|
Predicting small cell lung cancer stage by training extreme gradient boosting models based on radiomics features extracted from computed tomography data
|
T. Wu   , Y. Bao , X. Cao , L. Zhao , L. Yin , K. Jin , Y. Wang , F. Yang , W. Shao |
| Department of Radiation Physics, Harbin Medical University Cancer Hospital, Harbin, China , wenchengshao@outlook.com |
|
|
Abstract: (12 Views) |
Background: This study aimed to enhance small cell lung cancer (SCLC) staging accuracy by developing extreme gradient boosting (XGBoost)-based models leveraging radiomic features derived from computed tomography (CT) data. The aim was to provide a non-invasive tool that enhances staging precision through advanced machine learning and quantitative imaging biomarkers. Materials and Methods: A retrospective cohort of 219 patients with SCLC spanning various tumor stages determined by clinical evaluation was analyzed. High-resolution CT scans were used to extract 107 three-dimensional radiomic features using PyRadiomics. The cohort was split into training (80%) and testing (20%) subsets. XGBoost models were trained for classification tasks with hyperparameters optimized using a grid search. Model performance was assessed using accuracy, area under the receiver operating characteristic curve (AUC), recall, precision, and F1-scores. Results: The XGBoost model exhibited strong performance in tumor staging, achieving an accuracy of 0.90, precision of 0.84, AUC of 0.92, and F1-score of 0.84, with a recall value of 0.84, distinguishing early (I–II) from advanced (III–IV) stages. The deviations in accuracy, F1-score, and AUC were less than 0.05 when five-fold cross-validation was applied, indicating the robust predictive performance of the XGBoost model across diverse patient characteristics. Conclusions: XGBoost models trained on CT-derived radiomic features effectively predicted SCLC stages, offering clinicians a valuable, noninvasive tool. This approach enhances the diagnostic accuracy and supports treatment decisions, potentially improving patient outcomes in lung cancer management. |
|
| Keywords: Small cell lung cancer, neoplasm staging, radiomics, extreme gradient boosting, tomography, X-Ray computed, machine learning. |
|
|
Full-Text [PDF 910 kb]
(4 Downloads)
|
|
Type of Study: Original Research |
Subject:
Radiation Biology
|
|
|
|
|
|
|
| Send email to the article author |
|
|
|
| Add your comments about this article |
|
|
|
