Lung Cancer Screening
Conditions
Keywords
Lung Cancer Screening, Artificial Intelligence, Low-Dose Computed Tomography, Pulmonary Nodule Detection
Brief summary
The goal of this observational study is to clinically validate the accuracy of an AI-based decision support tool-the Lung Cancer Detection System (LCDS)-for detecting lung nodules in asymptomatic adults aged 50-79 with a history of heavy smoking who underwent low-dose chest CT (LDCT) scans. The main questions it aims to answer are: * Can the LCDS accurately detect the presence of solid pulmonary nodules on LDCT scans, as measured by sensitivity and specificity? * How does the LCDS's performance compare to existing AI systems using the Area Under the Curve-Receiver Operating Characteristic (AUC/ROC) Curve? Researchers will compare the AI-based interpretations to a ground truth established by consensus among radiologists' double-readings to see if the LCDS can accurately classify cases as 'lung nodule presence' or 'lung nodule absence'. Participants will: * Have their de-identified LDCT scans (collected between 2018 and 2023) reviewed retrospectively. * Be evaluated through the LCDS tool, which will classify cases based on lung nodule presence. Contribute to performance evaluation using sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and ROC analysis.
Interventions
DEVICELung Cancer Detection System (LCDS)
An AI-based decision support software designed to detect solid pulmonary nodules on LDCT chest scans. In this study, the LCDS is applied retrospectively to 100 previously acquired LDCT scans, and its performance is compared to a ground truth established by double-read radiologist reports with arbitration.
Sponsors
Genesis Medical AI
Study design
Observational model
OTHER
Time perspective
RETROSPECTIVE
Eligibility
Sex/Gender
ALL
Age
50 Years to 79 Years
Healthy volunteers
No
Inclusion criteria
* Undergone an LDCT scan between 2018 and 2023, while a diagnosis record exists. * Age is between 50-79 years old. * History of smoking at least a 20 pack-year smoking history and currently smoke or have quit within the past 15 years.
Exclusion criteria
* History of lung cancer: Subjects with a previous diagnosis of lung cancer may be excluded to ensure that the study focuses on detecting new cases or evaluating the progression of the disease. * Prior lung nodule detection: Individuals who have previously undergone LDCT scans with documented lung nodules that required medical intervention may be excluded to avoid potential confounding factors in the analysis.
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Sensitivity of LCDS for Detection of Solid Pulmonary Nodules | Through study completion, an average of 1 year | Proportion of true positive cases correctly identified by the AI-based Lung Cancer Detection System (LCDS) out of all subjects with radiologist-confirmed pulmonary nodules (Ground Truth). |
| Specificity of LCDS for Detection of Solid Pulmonary Nodules | Through study completion, an average of 1 year | Proportion of true negative cases correctly identified by the LCDS out of all subjects without pulmonary nodules, as defined by the radiologist consensus ground truth. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Area Under the ROC Curve (AUC) for LCDS Performance | Through study completion, an average of 1 year | The area under the receiver operating characteristic (ROC) curve comparing AI classifications with the radiologist-defined ground truth for nodule detection. |
| False Positive Rate per Case | Through study completion, an average of 1 year | The average number of false positive nodule detections made by the Lung Cancer Detection System (LCDS) per LDCT scan. A false positive is defined as a nodule detected by the AI system that was not confirmed by the radiologist-established ground truth. |
Countries
Israel
Outcome results
None listed