CLEAREST: Clinical evaluation of lung cancer detection and diagnosis software

BACKGROUND: Detecting lung cancer early remains the key determinant factor in lung cancer survival, and occurs most commonly opportunistically through incidental discovery on CT imaging.

However, well documented workforce challenges across the NHS create a bottleneck for timely review of incidental findings such as unsuspected lung nodules, which are present in approximately 30% of CT scans, and most of which are benign.

Improved risk stratification of pulmonary nodules would help prioritize onward referral of suspicious cases from radiology to pulmonology and their follow-up, potentially reducing the time-to-diagnosis of lung cancer, and unnecessary procedures where a malignancy can be reliably ruled-out non-invasively.

Zero-click Lung Cancer Prediction (zcLCP) is an AI-based computer-assisted detection and diagnosis (CADe/CADx) device developed by Optellum, which integrates into standard radiology workflows to assist imaging experts in the risk stratification and reporting of lung nodules detected on CT scans. zcLCP operates by processing CT scans before they are reviewed by the radiologist to detect, measure, and assess the risk that a lung nodule found on the CT scan is malignant.

When reading the case, imaging experts can consult the output of zcLCP to help inform their follow-up recommendations. zcLCP is in an advanced stage of development, pending clinical evaluation and regulatory approval.

AIM: To produce real-world evidence of the effectiveness of zcLCP in aiding medical imaging experts when estimating the risk of lung cancer for lung nodules detected on CT scans, reporting them, and issuing appropriate follow-up recommendations. This evidence will serve as the basis for the regulatory approval of zcLCP.

METHOD: The central element is the execution of a Multi-Reader Multi-Case (MRMC) reader study to determine whether zcLCP improves the lung cancer risk estimation of medical imaging experts (e.g. radiologists) when reading CT scans with lung nodules.

The study will also measure the effect of aided lung cancer risk stratification on the issuing of follow-up recommendations, and the effect of the automated lung nodule detection and measuring on the reporting of lung nodules from CT scans.

The results will be used in at least three different ways: For a Clinical Evaluation Report to be submitted as part of the documentation for UKCA/CE marking.

To Assess the potential to streamline the pathway, standardising follow-up recommendations regardless of observer experience, and increasing efficiency of the available workforce.

Disseminated to the clinical community through conference presentations and journal publications TIMELINE: Between July 2024 and December 2025.

ANTICIPATED IMPACT: The project will enable the timely regulatory certification of zcLCP, making it available for clinical use and for the gathering of real-world evidence of its benefits.

The anticipated benefits of this technology include: For patients: the improved prioritization of referrals based on lung cancer risk stratification leading to a reduced time-to-diagnosis of lung cancer, and hence the stage at which these are acted on.

For the NHS: net savings by decreasing the cost of lung cancer treatment for those cases that are found at early stages, while ruling-out unnecessary procedures for lung nodules that do not need to be followed-up.