Habitat-Based Radiomics for Predicting Visceral Pleural Invasion in Subpleural Nodules with Solid Component on...

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• Subpleural lung nodules with solid components require accurate assessment for visceral pleural invasion (VPI).
• Current methods for VPI prediction may be invasive or lack sufficient accuracy.
• Noninvasive prediction of VPI is crucial for personalized treatment strategies in lung cancer screening.

Purpose of the Study:

• To develop and validate a habitat imaging model using low-dose computed tomography (LDCT) for noninvasive prediction of VPI.
• To assess the diagnostic performance of the habitat model compared to conventional radiomic and radiological models.
• To provide a publicly available software tool for clinical application.

Main Methods:

• Retrospective enrollment of 313 patients with subpleural lung adenocarcinoma nodules across three centers.
• Development of a habitat model using unsupervised clustering to identify intratumoral heterogeneity from LDCT scans.
• Radiomic feature extraction and selection, followed by comparative analysis using ROC curves against whole-lesion radiomic and radiological models.

Main Results:

• The habitat model demonstrated superior performance in predicting VPI, achieving areas under the ROC curve of 0.893 and 0.908 in validation and test sets.
• The habitat model significantly outperformed both the whole-lesion radiomic (0.833, 0.772) and radiological (0.746, 0.624) models.
• A software tool based on the habitat model was developed and made publicly available.

Conclusions:

• The LDCT-based habitat imaging model effectively predicts VPI in subpleural lung adenocarcinoma by quantifying intratumoral spatial heterogeneity.
• This noninvasive approach shows promising diagnostic performance, surpassing conventional methods.
• The habitat model serves as a valuable preoperative tool for risk stratification and personalized treatment decisions in lung cancer screening.