Investigation of acoustic windows for photoacoustic imaging of intracranial blood vessels

Real-time intraoperative guidance during the en-donasal transsphenoidal approach to minimally invasive neurosurgery is often limited to endoscopy, which is suboptimal at locating underlying blood vessels to mitigate the risks and severe complications associated with accidental injury. Transcranial photoacoustic imaging is a promising technique for realtime visualization of these structures, but it is challenged by acoustic-bone interactions which degrade image quality. We are developing simulation methods that identify viable transcranial acoustic windows for intraoperative photoacoustic visualization of these underlying structures. In our previous work on this topic, simulation models were limited to an empty cadaver skull and the eyes of an intact cadaver head, while experiments with the same intact cadaver head were limited to the ocular acoustic window. In this paper, we present our advances to the simulation methodology, including quantitative assessments of images generated from simulated data and in silico investigations of additional acoustic windows with the intact cadaver head model. In addition, experimental images obtained with the ocular acoustic window of the cadaver head were compared to corresponding simulation results. With experimental and simulation similarity confirmed (i.e., point spread function area difference of 5.68 mm2), simuations were extended to explore the temple and sphenoid sinus acoustic windows. This exploration indicated that the sphenoid sinus and ocular acoustic windows provide images with the best target visibility (e.g., a generalized contrast to-noise ratio of $1.00 \pm 0.00$) and resolution. Considering that current transducer technology has limited ability to navigate to and within the sphenoid sinus, we conclude that the ocular cavity is the most feasible transducer location at this time.