟反射光谱的分割技术最初允许从皮质组织中区分血液（出血和血管）。然后，应用于来自脱氧血红蛋白的低频分量的阈值允许分割血管出血。一项关于区分所有三个分量所需的最小采集时间的研究确定，检测低频范围内的变化需要约 25 秒。其他频率范围，例如心跳（1 到 1. 7 Hz) 可用于将采集时间减少到几秒钟，但需要优化仪器以确保实现更大的信噪比。结论：基于多光谱反射信号和低频血红蛋白浓度变化的方法可用于区分出血、血管和皮质。这可以集成到光纤探头中，通过向用户提供测量是否被血液汇集破坏的指示来增强信号特异性，血液汇集是应用于手术的生物医学光学中的一个重要混杂因素。一种基于多光谱反射信号和低频血红蛋白浓度变化的方法可用于区分出血、血管和皮质。这可以集成到光纤探头中，通过向用户提供测量是否被血液汇集破坏的指示来增强信号特异性，血液汇集是应用于手术的生物医学光学中的一个重要混杂因素。一种基于多光谱反射信号和低频血红蛋白浓度变化的方法可用于区分出血、血管和皮质。这可以集成到光纤探头中，通过向用户提供测量是否被血液汇集破坏的指示来增强信号特异性，血液汇集是应用于手术的生物医学光学中的一个重要混杂因素。