AIDOT (CEO Jung Jae-hoon) announced that its AI-based gastroscopy screening solution, “GIGA dot AI,” co-developed with Professor Bang Chang-seok’s Gastroenterology team at Hallym University Chuncheon Sacred Heart Hospital, has been published in the SCI-level international journal JMIR under the title:
“Clinical decision support system for pre-occurrence stages of gastric cancer in real-time endoscopy: model construction and validation study.”

GIGA dot AI had already been listed earlier this August in the journal “Endoscopy.”

This study aims to build a clinical decision support system (CDSS) for real-time endoscopy covering all stages of gastric cancer.

Previous studies established AI-based CDSS for detecting and classifying gastric cancer, but did not consider precancerous conditions such as atrophic gastritis and intestinal metaplasia, and there was no model necessary to classify all stages of gastric cancer occurrence.

Therefore, in this study, an AI-based CDSS for gastric cancer and its pre-occurrence stages was academically established.

Based on this research, GIGA dot AI—together with Professor Bang Chang-seok’s team at Hallym University Chuncheon Sacred Heart Hospital—performed training and validation using a total of 11,868 endoscopy data cases. The main objectives of the study are twofold: the accuracy of classifying lesions (six classes: advanced gastric cancer, early gastric cancer, dysplasia, atrophic gastritis, intestinal metaplasia, normal) and the boundary segmentation rate for the precancerous stages, atrophic gastritis and intestinal metaplasia.

According to the results mentioned in the paper, lesion classification accuracy on internal validation data was 90.3% or higher, and the sensitivity and specificity for interpreting gastric cancer and precancerous stages achieved very high values of 99.86% and 96.95%, respectively.

In addition, class-specific classification accuracy for the precancerous stages atrophic gastritis and intestinal metaplasia was 96.5% and 90.4%, respectively, and the lesion boundary segmentation rate achieved 93.4% or higher.

In conclusion, this CDSS showed high performance as an AI-assisted diagnostic system for gastric cancer and its pre-occurrence stages.

An AIDOT representative added, “As MFDS Class III approval is expected to be finalized soon, it is expected to be actively adopted in Korea next year after undergoing the New Health Technology Assessment,” and also stated, “Based on these results, we are in close discussions with existing overseas partners, so there will be visible achievements in global expansion as well.”

October 19, 2023 / Korea Economic TV / Senior Reporter Yang Jae-joon