WCLC 2025 - Posters & ePosters-P1.07.11 AI-Based 3D Reconstruction in Lung Surgery

Pdf Summary

This study evaluates the use of AI-assisted 3D reconstruction technology for preoperative planning in minimally invasive lung segmentectomy surgery, specifically for early-stage non-small cell lung cancer (NSCLC). Segmentectomy, a lung-sparing alternative to lobectomy, has been shown to preserve lung function while offering comparable survival but is technically challenging due to high anatomical variability of segmental vessels. Misidentification of segments and vascular structures can lead to complications such as vascular injury, incomplete tumor removal, or misassignment of nodules.<br /><br />Researchers retrospectively analyzed data from 37 patients who underwent video-assisted thoracoscopic surgery (VATS) segmentectomy with preoperative AI-based 3D reconstruction between October 2023 and December 2024. Preoperative tumor morphology, size, and location guided indications for segmentectomy. Key perioperative parameters assessed included operative time, blood loss, R0 resection rates (complete tumor removal), conversion to open surgery, complications (graded by Clavien-Dindo classification), and hospital length of stay. These outcomes were compared to those of 46 patients who had segmentectomy without 3D reconstruction between 2020 and 2023.<br /><br />The AI system used (Infervision GmbH) employs deep learning to automatically detect anatomical structures from chest CT scans and generate 3D reconstructions in approximately 280 seconds. Its anatomical accuracy of 82.8% aligns with that of thoracic surgeons with 2-5 years’ experience. Improved preoperative visualization of anatomical variations and tumor localization via AI-based 3D models shows promise for reducing perioperative complications, improving resection margins, and shortening hospital stays.<br /><br />While initial results suggest AI-assisted 3D reconstruction enhances surgical planning and may improve outcomes in lung segmentectomy, further studies are needed to validate these findings and to better define AI's role in thoracic surgery workflows.

Asset Subtitle

Sayzana Kassahun

Meta Tag

Speaker Sayzana Kassahun

Topic Early-Stage Non-small Cell Lung Cancer

Keywords

AI-assisted 3D reconstruction

lung segmentectomy

minimally invasive surgery

non-small cell lung cancer

video-assisted thoracoscopic surgery

preoperative planning

anatomical variability

segmental vessels

perioperative outcomes

Infervision GmbH