Background Endoscopic skull base and sinus surgery (ESBSS) is one of the most common surgical procedures, with more than 300,000 surgeries performed in the United States yearly, for treatment of neoplasm, cancer, and chronic sinusitis (CRS). Studies have shown that incomplete bone removal in ESBSS is a common cause for persistent CRS symptoms and disease. In patients undergoing revision ESBSS, 70 to 90% of patients have residual bony partitions that could have been removed at the original surgery. However, surgical completeness is difficult to determine intraoperatively due to factors such as altered pathology in surgery and inaccuracy of navigation tracking systems. Although postoperative imaging can be used to determine surgical completeness, they are delayed, costly, and yield extraradiation to patients.
Methods This article presents an algorithm for automatic objective surgical assessment of completeness based on surgical motion data from commercial navigation systems (e.g., Stryker Nav II & III, Medtronic Fusion and S7). The surgical completeness, C, is defined as, where D is the set of desired treatment/removal (green area in the figure), T is the actual volume surgically addressed (red area). The definition of D depends on the surgical indications and goals, and is recognized by deformable coregistering the patient CT with an atlas, which was manually segmented by surgeons. However, the treatment T cannot be directly derived from surgical motion data because (1) instrument tracking errors exist; (2) commercial surgical navigation systems have low sampling frequency and cannot perfectly capture instrument trajectory; (3) the instrument trajectory cannot perfectly reflect the real treatment (e.g., backbiters remove bone which extends past the instrument’s grasp). Estimation algorithms, which use instrument positions as observations, were adopted to jointly estimate instrument position, velocity, and acceleration, to improve trajectory estimation precision and resolution. Three commonly used instruments: backbiters, microdebrider, and navigation probes were studied, and movement velocity, acceleration, and operation duration were used to estimate the actual treatment from the instrument trajectory.
Results Data were collected retrospectively on six pituitary tumor resections with navigation. Attending surgeons evaluated the completeness of the operations based on postoperative CT scans. The surgical completeness estimation results from the proposed method agreed with all surgeons’ assessment.
Conclusion This work proposed and verified that, through processing and learning from surgical motion data collected from commercial surgical navigation systems, surgical completeness can be automatically and objectively evaluated in real time. We anticipate this technique will improve surgical completeness and decrease revision rate in endoscopic skull base and sinus surgeries.
Fig. 1
