Olympus' AI-Powered Caddie System Receives FDA 510(k) Clearance, Boosts Large Adenoma/Carcinoma Detection Rate by 93%

On September 5, Olympus announced that its Odin Medical division’s Caddie computer-aided detection (CADe) device had received FDA 510(k) clearance.

This cloud-based AI technology can help gastroenterologists detect suspected colorectal polyps during colonoscopy. Olympus stated that this is the first cloud-based AI technology used for this purpose.

Colonoscopy is considered the gold standard for colorectal cancer screening. Olympus stated that one of the primary objectives behind launching Caddie is to enhance the effectiveness of colonoscopy and reduce the societal burden caused by colorectal cancer. Its intelligent endoscopy ecosystem, leveraging Caddie, can improve the accuracy of clinical decision-making.

Caddie is a cloud-based artificial intelligence medical device software that interfaces with the video stream generated by endoscopic video processors during colonoscopy procedures. The software is intended for use by trained and qualified healthcare professionals. As an aid in video endoscopy, Caddie is designed to draw physicians’ attention to areas exhibiting visual characteristics of colonic mucosal lesions, such as polyps and adenomas.

Caddie analyzes data from the endoscopic video processor in real time and provides information to assist endoscopists in detecting suspected colorectal polyps within the field of view. However, it should be noted that areas highlighted by Caddie should not be interpreted as “definite polyps or adenomas.” The responsibility for determining whether a highlighted area contains a polyp or adenoma rests with the endoscopist.

Endoscopists are responsible for reviewing suspected polyp regions marked by Caddie and, based on their own medical judgment, confirming the presence or absence of polyps as well as their classification. Caddie is not intended to replace a comprehensive patient evaluation, nor should it be relied upon for primary interpretation of endoscopic procedures, medical diagnoses, or treatment/management plans. The device is limited to use with standard white-light endoscopic imaging.

Olympus stated that Caddie may compete with Medtronic’s AI-powered GI Genius platform. Both devices share the same intended use: to provide real-time assistance in detecting colorectal lesions during screening and surveillance endoscopic mucosal evaluation in adult patients, and neither is intended to replace clinical decision-making. Both systems accept colonoscopy video as input, sourced from an endoscopic image processor, and output a green bounding box to highlight detected polyps.

From an indication perspective, GI Genius is suitable for adult patients undergoing screening and endoscopic mucosal assessment; the patient population for the Caddie device is limited to those aged 45 years and older who are recommended for endoscopic mucosal assessment. This aligns with the recommendations of the U.S. Multi-Society Task Force on Colorectal Cancer (MSTF), which represents the American College of Gastroenterology, the American Gastroenterological Association, and the American Society for Gastrointestinal Endoscopy, advising that colonoscopy screening for average-risk patients should begin at age 45 and continue until age 75.

From a technical perspective, the GI Genius system comprises both software and hardware components; in contrast, the Caddie device does not utilize custom hardware but is instead accessed via network connection on off-the-shelf computers, operating systems, and browsers, functioning as cloud-based software delivered through web access. However, this difference does not give rise to any additional safety or efficacy concerns, as clinical and non-clinical performance testing has already demonstrated the device’s intended use.

The efficacy and safety of Caddie were evaluated in a prospective, multicenter, MRMC, randomized controlled, parallel-group trial across eight medical centers in Europe. Clinical trial results demonstrated that Caddie was significantly superior to standard colonoscopy in detecting adenomas, with a 33% significant increase in the adenoma per colonoscopy (APC) rate. Furthermore, the Caddie device significantly increased the detection rates of small (≤5 mm) and large (≥10 mm) adenomas/adenocarcinomas by 29% and 93%, respectively.

“As a global leader in the field of endoscopy, our vision is to make AI a key component of our digital transformation, and the FDA clearance of Caddie represents a significant milestone in this journey,” said Miquel Àngel García, Global Head of the Olympus Endoscopy Solutions Ecosystem. “Against the backdrop of global industry trends such as cost pressures and staff shortages, a digitally transformed endoscopy ecosystem holds immense potential to support physicians and hospitals.”

In clinical practice, laboratory diagnostics provide over 70% of the information required for clinical decision-making. Characterized by massive data volumes, standardized testing processes, and a high degree of automation, clinical laboratories possess inherent advantages that facilitate deep integration with artificial intelligence (AI) technologies.

According to a report by Global Market Insights, the global market size for medical AI exceeded $5 billion in 2022. Among various applications, medical imaging is one of the most widely adopted fields, which, together with drug discovery, accounts for over 50% of the market. Typical application scenarios for AI in medical imaging include tumor detection and diagnosis, disease screening, assisted diagnosis, as well as treatment planning and evaluation.

China’s AI medical imaging sector started late but has developed rapidly. According to data from Huajing Industry Research Institute, the market size of China’s AI medical imaging industry was approximately RMB 3.62 billion in 2023 and is projected to reach RMB 12.68 billion by 2025, representing a compound annual growth rate (CAGR) of approximately 127.1% from 2020 to 2025. The 2024 Frost & Sullivan Blue Book on AI in Medicine indicates that, as of June 2024, 92 AI-based medical imaging auxiliary diagnosis software products had received Class III medical device certification from the National Medical Products Administration (NMPA) in China.

Among them, Shukun Technology, a leading AI healthcare enterprise in China, has developed the “Digital Doctor” product matrix, a portfolio of AI-assisted diagnostic and therapeutic products covering common major diseases across various body parts. This product portfolio comprises five major modules: “Digital Heart,” “Digital Brain,” “Digital Chest,” “Digital Abdomen,” and “Digital Musculoskeletal.” In addition, Shukun Technology’s self-developed chest CT image processing and analysis software (Lungdimension Doc) has officially obtained the Class III medical device registration certificate from the National Medical Products Administration (NMPA), marking its 13th Class III medical AI certification.

United Imaging Intelligence, the AI subsidiary of United Imaging Healthcare, has launched an AI-powered fracture detection assistance system. It is the industry’s first AI system capable of automatically labeling ribs on every image slice, assisting physicians in “counting ribs.” The system enables precise rib localization, second-level fracture detection, and intuitive lesion highlighting. It supports the detection and classification of various fracture types, achieving a comprehensive sensitivity of over 98% for rib fractures. Additionally, the system features multiple viewing modes, including MPR tri-planar views, VR reconstructions, and local dynamic images, with one-click automatic presentation of the optimal view for fracture assessment.

Airdoc, the first publicly listed AI medical imaging company, offers a core product, Airdoc-AIFUNDUS, which enables rapid diagnosis and assessment of diabetic retinopathy, hypertensive retinopathy, retinal vein occlusion, age-related macular degeneration, pathological myopia, and retinal detachment. Its version 1.0 received the National Medical Products Administration’s (NMPA) first Class III medical device registration certificate for fundus AI. By the end of 2022, Airdoc’s AI-powered retinal imaging products had served nearly 20 million users cumulatively.

Infervision has developed intelligent solutions for screening and diagnosis, covering areas such as thoracic fractures, pneumonia, thoracopulmonary diseases, and stroke. For instance, its InferRead CT Bone intelligent solution for thoracic fractures has received NMPA certification. It enables radiological identification of thoracic fractures and provides triage alerts. The solution features multiple functionalities, including multi-planar reconstruction (MPR), bone volume rendering (VR), rib analysis, deep learning-based image analysis, lesion lists, and examination reports, allowing for precise, second-level analysis of thoracic fractures (including fractures of the ribs, scapulae, clavicles, and sternum).