Sadeer Al-Kindi

MD

Houston Methodist Research Institute

Associate Professor

NLST (Learn more about this study)

NLST-1359

Nov 21, 2024

Chest CT Radiomics and Outcomes

This project aims to evaluate the diagnostic and prognostic potential of CT radiomics using data from the National Lung Screening Trial (NLST). Radiomics, which extracts high-dimensional quantitative imaging features from CT scans, offers a novel approach to enhance the assessment of lung cancer and cardiopulmonary diseases. By leveraging the extensive NLST dataset of low-dose CT scans and clinical outcomes, this study seeks to identify radiomic biomarkers that improve diagnostic accuracy for lung cancer and provide prognostic insights into survival, disease recurrence, and cardiopulmonary risk.

Radiomic features—such as texture, shape, and intensity metrics—will be extracted from CT scans using standardized methods and analyzed with machine learning models. These models will integrate radiomic data with clinical and demographic factors to stratify patients by disease risk and predict long-term outcomes. Validation will occur using independent NLST subsets, with performance compared to traditional clinical predictors. This work could advance precision medicine by demonstrating the value of radiomics in optimizing lung cancer screening, risk stratification, and management, potentially transforming clinical decision-making and public health strategies.

This project utilizes advanced segmentation models to analyze low-dose CT scans from the National Lung Screening Trial (NLST), extracting features from structures such as the liver, bone, epicardial fat, and cardiac chambers. We aim to examine the relationships between these features and key clinical outcomes, including mortality, cancer diagnosis, cause-specific death, and other systemic conditions.

Segment CT Features: Apply automated models to extract detailed anatomical and radiomic features from NLST CT scans, providing a comprehensive assessment of organ and tissue characteristics.
Link Features to Mortality and Cancer: Investigate the association of segmented features with all-cause mortality and cancer diagnosis, hypothesizing that specific patterns predict risk.
Examine Cause-Specific Outcomes: Assess the relationship between segmented features and cause-specific mortality, such as cardiovascular or cancer-related deaths, and broader clinical outcomes.
This work will identify novel imaging biomarkers that enhance risk prediction and improve clinical utility in low-dose CT screening programs.