Occult pleural dissemination (PD) in non-small cell lung cancer (NSCLC) patients is likely to be missed on computed tomography (CT) scans, associated with poor survival, and generally contraindicated for radical surgery. This study aimed to develop and compare the performance of radiomics-based machine learning (ML), deep learning (DL), and fusion models to preoperatively identify occult PDs in NSCLC patients. Patients from three Chinese high-volume medical centers (2016-2023) were retrospectively collected and divided into training, internal test, and external test cohorts. Ten radiomics-based ML models and eight DL models were trained using CT plain scan images at the maximum cross-sectional areas of the primary tumor. Moreover, another two fusion models (prefusion and postfusion) were developed using feature-based and decision-based methods. The receiver operating characteristic curve (ROC) and area under the curve (AUC) were mainly used to compare the predictive performance of the models.

Inclusion Criteria:

• pathologically confirmed primary NSCLC with malignant pleural dissemination;
• no preoperative treatment;
• clinicopathological data were complete.

Exclusion Criteria:

• pleural effusion detected preoperatively;
• preoperatively diagnosed with PD;
• poor CT quality or no CT scans within 1 month before surgery.

From January 2016 to December 2023, NSCLC patients with or without surgically confirmed occult PD from three high-volume centers in China were enrolled.

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.