AUTHOR=Jia Shanhang , Weng Yuanzhi , Wang Kai , Qi Huan , Yang Yuhua , Ma Chi , Lu Weijia William , Wu Hao 

TITLE=Performance evaluation of an AI-based preoperative planning software application for automatic selection of pedicle screws based on computed tomography images

JOURNAL=Frontiers in Surgery

VOLUME=Volume 10 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/surgery/articles/10.3389/fsurg.2023.1247527

DOI=10.3389/fsurg.2023.1247527

ISSN=2296-875X

ABSTRACT=Introduction: Recent neurosurgical applications based on artificial intelligence (AI) has demonstrated its potential in surgical planning and anatomical measurement. We aimed to evaluate the performance of an AI planning software in screw length/diameter selection and insertion accuracy in comparison with free-hand surgery. 
Methods: 45 patients with 208 pedicle screw placements on thoracolumbar segments were involved in this investigation. The novel AI planning software was developed based on a deep-learning model. AI pedicle screw placement was achieved on preoperative CT data and free-hand surgery screw placements were observed from postoperative CT data. Performance evaluation of AI pedicle screw placements in the aspect of screw length, diameter, and Gertzbein grade in comparison with that of free-hand surgery has been executed.    
Results: Among 208 pedicle screw placements, the average screw length/diameter of the AI model and free-hand surgery were 48.65±5.99 mm/7.39±0.42 mm and 44.78±2.99 mm/6.1±0.27 mm respectively. 85.1% of AI screw placements had a Gertzbein Grade A (no cortical pedicle breach), and 64.9% of free-hand surgery placements had a Gertzbein Grade A.
Conclusion: The novel AI planning software could provide an available and safe pedicle screw placement strategy in comparison with traditional free-hand pedicle screw placement strategy, its choice of pedicle screw dimensional parameters including length and diameter may be a potential inspiration for real clinical discretion.