Multimodal AI Predicts High-risk Pathology in Lung Cancer Associated With Cystic Airspaces

N/ARecruitingNCT07820423

Central South University600 enrolled

Overview

The goal of this observational study is to develop and validate an artificial intelligence (AI)-based multimodal radiomics model that integrates preoperative CT imaging features and clinical data to predict pathological high-risk features in patients with lung cancer associated with cystic airspaces (LCCA). The main questions it aims to answer are: Can an AI-based multimodal radiomics model accurately predict pathological high-risk features in LCCA before surgery? Does the integration of CT imaging features and clinical variables improve preoperative risk stratification compared with imaging or clinical information alone?

Lung cancer associated with cystic airspaces (LCCA) is an uncommon radiological presentation of lung cancer characterized by cystic or air-containing spaces with adjacent or surrounding abnormal lung tissue. Although many LCCA lesions are detected at an early clinical stage, their pathological behavior can vary considerably. Some tumors may contain high-risk pathological features associated with greater invasive potential and a less favorable prognosis. Identifying these features before surgery may therefore help improve individualized surgical planning and clinical decision-making. This observational study aims to develop and validate an artificial intelligence (AI)-based multimodal model for the preoperative prediction of pathological high-risk features in patients with LCCA. The model will integrate information derived from preoperative chest computed tomography (CT) images with routinely available clinical and radiological variables. Eligible patients with surgically resected LCCA will be included in the study. Preoperative CT images will be analyzed using AI-based imaging methods to extract tumor-related imaging information. Clinical and radiological characteristics available before surgery will also be collected. These data will be integrated to construct a multimodal prediction model. The reference standard will be the final postoperative pathological examination of the resected tumor. Pathological high-risk features will include predefined histological and invasive characteristics associated with more aggressive tumor behavior. The ability of the multimodal model to predict these pathological features before surgery will be evaluated using measures including the area under the receiver operating characteristic curve, sensitivity, specificity, accuracy, positive predictive value, and negative predictive value. The performance of the multimodal model will also be compared with models based on CT imaging information alone and clinical or tabular information alone to determine whether combining multiple sources of preoperative information provides additional predictive value. Model interpretation methods will be used to identify imaging regions and clinical variables that contribute to individual predictions, thereby improving the transparency and clinical interpretability of the AI model. The study includes model development and independent evaluation. The overall objective is to determine whether multimodal AI can provide a noninvasive and clinically interpretable approach for identifying patients with LCCA who are more likely to harbor pathological high-risk features before surgical treatment.

Eligibility

Sex: ALLMin age: 18 Years

Medical Language ↔ Plain English

Inclusion Criteria: * 1\. Patients with non-small cell lung cancer (NSCLC) confirmed by biopsy or postoperative pathological examination. 2\. Patients who underwent surgical resection of a pulmonary tumor, including lobectomy, segmentectomy, or wedge resection. 3\. Patients with complete preoperative chest CT imaging data. 4. Patients whose preoperative chest CT showed a well-defined air-containing cystic component within the tumor, consistent with the radiological features of lung cancer associated with cystic airspaces (LCCA). 5\. Patients with available clinical and pathological data required for analysis. Exclusion Criteria: * 1\. Patients with a history of pulmonary diseases that may cause cystic lung lesions, such as pulmonary tuberculosis, pulmonary fungal infection, lymphangioleiomyomatosis (LAM), or Birt-Hogg-Dubé (BHD) syndrome. Emphysema will not be considered an exclusion criterion; however, patients with severe emphysema will be excluded if it significantly affects the identification, boundary delineation, or imaging feature assessment of the target lesion. 2\. Patients who received systemic antitumor therapy before enrollment, including chemotherapy, radiotherapy, targeted therapy, or immunotherapy. 3\. Patients with other primary malignancies. 4. Patients with missing preoperative chest CT images or CT images of insufficient quality for analysis. 5\. Patients without a definite pathological diagnosis or with incomplete pathological results. 6\. Patients with missing clinical data.

Outcomes

Primary Outcomes

Predictive performance of the AI-based multimodal radiomics model for pathological high-risk features in LCCA

Pathological high-risk features will be determined based on the final postoperative pathological examination. The predictive performance of the AI-based multimodal model will be quantified using the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, accuracy, positive predictive value, and negative predictive value.

Time frame: Within 30 days after surgery

Secondary Outcomes

Area under the receiver operating characteristic curve of the multimodal, CT imaging, and tabular models for pathological high-risk features in LCCA

The predictive performance of the AI-based multimodal model will be compared with that of the CT imaging model and the tabular model for predicting pathological high-risk features. Model performance will be quantified using the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, accuracy, positive predictive value, and negative predictive value. Differences in AUC between models will be evaluated using DeLong's test.