Subscribe to RSS
DOI: 10.1055/s-0040-1704035
REAL-TIME COMPUTER AIDED DETECTION OF COLORECTAL NEOPLASIA AT COLONOSCOPY ACCORDING TO THE INDICATION: A RANDOMIZED CLINICAL TRIAL
Publication History
Publication Date:
23 April 2020 (online)
Aims One fourth of colorectal neoplasia is missed at colonoscopy, representing the main cause of interval Colorectal-Cancer(CRC). Deep-learning (DL) systems allow Computer-Aided Polyp-Detection (CADe) with high-accuracy in retrospective studies. We performed a multicenter Randomized clinical trial (RCT) of 700 patients who underwent colonoscopy for: gastrointestinal-symptoms (23%), fecal immunochemical test (29.6%), primary screening (22%),postpolypectomy-surveillance (23.4). Aim was to assess safety and efficacy of CADe system in improving lesion-detection.
Methods Consecutive 40-to 80 years old subjects undergoing colonoscopy. Patients were randomized 1:1 between High-Definition CADe-colonoscopy(GI-Genius Medtronic) and control (standard-HD). A minimum Withdrawal Time (WT) of 6 minutes was required. Study end-points were Adenoma Detection Rate (ADR) and Adenomas Per Colonoscopy (APC).
Results We included 685/700 randomized patients in the final analysis (age: 61.32 ± 10.2 y.o.; gender M/F: 337/348). Considering the whole sample, ADR was statistically significantly higher in the CADe-group (194/341, 56.9%) than in the control group (141/344, 40.91%; OR:1.9; 95% CI:1.4 to 2.57; p< 0.001), as well as APC (1.13±1.63 vs 0.73±1.12; OR: 2.1; 95% CI: 1.6 to 2.72; p< 0.001).
We studied the two groups within each indication: symptomatic: ADR(47.36% vs 31.76%; OR:1.9; 95%CI: 1.02 to 3.67; p=0.05), APC(0.76±1.06vs0.52±0.11; OR 1.91; 95% CI:1.1 to 3.3; p=0.04). FIT+: ADR(61.76% vs 44.76%; OR 1.99; 95% CI:1.14 to 3.47; p=0.01), APC(1.62±2.3vs0.9±1.39; OR 2.6; 95%CI 1.6 to 4.17; p=0.008). Primary screening: ADR(44.16% vs 32.89%; OR 1.6; 95%CI: 0.8 to 3.1); APC(0.7±0.11 vs 0.48±0.82; OR 1.7; 95% CI: 1 to 3.1; p=0.15). Surveillance ADR(70.93% vs 55.81%; OR: 1.5; 95%CI: 1 to 2.92; p=0.05); APC(1.29±1.29 vs 0.99±1.4; OR 1.72; 95% CI: 1 to 3.16; p=0.02). No difference in withdrawal time (CADe: 417±101sec vs control: 435±149; p=0.1) was observed.
Conclusions We showed the safety and efficacy of integrating CADe in real-time colonoscopy in terms of substantial improvement of ADR and APC also within each single indication.