2024 World Conference on Lung Cancer (WCLC) - ePosters-EP.07A.11 AI Models for Intraoperative Diagnosis Based on Surgical Resection Images in Stage IA LUAD: A Prospective, Multicentric Study

EP.07A.11 AI Models for Intraoperative Diagnosis Based on Surgical Resection Images in Stage IA LUAD: A Prospective, Multicentric Study

Back to course

Pdf Summary

The study "AI Models for Intraoperative Diagnosis Based on Surgical Resection Images in Stage IA LUAD (SuRImage)" explores the development of artificial intelligence (AI) models to assist with intraoperative decision-making in lung adenocarcinoma (LUAD) cases. It's a prospective and multicentric trial focused on stage IA LUAD.<br /><br />**Introduction**<br />Intraoperative diagnoses are crucial for deciding surgical strategies, such as segmentectomy or lobectomy for stage IA lung cancer. These strategies are guided by the IASLC grading system, which suggests sublobar resection and lymph node sampling for Grade 1 lepidic predominant LUAD. However, traditional intraoperative frozen sections are often time-consuming and insufficient for accurate IASLC grading. Identifying macroscopic features linked to final pathology poses significant challenges for surgeons and pathologists.<br /><br />**Objective**<br />The research aims to create AI models that use surgical resection images for real-time intraoperative diagnosis and risk stratification of stage IA LUAD.<br /><br />**Results and Discussion**<br />The study developed the SuRImage model, which has shown efficacy in identifying, diagnosing, and classifying invasive adenocarcinoma (IAC). SuRImage has demonstrated a higher concordance with final pathology results compared to traditional frozen section methods. A key feature of the AI model is its ability to focus on macroscopic features essential for analysis, highlighted in attention maps.<br /><br />**Conclusion and Future Directions**<br />The SuRImage model offers novel intraoperative diagnostic capabilities that can significantly impact surgical decision-making. Future research aims to validate the model internationally, enhance feature extraction, and explore multimodal approaches for improved model performance. This research represents a step forward in using AI for real-time decision support in surgery, potentially improving outcomes for patients with stage IA LUAD.

Asset Subtitle

Lintong Yao

Meta Tag

Speaker Lintong Yao

Topic Early-Stage Non-small Cell Lung Cancer

Keywords

AI models

intraoperative diagnosis

lung adenocarcinoma

SuRImage

surgical resection images

stage IA LUAD

invasive adenocarcinoma

macroscopic features

real-time decision support

surgical decision-making