Radiomics-based optimization of target selection in CT-guided percutaneous lung cancer biopsy: a retrospective study

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• CT-guided percutaneous transthoracic needle biopsy (PTNB) is crucial for lung cancer diagnosis.
• Tumor heterogeneity and target selection errors can lead to false-negative PTNB results.
• Improving the diagnostic accuracy of PTNB is essential for effective lung cancer management.

Purpose of the Study:

• To develop and validate a clinical-radiomics model for predicting tumor-rich targets in CT-guided PTNB.
• To enhance the diagnostic yield of PTNB by improving target selection.
• To reduce false-negative outcomes in lung cancer diagnosis via precise biopsy planning.

Main Methods:

• Retrospective analysis of 350 lung cancer patients undergoing CT-guided PTNB.
• Extraction of radiomic features from biopsy-slot regions of interest (ROIs).
• Development of a combined clinical-radiomics model using LASSO regression and multivariate logistic regression, validated with ROC and DCA.

Main Results:

• The combined clinical-radiomics model achieved high discriminative performance (AUCs of 0.942 and 0.926 in training and validation sets).
• The model significantly outperformed clinical-only and radiomics-only models.
• An optimal radiomics score cutoff was established, providing high sensitivity and specificity for target prioritization.

Conclusions:

• The developed clinical-radiomics model accurately predicts optimal biopsy targets for CT-guided PTNB in lung cancer.
• This model supports personalized biopsy planning and precision-guided sampling, potentially reducing nondiagnostic procedures.
• Further validation is needed for inoperable patients and broader populations due to the retrospective, single-center nature of the study.