gms | German Medical Science

Objective: Recent techniques, such as image-guided- or robot-assisted screw insertion have helped to reduce the rate of misplaced pedicle screws. However, most centers control the resulting screw position by intraoperative 2D fluoroscopy and 3D reconstructions are viewed only on postoperative CTs. While intraoperative 2D imaging is sufficient in most cases, 3D reconstructions allow a more detailed assessment of the screw position. We therefore applied a novel software solution (C-OnSite, Mazor Robotics, Israel), reconstructing 3D images out of standard C-arm fluoroscopies and superimposing them on the preoperative CT, for the intraoperative control of lumbar and thoracic pedicle screws.

Method: We performed C-OnSite scans in 12 consecutive cases (56 screws). Screw position was rated 0–4 (completely within bone, encroachment of the cortical bone, deviation <3 mm, deviation 3–6 mm, deviation>6 mm). Postoperatively, a CT-scan was performed in all cases. The screw position was likewise rated by a neurosurgeon blinded to the C-OnSite findings. The rating of each screw on C-OnSite- and CT-images was compared.

Results: Of the 12 cases, C-OnSite scans succeeded in 11 cases (52 screws), yielding image quality sufficient for the intraoperative decision whether to reposition a screw or not. In these 11 cases rating of screw positions on CT and C-OnSite showed a maximum difference between C-Onsite and CT rating of 1 grade. In 19.2% the C-OnSite was rated 1 grade worse (false positive, 8x 1 instead of 0, 2x 2 instead of 1) and in 5.8% one grade better than the CT scan (false negative, 1x 0 instead of 1, 2x 1 instead of 2), while 75% matched exactly. In one case we revised a grade 2 deviated screw because of the intraoperative C-OnSite findings. The final resulting position was than rated conclusively 0 on C-OnSite and CT scans. The unsuccessful C-OnSite scan was in a revision operation due to secondary screw loosening creating a mismatch between the preoperative CT scan and the intraoperative imaging, preventing C-OnSite’s registration process.

Conclusions: In this pilot study, C-OnSite scanning was successful in 11 out of 12 cases, achieving a same positioning rate in 75% of the 52 imaged screws and within one grade in the other 25%. C-OnSite provides valuable information for assessment of screw position during operation and thus could conceptually replace postoperative CTs and help to prevent reoperations. However, more research is needed, especially in cases of significantly breaching screws (rated 3 or 4).