Efficient Non-Invasive Registration with A-Mode Ultrasound in Skull Surgery

Fast and accurate registration of cranial CT images with A-mode ultrasound

“…This is an important finding because minimizing the cost function is not generally equivalent to maximizing registration accuracy. The accuracy obtained in this work even after very rough pre-registrations appears similar to the highest accuracy obtained in other works [2,22,24]. However, when absolute values are compared, particular attention must be paid to the various different accuracy metrics including the positions where accuracy is evaluated.…”

“…Overall registration times could be reduced to approximately 2 min. In a previous cadaver study, where basically the same palpation protocol was followed, almost 7 min were needed for palpation [24]. Probably overall timeefficiency could be improved by optimizing the trade-off between a reduction of the computation time (needed by the CT-based alignment tool) and a reduction of the interaction time (needed by the user to digitize points).…”

“…However, during minimally-invasive interventions the bone surface in the regions around the relevant anatomical landmarks might not be exposed, and digitization of those points by means of mechanical devices then increases the invasiveness [8]. Moreover, as precise localization of the anatomical landmarks in CT images and in physical space is difficult, only moderate registration accuracy can be expected [24]. In contrast, surface-based registration does not require artificial markers and avoids the problem of precise manual landmark localization.…”

Robot- and computer-assisted craniotomy: resection planning, implant modelling and robot safety

“…With our user guidance software module, mean registration duration in the phantom study was <5 min, in the cadaver and patient studies <7 min. Further development of the navigation module is focused on enhancement of the existing surface-matching algorithms and improvements of the palpation protocol (23).…”

“…Skin-adhesive markers and anatomical landmarks are non-invasive and practical for use but are of limited accuracy. An alternative to traditional methods is A-mode ultrasound (US) transcutaneous surface-based registration, using an optically tracked probe (22,23). Within the CRANIO project, a study was performed to assess the accuracy of different approaches in neuronavigation and the practicability of those approaches for the specific task of robot-aided surgery (23).…”