Colorectal Dysplasia, Colorectal Neoplasms
Conditions
Keywords
Artificial Intelligence, Colonoscopy, Endoscopic mucosal resection, Computer-assisted diagnosis
Brief summary
In the last decade, many innovative systems have been developed to support and improve the diagnosis accuracy during endoscopic studies. CAD-Eye™ (Fujifilm, Tokyo, Japan) is a computer-assisted diagnostic (CADx) system that uses artificial intelligence for the detection and characterization of polyps during colonoscopy. However, the accuracy of CAD-Eye™ in the recognition of remaining lesions after endoscopic mucosal resection (EMR) has not been broadly evaluated. Finally, based on the importance of complete resection of the colonic mucosal lesions, namely suspicious high-grade dysplasia or early invasive cancer, the investigators aimed to assess the accuracy of CAD-Eye™ in the detection of remaining lesions after the procedure.
Detailed description
Nowadays, the increased polyp and adenoma detection rate, and its early treatment have reduced considerably colorectal cancer-related mortality. For lesions suspicious of high-grade dysplasia or early invasive cancer, the endoscopic mucosal resection (EMR), along with snare polypectomy, is now considered one of the established standard treatments. However, there are many ´difficult-to-treat lesions´ such as the large and fibrotic ones, which can lead to incomplete resections. Based on the above, many newly diagnostic techniques guided by artificial intelligence (AI), currently proposed to improve the polyp detection rate during colonoscopy, can be applied for the detection of remaining lesions after endoscopic treatment. CAD-Eye™ is CADx for polyp detection and characterization. It improves polyp visualization by using techniques such as blue-laser imaging (BLI-LASER), blue-light imaging (BLI-LED), and linked-color imaging (LCI). This device aimed to improve real-time polyp detection, helping experts identify multiple polyps simultaneously and common inadvertently missed lesions (flat lesions, polyps in difficult areas). CAD-Eye™ had demonstrated in previous studies an accuracy of 89% to 91.7% in polyp detection. However, few studies had demonstrated its performance in the detection of remaining lesions after EMR. The investigators aimed to take advantage of this system in the detection of remaining lesions immediately after EMR and in its endoscopic control after three months.
Interventions
DIAGNOSTIC_TESTEMR with CAD-Eye™
Patients of group 1 undergoing Intervention 1 are subjected to an EMR with CAD-Eye™ to detect the remaining lesions immediately after the endoscopic procedure. The suspected remaining lesions in the post-procedure defect detected with CAD-Eye™ are removed and sent to pathology to confirm the diagnosis.
DIAGNOSTIC_TESTEMR without CAD-Eye™
Patients of group 2, undergoing intervention 2, subjected to an EMR alone. The immediate detection of remaining lesions is based on the visual impression of the expert. The suspected remaining lesions in the post-procedure defect are removed and sent to pathology to confirm the diagnosis.
DIAGNOSTIC_TESTFollow-up colonoscopy with CAD-Eye™
Patients undergoing Interventions 1 and 2, with a previous EMR, are assigned for a three-month follow-up using the CAD-Eye™ as a complementary procedure to detect remaining lesions. For the detection of residual lesions, the colonoscope with the CAD-Eye™ assistance is used during the post-procedural scar evaluation. Suspicious lesions detected are removed and sent to pathology for final diagnosis.
Sponsors
Instituto Ecuatoriano de Enfermedades Digestivas
Study design
Allocation
NON_RANDOMIZED
Intervention model
PARALLEL
Primary purpose
DIAGNOSTIC
Masking
NONE
Intervention model description
Non-blinded, single center, non-randomized prospective pilot study
Eligibility
Sex/Gender
ALL
Age
18 Years to 90 Years
Healthy volunteers
No
Inclusion criteria
* Patients referred to our center with an indication of colonoscopy and EMR for the treatment of lesions suspicious of high-grade dysplasia and early invasive cancer. * Patients who authorize EMR and colonoscopy. * Signed informed consent
Exclusion criteria
* Any clinical condition which makes EMR inviable. * Poor bowel preparation score defined as the total Boston bowel preparation score (BBPS) \<6 and the right-segment score \<2 * Patients with more than one previous EMR * Lost on a three-month follow-up after EMR * Pregnancy or nursing
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Lesions recurrence after EMR | up to 1 week | Detection of remaining lesions immediately after EMR procedure based on endoscopist expertise (EMR without CAD-Eye™ group) or CAD-Eye™ (EMR + CAD-Eye™ group). Lesions will be confirmed by biopsy. Data will be summarized as frequencies. |
| Lesions recurrence in a three-month follow-up after EMR | up to 3 months | Evaluation of CAD-Eye™ in the detection of recurrent lesions after EMR procedure. Remaining lesions detected by CAD-Eye™ in the three-month follow-up. Lesions will be confirmed by biopsy. Data will be summarized as frequencies. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Recurrence risk after EMR | up to 1 week | Calculate de recurrence risk by the Sydney EMR recurrence tool (SERT) in a scale from 0 to 4 * 2 points: size of 40 mm or larger * 1 point: Intraprocedural bleeding (IPB) * 1 point: high-grade dysplasia (HGD) in histopathology |
Countries
Ecuador
Contacts
Primary ContactCarlos Robles-Medranda, MD FASGE
Outcome results
None listed