2024 World Conference on Lung Cancer (WCLC) - Posters-P1.06B.17 Combing Circulating Tumor Cell Associated Liquid Biopsy and Radiomics to Predict the Pathology Grade of LUAD Patients

P1.06B.17 Combing Circulating Tumor Cell Associated Liquid Biopsy and Radiomics to Predict the Pathology Grade of LUAD Patients

Back to course

Pdf Summary

The study discussed in the document focuses on combining circulating tumor cell (CTC)-associated liquid biopsy with radiomics to predict the pathology grade of lung adenocarcinoma (LUAD) patients. Conducted by Hongchang Wang, Yan Gu, Wenhao Zhang, Guang Mu, and Jun Wang, this is a single-center retrospective study undertaken at Jiangsu Province Hospital. The researchers employed habitat radiomics, primary tumor radiomics, and peri-tumor radiomics with varying thicknesses (1/3/5mm) to assess prediction performance.<br /><br />Key methodologies included the use of Borderline-SMOTE for resampling an unbalanced dataset and Bayesian optimization for selecting optimal parameters in machine learning models. The study concluded that the Habitat-CTC combined model achieved the best performance with an Area Under Curve (AUC) of 0.98, indicating high predictive accuracy within the 95% confidence interval of 0.95-1.00. This combination model also demonstrated excellent clinical decision-making efficacy.<br /><br />The document underscores the necessity of early detection of high-grade components in LUAD, as these significantly worsen patient prognosis and influence clinical decisions, especially concerning surgical interventions. The actionable takeaway from this study is the potential of this workflow in predicting other clinical outcomes, emphasizing the integral role of model construction and radiomics feature engineering.<br /><br />For future research, the authors propose a focus on integrating deep learning models with basic machine learning models to enhance prediction capabilities and potentially extend the applications of this approach to other clinical outcomes. Lung cancer's status as the leading cause of cancer incidence and mortality worldwide highlights the importance of developing advanced predictive models to improve patient management and outcomes.

Asset Subtitle

Wang Hongchang

Meta Tag

Speaker Wang Hongchang

Topic Pathology & Biomarkers

Keywords

circulating tumor cells

liquid biopsy

radiomics

lung adenocarcinoma

pathology grade prediction

machine learning

Bayesian optimization

early detection

deep learning integration

cancer prognosis