Using laser imaging and AI during bronchoscopy to detect lung cancer in biopsy samples

A Multicenter Study in Bronchoscopy Combining Stimulated Raman Histology With Artificial Intelligence for Rapid Lung Cancer Detection

This will test whether a special laser imaging system combined with artificial intelligence can quickly identify cancerous cells in biopsy samples taken during bronchoscopy for people with suspected lung cancer.

Quick facts

• Study type: Observational
• Enrollment: 900 (estimated)
• Ages: 22 Years and up
• Sex: All
• Sponsor: Invenio Imaging Inc. (industry)
• Locations: 7 sites (San Diego, California and 6 other locations)
• Trial ID: NCT07045103 on ClinicalTrials.gov

What this trial studies

ON-SITE is a prospective, observational multicenter effort to train, tune, and validate deep learning algorithms that detect cellular and tissue features suspicious for cancer in biopsies imaged with the NIO Laser Imaging System directly in the procedure room without conventional histology processing. The study collects images from four biopsy settings: transbronchial forceps biopsies, transbronchial needle aspiration of peripheral nodules, EBUS-guided needle aspiration of mediastinal/hilar nodes, and transbronchial cryobiopsies. Collected images are used to develop and validate automated models for rapid on-site interpretation rather than to change clinical management. Participating sites include several academic and regional centers in the United States.

Who should consider this trial

Good fit: Adults aged 22 or older who are scheduled for a clinically indicated bronchoscopy with peripheral lung biopsy or EBUS-TBNA and have an intermediate to high pretest probability of lung cancer are the intended participants, provided they can tolerate the procedure and consent to participation.

Not a fit: People under 22, those not undergoing the specified bronchoscopy/biopsy procedures, those with low pretest probability of lung cancer, prisoners, or patients who cannot tolerate the procedure are unlikely to receive benefit from participation.

Why it matters

Potential benefit: If successful, the approach could give faster, on-site feedback about whether a biopsy contains cancer, potentially shortening the time to diagnosis and helping guide immediate procedural decisions.

How similar studies have performed: Prior pilot and single-center work combining stimulated Raman histology with AI has shown promising results in identifying cancerous tissue, but large multicenter validation in routine bronchoscopy biopsies is still limited.

Eligibility criteria

Inclusion Criteria:

1. The patient or legal guardian is willing and able to understand, sign and date the Ethics committee approved study specific Informed Consent Form.
2. The patient is 22 years of age or older.
3. The patient is scheduled for routinely indicated staging of the mediastinum (EBUS-TBNA) and planned peripheral lung biopsy procedure (peripheral TBBx/TBNA) with an intermediate to high pretest probability of lung cancer based on clinician suspicion.
4. The patient can tolerate the clinical procedure as indicated.

Exclusion Criteria:

1. Patient is a prisoner.
2. The participant, in the judgment of the Investigator, may be inappropriate for the intended study procedures

Where this trial is running

San Diego, California and 6 other locations

• UC San Diego — San Diego, California, United States (RECRUITING)
• Mayo Clinic Jacksonville — Jacksonville, Florida, United States (NOT_YET_RECRUITING)
• Corewell Health — Grand Rapids, Michigan, United States (RECRUITING)
• Memorial Sloan Kettering Cancer Center — New York, New York, United States (SUSPENDED)
• Montefiore Medical Center — The Bronx, New York, United States (RECRUITING)
• UNC Medical Center — Chapel Hill, North Carolina, United States (RECRUITING)
• MD Anderson Cancer Center — Houston, Texas, United States (ACTIVE_NOT_RECRUITING)

Conditions: Lung Biopsy