This study evaluates the impact of CADx assistance on endoscopists' histologic characterization of diminutive colorectal polyps (≤5 mm) during colonoscopy. The primary objective is to determine whether CADx assistance increases the proportion of endoscopists who meet PIVI-related performance thresholds, thereby supporting implementation of the "resect and discard" and "diagnose and leave" strategies in routine clinical practice.
This study evaluates the impact of CADx assistance on endoscopists' histologic characterization of diminutive colorectal polyps (≤5 mm) during colonoscopy. The primary objective is to determine whether CADx assistance increases the proportion of endoscopists who meet PIVI-related performance thresholds, thereby supporting implementation of the "resect and discard" and "diagnose and leave" strategies in routine clinical practice. In this randomized controlled trial, endoscopists will be assigned to one of three arms: no CADx assistance, CADx assistance without predicted probability, or CADx assistance with predicted probability. The CADx system provides NICE-based histology predictions (Type 1 vs Type 2), which endoscopists may use to support optical diagnosis and subsequent management decisions, including surveillance interval recommendations when applicable. Outcomes will compare endoscopist-level pass rates and diagnostic performance metrics relevant to PIVI-based adoption, with histopathology as the reference standard where applicable.
Study Type
INTERVENTIONAL
Allocation
RANDOMIZED
Purpose
DIAGNOSTIC
Masking
SINGLE
Enrollment
399
CADx-assisted optical diagnosis (NBI; predicted probability not displayed). Endoscopists perform optical diagnosis of diminutive colorectal polyps (≤5 mm) during colonoscopy using narrow-band imaging (NBI) with CADx-displayed NICE classification predictions for each polyp. In this arm, the CADx output is displayed without any predicted probability information.
CADx-assisted optical diagnosis (NBI; predicted probability displayed). Endoscopists perform optical diagnosis of diminutive colorectal polyps (≤5 mm) during colonoscopy using narrow-band imaging (NBI) with CADx-displayed NICE classification predictions for each polyp. In this arm, the CADx output is displayed with accompanying predicted probability information.
Ruijin Hospital, Shanghai Jiao Tong University School of Medicine
Shanghai, Shanghai Municipality, China
Pass rate for the "resect and discard" strategy
Proportion of endoscopists who achieve ≥90% accuracy in correctly predicting the patient's recommended surveillance colonoscopy interval according to the 2020 U.S. Multi-Society Task Force (USMSTF) consensus recommendations.
Time frame: 14 days
Pass rate for the "diagnose and leave" strategy
Proportion of endoscopists who achieve a negative predictive value (NPV) ≥90% for neoplastic lesions among rectosigmoid polyps, when predictions are made with high diagnostic confidence
Time frame: 14 days
Pass rate for the ESGE 2020-based "resect and discard" strategy
The proportion of endoscopists who achieve at least 90% accuracy in correctly predicting the patient's recommended next surveillance colonoscopy interval, using surveillance recommendations based on the ESGE 2020 guideline.
Time frame: 14 days
Pass rate for the APWG 2022-based "resect and discard" strategy
The proportion of endoscopists who achieve at least 90% accuracy in correctly predicting the patient's recommended next surveillance colonoscopy interval, using surveillance recommendations based on the APWG 2022 consensus.
Time frame: 14 days
Pass rate for the China 2023-based "resect and discard" strategy
The proportion of endoscopists who achieve at least 90% accuracy in correctly predicting the patient's recommended next surveillance colonoscopy interval, using surveillance recommendations based on the Chinese 2023 consensus.
Time frame: 14 days
SODA-1 achievement rate
The proportion of endoscopists who, when reporting high diagnostic confidence, meet the SODA-1 performance thresholds for diagnosing neoplastic pathology in diminutive rectosigmoid polyps, as assessed by sensitivity of at least 90% and specificity of at least 80%, with histopathology as the reference standard.
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.
Time frame: 14 days
SODA-2 achievement rate
The proportion of endoscopists who, when reporting high diagnostic confidence, meet the SODA-2 performance thresholds for diagnosing neoplastic pathology in diminutive colorectal polyps, as assessed by sensitivity of at least 80% and specificity of at least 80%, with histopathology as the reference standard.
Time frame: 14 days
Proportion of high-confidence optical diagnoses
For each endoscopist, the percentage of polyp assessments that are labeled as high diagnostic confidence among all polyp assessments performed.
Time frame: 14 days
Accuracy of high-confidence optical diagnosis (neoplastic vs non-neoplastic)
For each endoscopist, the percentage of correct classifications (neoplastic vs non-neoplastic) among assessments labeled as high diagnostic confidence, using histopathology as the reference standard.
Time frame: 14 days