The aim of this study is to assess the performance of our CADe/CADx system in detecting and diagnosing Barrett neoplasia during live endoscopic procedures.
ID
Source
Brief title
Condition
- Malignant and unspecified neoplasms gastrointestinal NEC
- Gastrointestinal neoplasms malignant and unspecified
Synonym
Research involving
Sponsors and support
Intervention
Outcome measures
Primary outcome
- Per patient combined CADe/CADx sensitivity (i.e. number of patients with a
visible abnormality detected by CAD on at least one Barrett level, divided by
the total number of neoplasia patients according to the gold standard).
Secondary outcome
- Per patient combined CADe/CADx specificity (i.e. number of patients without a
visible abnormality where CAD does not identify a lesion on any level, divided
by the total number of NDBE patients according to the gold standard).
- Per level combined CADe/CADx sensitivity and specificity (i.e. number of
levels with a visible abnormality identified by CADe/CADx, and number of levels
without a visible abnormality where CADe/CADx does not identify a lesion; both
divided by the total number of neoplasia/NDBE levels according to the gold
standard).
- Per patient and per level CADe sensitivity and specificity
Background summary
Barrett's esophagus (BE) is a known precursor for esophageal adenocarcinoma
(EAC). When detected in an early stage, EAC is often still treatable
endoscopically. Therefore, BE patients undergo regular endoscopic surveillance
by the gastroenterologist. Early Barrett neoplasia however is difficult to
recognize by the general endoscopist due to its subtle appearance.
Our research group has developed a computer aided detection (CADe) system that
can detect and localize early Barrett*s neoplasia using artificial intelligence
(AI). This CADe system assists the endoscopist during the procedure and
provides alerts when an abnormality is detected. We have furthermore integrated
a previously developed computer aided diagnosis (CADx) system that can
characterize detailed imagery and thereby confirm of dismiss any CADe
detections. Both systems have been evaluated extensively in an ex-vivo setting,
showing superior performance when compared to endoscopists. Furthermore, an
earlier clinical pilot study of 30 patients has already shown the feasibility
of real-time utilization of the CADe system in the endoscopy room. The next
step is to evaluate the systems, that are now further developed, clinically
during live endoscopic procedures in a more extensive, prospective, multicenter
cohort.
Study objective
The aim of this study is to assess the performance of our CADe/CADx system in
detecting and diagnosing Barrett neoplasia during live endoscopic procedures.
Study design
Multicenter prospective cohort study
Study burden and risks
During the endoscopic procedure, the computer system essentially monitors the
endoscopic imagery through its own monitor. The imagery is immediately analyzed
by the system for possible abnormalities. The computer system does not
introduce additional risks to the patient and never operates independently.
This means that each part of the Barrett's segment is first evaluated by the
endoscopist for abnormalities. At most, in the case that CAD detects a visible
abnormality and the endoscopist assesses the Barrett's segment as normal, a
target biopsy may be taken, but always in addition to the random biopsies taken
as part of standard care. The target biopsy counts as one of the 4 random
biopsies taken per 2 cm of Barrett's segment (standard protocol). This way, no
extra biopsies are taken compared to regular Barrett's care. Even a
false-positive CAD finding therefore does not pose additional risks to the
patient. The endoscopic examination may take 5-10 minutes longer due to
following the CAD workflow.
de Boelelaan 1117
Amsterdam 1081HV
NL
de Boelelaan 1117
Amsterdam 1081HV
NL
Listed location countries
Age
Inclusion criteria
- BE patients without prior treatment that are undergoing regular endoscopic
surveillance, or an endoscopic treatment in the Barrett segment.
- Age >= 18 years
- Signed informed consent
Exclusion criteria
- Prior endoscopic or surgical treatment of the esophagus for neoplasia
- Reflux esophagitis > grade B (LA classification)
- Inability to undergo EMR/ESD and/or obtain biopsies (e.g. due to
anticoagulation, coagulation disorders, varices)
Design
Recruitment
Medical products/devices used
Followed up by the following (possibly more current) registration
No registrations found.
Other (possibly less up-to-date) registrations in this register
No registrations found.
In other registers
Register | ID |
---|---|
CCMO | NL86341.018.24 |