臨床研究等提出・公開システム

Performance test of a diagnostic support program for evaluating the depth of invasion of early gastric cancer

Performance test of a diagnostic support program for evaluating the depth of invasion of early gastric cancer

204

Site of lesion(U 16 cases M 106 cases L 82 cases.) Classification(Type 0, 200 cases, Types 1-5, 4 cases.) Subtypes(1 10 cases, 2a 51 cases, 2b 6 cases, 2c 133 cases.) Lesion size,mm(10 or less 65 cases, 11-20 59 cases, 21-50 51 cases, 51 or more 12 cases, unknown 17 cases.) Histology (Differentiated type 167 cases, Undifferentiated type 37 cases.) Presence of ulceration (M cancer only) (UL- 126 cases UL+ 10 cases.)Vendor (Olympus 150 cases, Fujifilm 54 cases.) Above mentioned was the only information collected, and no additional background information was obtained.

As planned, data from 204 cases were collected, and the proportion of M cancer was 66.7% (136/204).

none

We calculated the sensitivity of the program for M cancer and its two-sided 95% confidence interval (Wilson method). The estimated value was 91.2% (124/136), and the two-sided 95% confidence interval (Wilson) was 85.2-94.9%, exceeding the target of 75% with a lower limit of 85.2%. The estimated value of specificity was 66.2% (45/68), and the lower limit of the one-sided 95% confidence interval was above 56.3% and 54%, demonstrating non-inferiority with p <0.001. The diagnostic accuracy rate was estimated to be 82.8% (169/204), with a confidence interval of 77.1-87.4%.

We calculated the sensitivity of the program to M cancer. The estimated value was 91.2% (124/136), and the two-sided 95% confidence interval (Wilson) was 85.2% to 94.9%. The lower limit of the confidence interval exceeded the target of 75% and was 85.2%. The estimated value of specificity was 66.2% (45/68), and the lower limit of the one-sided 95% confidence interval was higher than 56.3% and 54%, demonstrating non-inferiority with p<0.001.

April. 30, 2023

No

https://jrct.mhlw.go.jp/latest-detail/jRCT1060220108

Kawahara Yoshiro

Okayama University Hospital

2-5-1, Shikata-cho, Kita-ku, Okayama

yoshirok@md.okayama-u.ac.jp

Hirata Shoichiro

Okayama University Hospital

2-5-1, Shikata-cho, Kita-ku, Okayama

+81-86-235-7219

p6q09uf6@okayama-u.ac.jp

Complete

Jan. 01, 2023

Observational

1. Individuals who are 20 years of age or older at the time of endoscopic image acquisition.
2. Individuals who can provide early gastric cancer tissue suitable for pathological diagnosis by a pathological evaluator.
3. Individuals who can provide images that meet the following criteria for endoscopic images:
Images captured using White Light mode
Vertical resolution of 700 pixels or more and horizontal resolution of 800 pixels or more (image area)
Images captured using an Olympus or Fujifilm endoscope
Five or more images per lesion
Suitable for deep depth diagnosis, as shown below
The entire gastric tumor is generally visible (without using close-up or magnifying endoscopy, etc.)
The details of the tumor's shape can be observed (without moving the lens too far from the tumor)
Sufficient insufflation is performed.

1. The following images are excluded
Images that do not show the gastric tumor (such as normal areas)
Images taken after treatment such as biopsy or clipping, or images with severe bleeding
Images that show foreign objects such as food residue
Blurred or unclear images due to poor focus or other factors
Images that only show a portion of the lesion due to magnifying endoscopy or close-up photography
2. Individuals who have undergone gastric resection or gastric tube reconstruction at the time of endoscopic image acquisition or specimen collection.
3. Individuals with lesions extending to the esophagus or duodenum at the time of endoscopic image acquisition or specimen collection.
4. Individuals with a history of chemotherapy or radiation therapy to the upper abdomen within the year prior to the target endoscopic examination.
5. Patients who have refused to participate in the study.

Both

early gastric cancer

D013274

Sensitivity of mucosal gastric cancer (M cancer) diagnosed by this program.

1) Specificity of mucosal gastric cancer (M cancer) diagnosed by this program.
2) Diagnostic accuracy rate of mucosal gastric cancer (M cancer) diagnosed by this program.

+81-86-235-7133

Dec. 16, 2022

none