AI-Powered End-to-End Colonoscopy Automation
ArtInCol is an AI-powered tool that assists endoscopists in detecting epithelial lesions during colonoscopy through а real-time computer vision technology. It captures and stores video and image data from each procedure and is compatible with a range of video systems, including those by Olympus, Pentax, and Fujifilm.
Legendo serves as an endoscopic data archive integrated with ArtInCol, securely storing both video and image records. It enables physicians to review procedure recordings, complete structured reports in a semi-automated format, and access procedure statistics for ongoing insights and quality tracking.
Legendo serves as an endoscopic data archive integrated with ArtInCol, securely storing both video and image records. It enables physicians to review procedure recordings, complete structured reports in a semi-automated format, and access procedure statistics for ongoing insights and quality tracking.
Developed by Alnisoft company in collaboration with the Ryzhikh National Medical Research Center of Coloproctology, this solution brings the expertise of top endoscopy specialists to life. A skilled team of developers and engineers applied advanced AI model architectures for real-time video analysis, creating a powerful tool for endoscopic diagnostics.
How it works
AI Assistant ArtInCol
1
During a procedure, ArtInCol captures the video feed, analyzes it, and highlights potential lesions to assist doctors in spotting them. ArtInCol also enables video and photo recording, with photos taken simultaneously with the press of a button on the endoscope handle, requiring no additional devices for control.
AI Assistant ArtInCol
2
ArtInCol enables live streaming, allowing remote viewing of the procedure within the clinic’s network. The streaming feature supports a video wall mode for full-screen display and offers the option to switch directly to a specific procedure. The system also allows users to pause the video feed and capture images directly from a local computer.
Legendo Archive
3
The procedure registry receives photo and video records from ArtInCol AI assistants, automatically organizing them by procedure. Filters and quick search features make it easy to navigate and locate specific studies. Thumbnail images facilitate case review and enable comparison between different procedures.
Legendo Archive
4
The protocol editor enables quick, consistent documentation of procedures using a library of standardized blocks, including forms and text templates. Developed by leading endoscopists, this template library is customizable and can be expanded as needed.
Legendo Archive
5
The player allows users to review photo and video records from ArtInCol AI assistants. AI-detected findings and physician-captured images are marked along the timeline, making it easy to navigate even lengthy procedures.
Legendo Archive
6
Legendo provides statistics based on structured protocols: physicians can view their performance metrics, while department heads can access key indicators for the entire team. These metrics follow quality assurance standards, allowing users to track trends and assess the impact of organizational changes.
Facts About Colorectal Cancer
Colonoscopy is considered the gold standard for diagnosing diseases of the colon. The primary goal of this procedure is the early detection of colorectal cancer and precancerous conditions. Missed lesions, including small and flat ones, are a leading cause of cancer development despite screening.
- #3Colorectal cancer ranks as the third most common cancer in terms of incidence and second in terms of mortality across the Middle East and Northern Africa (MENA) region.
- 1 in 20Approximately one in 20 people will be diagnosed with colorectal cancer at some point in their lifetime.
- 90%When detected early, colorectal cancer has a 90% chance of being fully cured.
- 17% moreColonoscopy assisted by AI tools detects 17% more polyps and adenomas, reducing the risk of human error.
About
The solution was developed by Alnisoft in close collaboration with the Ryzhikh National Medical Research Center of Coloproctology—Russia’s leading specialized institution focused on the diagnosis and treatment of colorectal diseases. SocIT also contributed to the project.
The foundation for ArtInCol was a 2022 research study conducted at the A.N. Ryzhikh Center, which demonstrated that detecting epithelial lesions in video streams is a relevant and achievable goal for automation through computer vision.
By spring 2023, a prototype capable of real-time lesion detection using neural networks was created, affirming its feasibility and leading to the development of the ArtInCol AI for colonoscopy.
The foundation for ArtInCol was a 2022 research study conducted at the A.N. Ryzhikh Center, which demonstrated that detecting epithelial lesions in video streams is a relevant and achievable goal for automation through computer vision.
By spring 2023, a prototype capable of real-time lesion detection using neural networks was created, affirming its feasibility and leading to the development of the ArtInCol AI for colonoscopy.
In summer 2023, Skolkovo supported a pilot project to introduce ArtInCol AI assistant into clinical practice (Skolkovo is a significant science and technology innovation hub located just outside Moscow, Russia).
The product roadmap includes not only advancing ArtInCol AI assistant but also developing an endoscopic research archive, named Legendo Artis. Legendo includes a study registry for storing photos and videos during procedures, a video player with AI-labeled highlights, a protocol editor, and a dashboard displaying key performance metrics for endoscopy departments based on archived studies. Additionally, a live video streaming system enables doctors to monitor endoscopic procedures in real-time from their workstations.
A multicenter study has been launched to evaluate ArtInCol's impact on epithelial lesion detection, including trials at the Ryzhikh Center, Botkin Hospital’s Endoscopy Center, State Hospital No. 31 named after Academician G.M. Saveleva, and the Ufa Cancer Center.
On August 16, 2024, ArtInCol was registered as a medical device by Roszdravnadzor.
The product roadmap includes not only advancing ArtInCol AI assistant but also developing an endoscopic research archive, named Legendo Artis. Legendo includes a study registry for storing photos and videos during procedures, a video player with AI-labeled highlights, a protocol editor, and a dashboard displaying key performance metrics for endoscopy departments based on archived studies. Additionally, a live video streaming system enables doctors to monitor endoscopic procedures in real-time from their workstations.
A multicenter study has been launched to evaluate ArtInCol's impact on epithelial lesion detection, including trials at the Ryzhikh Center, Botkin Hospital’s Endoscopy Center, State Hospital No. 31 named after Academician G.M. Saveleva, and the Ufa Cancer Center.
On August 16, 2024, ArtInCol was registered as a medical device by Roszdravnadzor.
Our Certifications
ArtInCol is listed in the Unified Register of Russian Software for Computers and Databases, by order of the Ministry of Digital Development, Communications, and Mass Media of Russia, dated August 31, 2020, No. 429: Registry Entry No. 20424 as of December 14, 2023.
Medical Device Registration Certificate: RZN (РЗН) 2024/23409, issued on August 16, 2024.
Experts
Specialists from the Endoscopic Diagnostics and Surgery Department at the Ryzhikh Center, participate in every stage of the solution’s development—from designing core functionalities to testing and evaluating the effectiveness of newly developed features.
- Darya SuslovaProduct ManagerWorking on this project has given us the opportunity to collaborate with many talented physicians. We’ve aimed to integrate their insights into the system’s features, as automating endoscopic procedures is truly a dream project—one that leverages the full potential of modern user interface design to create a user-friendly workspace for endoscopists.
We envision our system as a comprehensive automation tool that ensures high standards in lesion detection while accumulating annotated video materials for research and specialist training. - Nikita PolyakovProduct ArchitectAt the outset, we believed that real-time object detection in video streams was a well-solved problem—widely used in industries like automotive, where it’s employed not only by companies like Tesla but also by various startups worldwide. However, during prototype testing, physicians informed us that any additional delay caused by frame capture, model processing, bounding box overlay, and display was unacceptable, even when reduced to 80ms. Our team overcame this challenge, mastering not only machine learning and code but also existing video equipment and video processing techniques. As a result, the assistant now operates with zero added latency, preserving the quality of images from the endoscopic tower. Additionally, we enabled the system to capture images from Olympus, Pentax, and Fujifilm video systems via a button on the endoscope handle—a favorite feature among physicians.
Contact Information
Email: artincol@alnisoft.ru
Address: Kamenoostrovsky Ave., Bldg. 11, Corp. 2, Lit. A, Kronverkskoye Municipal District, Saint Petersburg, Russia
© All rights reserved 2024