Save patients around the world

Endoscopy is an advanced medical field in which Japan is leading the world.
In reality, however, more than 20% of lesions are overlooked and endoscopists are exhausted due to the burden of a large amount of double checks.
AI medical service is founded to contribute to the evolution of endoscopy around the world by addressing the problem with AI (artificial intelligence) and eliminate cancers by detecting them at an early stage.


Departure was "voice from endoscopists"

As a clinician, I have performed more than 20,000 endoscopy exams. Every time I have faced the problems, I have considered solutions with the help of doctors around me. (For example, I have systematized the methodology to prevent accidents of perforation and published a "patient-friendly painless colonoscopic insertion method", and presented Cold Polypectomy to reduce bleeding after colon polypectomy to almost zero.) Currently my challenge is to eradicate "missing cancer". Endoscopists work hard every day not to "miss any cancer". However, finding small lesions in the inflamed gastric lining is difficult even for specialists with more than 10 years of experience, and doctors in charge of double check are also exhausted by reading around 3000 images per day. ..In addition, doctors in rural areas with few specialists are really having a hard time in the process of endoscopic examination.

Encouraged by the renowned AI expert

The reason for my attempt was the story of Associate Professor (at the time) Matsuo of the University of Tokyo, saying that "the capability of image recognition by AI has begun to surpass that of humans." I thought "we might be able to solve the pain in the clinical site!". We started to scrutinize a huge number of images one by one, built a database and conducted research using deep learning. Finally AI has overcome the accuracy of average endoscopists. Since improvements are still necessary for daily usage in clinical settings, we are accelerating development at a rapid pace. Not only does Japan's endoscopes account for 70% of the world's share of equipment, but the skills of specialists are also the highest in the world. By gathering that wisdom, we aim to contribute to endoscopy around the world.


Tomohiro Tada CEO


Cutting-edge Records


Treatises recognized one after another

The world's first paper on artificial intelligence for gastric cancer was published in Gastric Cancer, and the paper on AI diagnosis of Helicobacter pylori was published in EBioMedicine (a sister magazine of Lancet and Cell). Furthermore, we have submitted and published many the world's first papers by the topics such as colorectal inflammation and picking up of esophageal cancer.


Multiple patent applications

Not only domestic but global patents have been applied. We pride ourselves on being a world leader in the fierce competition of medical AI development around the world.


"Accuracy" beyond average doctors

Confrontation between AI and Shogi masters is famous. Similarly when our AI also confronted more than 20 endoscopists, we have achieved the higher accuracy than the average of doctors in the judgment of Helicobacter pylori gastritis. AI has proven to be a good support partner in the clinical setting of doctors.


"Speed" applied for realtime endoscopic exams

Real-time assessments are already feasible due to improved hardware performance and the evolution of CNN (convolutional neural network) models. In other words, we believe that AI has advanced to the point where it can support realtime detection of lesions during endoscopy, and will become an indispensable tool in clinical practice around the world.

Foundation of our technology


Make full use of deep learning

CNN in the field of image recognition is being researched all over the world and is evolving day by day. In particular, the level of image recognition by deep learning has surpassed conventional machine learning and even human capabilities. We are taking the latest AI knowledge as much as possible.


The quantity and quality of data are essential

In AI development, the quality and quantity of teacher data (data to be remembered by AI) is the key. Since we are cooperating with leading hospitals nationwide and dozens of endoscopists, we are continuously collecting a huge number of high-quality images, steadily sorting them and finally built the database for AI.


Operational platform developed in-house

In order to increase the development speed, we have developed the operational platform in-house and continue to improve it so that we can work efficiently. For example, we have developed "image anonymization processing software" and "image sorting WEB system" to maximize the work efficiency of cooperating doctors handling of medical data. We will continue to make speedy adjustments and improvements by taking advantage of our in-house development.