Advanced solutions for Real-Time intraoperative imaging: ultrasounds perspectives

Minimally invasive surgery is the future of the surgery but now it implies several challenges. Indeed, the surgeon loses visual and haptic feedback inside the patient. In this context, the goal of augmented reality is to compensate for this loss of feedback by enhancing the real view of the surgeon by virtuality. The preoperative information may contain a lot of information and is used by the surgeon for the planning of the intervention. However, during the intervention, the patient anatomy changes, and the preoperative images don't fit anymore the patient. Intraoperative real-time information is then needed. In this chapter, we present the available intraoperative modalities achieving the real-time or quasi real-time constraint, giving their main advantages and limitations for the intraoperative use. This comparison shows how ultrasound results to be the most suited modality and special focus is given on main ultrasound techniques in surgical. Then, a brief theoretical background is given. We demonstrate that real-time segmentation of region of interest in intraoperative ultrasounds images is needed in intraoperative use, both for information enhancement and for registration of preoperative and intraoperative images. Some key approaches for real time segmentation in intraoperative US are then presented. Solutions based on the image processing are summarized, differencing the approaches based on time-efficient image processing of a static image from the one based on the tracking of an initial contour in temporal series. We also describe approaches that use some specific hardware to achieve the real-time constraint. The conclusion gives the perspective of the IFC-CNR Lecce on this research question.