Retrospective Clinical Trial Comparing Radiologists' Diagnosis Accuracy in Lung Cancer Screening Population With and Without the Help of an AI/ML Tech-based SaMD

Median Technologies480 enrolled

Overview

This is a two arm, randomized, controlled, blinded, multi-case multi reader (MRMC), retrospective study for the evaluation of the efficacy and safety of an AI/ML technology-based CADe/x developed to detect, localize and characterize malignancy score of pulmonary nodules on LDCT chest scans taken as part of a lung cancer screening program. LDCT DICOM images of patients who underwent routine lung cancer screening will be selected and enrolled into the study. Enrolled scans analyzed by radiologists with the assistance of the Median LCS (formerly iBiopsy) device are compared to the analysis by radiologists without the assistance of the Median LCS device. Figures of merit for patient level and lesion level detection and diagnostic efficacy will be calculated and compared, sub-class analysis will be performed to ensure device generalizability.

Study Type

OBSERVATIONAL

Enrollment

480

Conditions

High Risk Cancer Lung Cancer

Interventions

Median LCSDEVICE

Eligibility

Sex: ALLMin age: 50 YearsMax age: 80 YearsHealthy volunteers:

Medical Language ↔ Plain English

Inclusion Criteria: * ≥50-80 Years of age; * Current or ex-smoker (\>=20 pack years); * Patient screened and surveilled for lung cancer screening following lung cancer screening guidelines (equivalent to United States Preventive Services Task Force (USPSTF) 2021 Criteria); * Received LDCT due to inclusion in high-risk category for lung cancer. Exclusion Criteria: * Prior lung resection; * Pacemaker or other indwelling metallic medical devices in the thorax that interfere with CT acquisition; * Patients/images used during AI model development; * Patients with only hilar and/or mediastinal cancer(s); * Patients with only ground glass cancer(s); * Patients with nodules, solid or part-solid \>30mm (masses); * Patients that are not accompanied with the required clinical information; * Patients with imaging with any of the following: missing slices, slice thickness \>3mm; * Partial cover of the lung.

Locations (5)

University of Pennsylvania - Penn Center for Innovation

Philadelphia, Pennsylvania, United States

Baptist Clinical Research Institute

Memphis, Tennessee, United States

The University of Texas M.D. Anderson Cancer Center

Houston, Texas, United States

Fundacion instituto de investigacion sanitaria de la fundacion jimenez diaz (FJD)

Madrid, Spain

Universidad de Navarra

Pamplona, Spain

Outcomes

Primary Outcomes

∆ AUC of ROCs > 0. Delta Area between the Response operating curve (AUROC) value with Median LCS and AUROC without Median LCS at patient level data is superior to 0.

Demonstrate that patient diagnosis with Median LCS is improved compared to without Median LCS.

Time frame: 12 months

Secondary Outcomes

Sensitivity at max Youden

Demonstrate that Median LCS aided sensitivity is non inferior (H2) , superior (H8) to radiologist alone. (Sensitivity with Median LCS-Patient) non inferior using non-inferiority margin delta = 0.1 to (Sensitivity Control Arm-Patient). First, non-inferiority. If passed, superiority will be performed.

Time frame: 12 months

Specificity at max Youden

Demonstrate that Median LCS assisted specificity is not inferior (H3), superior (H9) to radiologist alone. (Sensitivity with Median LCS-Patient) non inferior using non-inferiority margin delta = 0.1 to (Sensitivity Control Arm-Patient). First, non-inferiority. If passed, superiority will be performed.

Time frame: 12 months

∆ AUC of LROC > 0

Demonstrate that Median LCS improves clinician's performance in finding detection and diagnosis.

Time frame: 12 months

Recall rates for non-cancer patients (Specificity)

Demonstrate that Median LCS aids to rule out non-cancer patients compared to radiologist alone. "Non-Cancer-Recall-Rate will be calculated and compared between the two modalities using margin of 10%". First, non-inferiority. If passed, superiority will be performed.

Time frame: 12 months

Recall rates for cancer patients (Sensitivity)

Demonstrate that Median LCS aid to diagnose cancer patients compared to radiologist alone. "Cancer-Recall-Rate will be calculated and compared between the two modalities using margin of 10%". First, non-inferiority. If passed, superiority will be performed.

Time frame: 12 months

Time analysis

Demonstrate that Median LCS decreases the time of analysis per patient.

Time frame: 12 months