Impact of artificial intelligence-based automated segmentation on the accuracy of robotic computer-aided implant surgery: a retrospective study

OBJECTIVES: This retrospective study aimed to evaluate the impact of artificial intelligence (AI)-based automated segmentation (AS) on the accuracy of robotic computer-aided implant surgery (r-CAIS). METHODS: Patients who underwent r-CAIS were enrolled. Preoperative CBCT images were segmented using either threshold-based SS approach (SS group) or the AI-based AS approach (AS group). Osteotomy preparation and implant placement were performed using a semi-active robot. Registration errors and deviations between planned and postoperatively placed implant positions were calculated. Intergroup comparisons were performed using Wilcoxon Mann-Whitney U tests, with P < 0.05 considered significant. RESULTS: A total of 140 patients (79 females, 61 males) receiving 224 implants were included (SS group: 69 patients, 107 implants; AS group: 71 patients, 117 implants). The registration error in the AS group (0.10 ± 0.04 mm) was significantly lower than that in the SS group (0.12 ± 0.04 mm) (P < 0.05). The AS group demonstrated lower deviations at the platform (0.51 ± 0.16 vs.0.56 ± 0.18 mm), apex (0.60 ± 0.23 vs. 0.66 ± 0.23 mm), and angulation (1.13 ± 0.59° vs. 1.19 ± 0.53°), with a statistically significant difference observed in global platform deviation (P < 0.05). CONCLUSIONS: r-CAIS demonstrated consistently high accuracy, which was further improved by AI-based AS. This improvement may be associated with enhanced registration precision achieved through AI-based AS. CLINICAL SIGNIFICANCE: These findings support the potential clinical application of AI-based AS in r-CAIS.