Surgeons across multiple centres are reporting significant gains in accuracy and efficiency thanks to a new AI-powered 3D reconstruction system designed to assist in preoperative planning for pulmonary resections.
According to a large-scale, retrospective multi-reader multi-case (MRMC) study published this month, the technology significantly enhanced thoracic surgeons’ ability to identify pulmonary anatomical variants and select appropriate surgical procedures, reducing planning time and boosting confidence.
The study involved 10 thoracic surgeons evaluating 140 real-world cases, both with and without AI assistance.
Results showed an 8% increase in accuracy for anatomical variant identification and a 41% reduction in related errors when the AI-3D system was used.
Procedure selection accuracy improved by 8%, with a 35% reduction in selection errors, particularly minimising incorrect or insufficient resections.
Notably, planning time was reduced by 25%, and surgeon satisfaction reached 99%, highlighting the tool’s usability and clinical value.
The study involved a cohort of 450 patients from three Chinese hospitals, featuring a representative mix of lobectomy and segmentectomy cases.
AI-3D assistance consistently benefited surgeons with 6 to 19 years of experience.
Regarding time efficiency, the AI system’s 3D reconstruction took an average of under four minutes, compared to as much as 30 minutes for manual segmentation, and it saved an additional 63 seconds per case in planning time, particularly in more challenging scenarios.
The AI-3D system was associated with a significant increase in surgeon confidence levels, with ‘confident’ case assessments increasing by more than 30% in anatomical assessments and 17% in procedure selections.
Additionally, interobserver agreement improved, suggesting that the tool contributes to more standardised surgical planning among clinicians.
Despite the potential of 3D reconstruction, its adoption in lung surgery has historically been limited – employed in fewer than 25% of major operations – due to time constraints and a lack of validated tools.
This AI system helps to overcome those limitations by automating complex image segmentation and enhancing accuracy, even in non-contrast CTs. Compared to conventional methods, the system reduces reconstruction time and improves anatomical clarity while minimising errors associated with manual planning.
The system’s architecture guarantees that it serves as an assistive tool rather than a substitute for surgical judgement.
While the study emphasised the accuracy of preoperative planning rather than operative outcomes, the implications are significant. Its level of planning accuracy offers the potential for enhanced surgical safety and efficiency.
The authors advocate for routinely integrating AI-3D tools in lobectomy and segmentectomy, particularly in complex or ambiguous cases.
They also noted that the benefits were consistent across all experience levels, suggesting broad applicability in daily surgical practice.
