Sensitivity of Frequency Domain Near Infrared Spectroscopy for Neurovascular Structure Detection in Biotissue Volume: Numerical Modeling Results

Abstract

Through numerical modeling, it has been determined that near infrared spectroscopy with a frequency domain approach can detect neurovascular structures with diameters from 0.5 mm at source‐detector distances of 5–8 mm, depending on optical parameters and technical implementation of the method. Among the five classical machine learning methods considered, quadratic discriminant analysis is the most effective for detection. Furthermore, it has been demonstrated that the use of a photomultiplier tube and the registration of both amplitude and phase signal components exhibit the highest sensitivity. Spectroscopy can rival modern ultrasound for detecting arterial vessels. A cross‐shaped probe configuration improves sensitivity, and the ratio of reduced scattering coefficient values at different wavelengths is informative for blood‐filled vessel detection. These findings are consistent with and significantly extend previous experimental in vivo and in situ studies and could be valuable for intraoperative diagnostic tasks, particularly in neurosurgery.