In the Journals

Study finds algorithm useful for decision support about vertebral localization

Results from this study demonstrated LevelCheck, an automatic radiographic labeling algorithm, could be a useful tool to help support decisions about target localization in spine surgery.

“The study shows that LevelCheck can provide useful decision support in situations beyond what we originally anticipated,” Jeffrey H. Siewerdsen, PhD, FAAPM, FAIMBE, professor in the Department of Biomedical Engineering at Johns Hopkins University, told Healio.com/Spine.

Jeffrey H. Siewerdsen

“Our original motivation mainly considered assistance in the mid-thoracic spine, where a lack of unambiguous landmarks can challenge confident target localization. However, the study found utility to be high in several additional scenarios, including counting across long segments (eg, L5 to the lower thoracic), resolving common anatomical variations (such as a thirteenth thoracic vertebra or absence of L5) and radiographic scenes with high implant density (where visibility of underlying anatomy can be a challenge) or generally low image quality (as with obese patients or a lateral radiograph through the shoulders).” He added, “Moreover, the algorithm was unobtrusive in situations where the target was conspicuous, and the independent check provided by the algorithm was helpful, even if simply confirmatory.”

Siewerdsen and colleagues collected 398 intraoperative radiographs and 178 preoperative CT images of patients who underwent spine surgery. Investigators overlaid the preoperative CT images with annotated vertebral labels onto the radiographs by way of 3-D and 2-D registration. Using a questionnaire that evaluated performance, utility and suitability to surgical workflow, three spine surgeons evaluated the radiographs and LevelCheck labels. Each case was measured for geometric accuracy and registration run time.

Findings showed in 42.2% of cases, the LevelCheck was helpful. In 30.6% of cases, the algorithm helped improve confidence in the decision. Performance did not decrease in any of the cases. Investigators noted the LevelCheck shows potential as an independent check and can support decision-making in spine surgery. According to researchers, LevelCheck showed 100% geometric accuracy. by Monica Jaramillo

 

Disclosure: Siewerdsen reports no relevant financial disclosures. 

 

Results from this study demonstrated LevelCheck, an automatic radiographic labeling algorithm, could be a useful tool to help support decisions about target localization in spine surgery.

“The study shows that LevelCheck can provide useful decision support in situations beyond what we originally anticipated,” Jeffrey H. Siewerdsen, PhD, FAAPM, FAIMBE, professor in the Department of Biomedical Engineering at Johns Hopkins University, told Healio.com/Spine.

Jeffrey H. Siewerdsen

“Our original motivation mainly considered assistance in the mid-thoracic spine, where a lack of unambiguous landmarks can challenge confident target localization. However, the study found utility to be high in several additional scenarios, including counting across long segments (eg, L5 to the lower thoracic), resolving common anatomical variations (such as a thirteenth thoracic vertebra or absence of L5) and radiographic scenes with high implant density (where visibility of underlying anatomy can be a challenge) or generally low image quality (as with obese patients or a lateral radiograph through the shoulders).” He added, “Moreover, the algorithm was unobtrusive in situations where the target was conspicuous, and the independent check provided by the algorithm was helpful, even if simply confirmatory.”

Siewerdsen and colleagues collected 398 intraoperative radiographs and 178 preoperative CT images of patients who underwent spine surgery. Investigators overlaid the preoperative CT images with annotated vertebral labels onto the radiographs by way of 3-D and 2-D registration. Using a questionnaire that evaluated performance, utility and suitability to surgical workflow, three spine surgeons evaluated the radiographs and LevelCheck labels. Each case was measured for geometric accuracy and registration run time.

Findings showed in 42.2% of cases, the LevelCheck was helpful. In 30.6% of cases, the algorithm helped improve confidence in the decision. Performance did not decrease in any of the cases. Investigators noted the LevelCheck shows potential as an independent check and can support decision-making in spine surgery. According to researchers, LevelCheck showed 100% geometric accuracy. by Monica Jaramillo

 

Disclosure: Siewerdsen reports no relevant financial disclosures.