Validation study of computer-aided diagnostic system for endoscopic image of gastric lesion

Validation study of computer-aided diagnostic system for endoscopic image of gastric lesion - Validation study of Tango

Status

Active, not recruiting

Phases

Unknown

Study type

Observational

Source

JPRN

Registry ID

JPRN-UMIN000044705

Enrollment

316

Registered

2021-06-29

Start date

2021-06-08

Completion date

Unknown

Last updated

2026-06-29

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

gastric lesion

Interventions

None listed

Eligibility

Sex/Gender

All

Inclusion criteria

Inclusion criteria: Patients and images that meet all of inclusion criteria below. (1) Endoscopic images taken with Olympus and Fujifilm endoscopic systems at The Cancer Institute Hospital Of JFCR and Jichi Medical University Hospital, respectively. (2) Endoscopic images taken from January 1st, 2019 to August 31st, 2020. (3) Endoscopic images saved in the on-site endoscopic management systems. (4) Pathological samples, which were taken by biopsy (non-neoplasm) or ESD/EMR (adenoma or adenocarcinoma) or those images are stored. (5) One lesion for one patient when multi-lesions were diagnosed in the patient. The higher priority for treatment will be selected. (6) White light endoscopy (7) Non-chromoendoscopy for the lesion (8) Equal to or greater than 256 x 256px (9) Inadequate images for diagnosis judged by the certified endoscopists (out of use condition, artifacts like devices or foreign bodies, inadequate wash, halation/blurring, out of focus, inable to anonymize such as rare case were excluded ) (10) Regardless of sizes/location of lesion or age/gender of patients

Exclusion criteria

Exclusion criteria: (1) Advanced gastric cancer (2) Consent has been revoked by opt-out (3) Lesions that is too large to fit in rectangular area inscribed in endoscopic images. (4) Smaller than 256 x 256px after anonymization. (5) Pathological diagnosis is "undecidable", which is Group X or Group 2 by the Group Classification or disagreement of pathologists

Design outcomes

Primary

Measure	Time frame
(1) Sensitivity; demonstrate Tango is not inferior to experts	—

Secondary

Measure	Time frame
(1) Accuracy; demonstrate Tango is not inferior to experts (2) Sensitivity/Accuracy; demonstrate Tango is superior to non-experts (3) Specificity/PPV; Tango, experts, and non-experts (4) Response time; Tango, experts, and non-experts (5) Differences in performance; Olympus system and Fujifilm system	—

Countries

Japan

Contacts

Public ContactYusuke Kato

AI Medical Service Inc. R&D

ykato@ai-ms.com03-6903-1028

Outcome results