MediAI-OA
by Crescom Co., Ltd.
· Based in South Korea →
— AI-powered software for quantitative knee osteoarthritis analysis.
Overview
MediAI-OA, developed by Crescom Co., Ltd., is an AI-powered software designed for the quantitative analysis of knee osteoarthritis (OA) using X-ray images.
- What it does: The tool automatically processes radiological images to provide quantitative data on key indicators of knee OA. This includes predicting the Kellgren-Lawrence (KL) grade, automatically segmenting knee joint space, and quantitatively analyzing joint space narrowing (JSN) and osteophytes. It also detects the presence of osteophytes and sclerosis.
- Who it is for: MediAI-OA is intended for clinicians, particularly radiologists and orthopedic surgeons, in various care settings. It is also utilized by the Health Insurance Review and Assessment Service (HIRA) in South Korea for knee osteoarthritis insurance reimbursement review processes.
- How it fits a clinical or practice workflow: The software integrates into the workflow by automatically analyzing knee X-ray images and providing results within approximately 5 seconds. This aims to support diagnostic decision-making by offering objective and consistent information on OA indicators, potentially reducing reading deviations.
- Notable capabilities: MediAI-OA provides a probability for the KL grade and offers quantitative measurements of joint space narrowing. It has received a Class II medical device manufacturing license in April 2024 and obtained U.S. Food and Drug Administration (FDA) 510(k) clearance in June 2026.
Reviewed by Pouyan Golshani, MD — Interventional Radiologist
Key Features
- AI-powered radiological image processing
- Automatic knee osteoarthritis severity assessment
- Kellgren-Lawrence (KL) grade prediction
- Automatic segmentation of knee joint space
- Quantitative analysis of Joint Space Narrowing (JSN)
- Automatic detection of medial/lateral JSN(%) and femur/tibia osteophytes
- Improved diagnostic accuracy
- Web-based AI software
Use Cases
- Assessing knee osteoarthritis severity from X-ray images
- Improving efficiency and accuracy of radiological image analysis
- Supporting clinical decision-making for osteoarthritis treatment
- Reducing workload for medical staff in radiology and orthopedics
- Evaluating osteoarthritis characteristics and KL grading
What Physicians Need to Know
Evidence Base
MediAI-OA is a deep learning (DL) software developed to analyze radiographic features of knee osteoarthritis (OA) and grade its severity based on the Kellgren-Lawrence (KL) system. It was trained and validated using data from the Osteoarthritis Initiative, specifically 44,193 radiographs for training and 810 for validation. The software quantifies joint space narrowing (JSN), detects osteophytes in four knee regions, and classifies KL grades.
Clinical Validation Studies
MediAI-OA demonstrated satisfactory performance comparable to experienced orthopedic surgeons and radiologists in analyzing knee OA features, KL grading, and OA diagnosis. In testing with 400 datasets, the software achieved an OA diagnosis accuracy of 0.92. The accuracy of KL grading was 0.83, with a kappa coefficient of 0.768, indicating substantial consistency with ground truth. It also showed an average osteophyte detection accuracy of 0.84. The software's accuracy for classifying KL Grade < 2 and u2265 2 was 0.92, which is considered near-perfect.
Alert Fatigue Management
While the provided information does not specifically detail MediAI-OA's alert fatigue management strategies, AI-driven clinical decision support systems generally aim to reduce alert fatigue by providing intelligent triage and clinical context to filter out unnecessary alerts. Effective strategies include increasing alert specificity, tailoring alerts to patient characteristics, tiering alerts by severity, and making only high-level alerts interruptive. Human oversight remains essential, allowing clinicians to review alert histories, adjust thresholds, and override recommendations.
Drug Interaction Checking
The provided information does not indicate that MediAI-OA includes drug interaction checking capabilities. Drug interaction checkers are separate tools designed to identify potential interactions between medications, supplements, and food, often powered by FDA drug label data.
Differential Diagnosis Support
MediAI-OA focuses on the diagnosis and severity grading of knee osteoarthritis from radiographic images. It provides quantitative analysis of joint space narrowing and osteophyte detection, and classifies the Kellgren-Lawrence grade. While it aids in OA diagnosis, the available information does not explicitly state its capabilities for providing differential diagnosis support for a broader range of conditions.
Guideline Update Frequency
The guideline update frequency for MediAI-OA is not specified in the provided information. However, deep learning models can be continuously refined and updated with new data to improve performance.
Clinical Workflow Integration
MediAI-OA is designed to reduce the burden on radiologists and orthopedic surgeons by providing reliable KL grading and analysis of knee OA features. It automatically quantifies joint space narrowing, detects osteophytes, and classifies KL grades, presenting these results together. This suggests integration into imaging interpretation workflows to enhance efficiency and accuracy.
Decision Audit Trail
The provided information does not explicitly detail a decision audit trail feature for MediAI-OA. However, for AI systems in healthcare, a robust audit trail is crucial. It should record how an AI-assisted decision was produced, reviewed, challenged, and approved, including the request context, output, logic applied, and final action. This ensures traceability, supports compliance, and enables reconstruction of decisions for review.
Physician Tip
MediAI-OA can significantly enhance the efficiency and accuracy of knee osteoarthritis diagnosis and severity grading from radiographs. Physicians should utilize its quantitative analysis of joint space narrowing and osteophyte detection to complement their own assessments. The high accuracy in KL grading can aid in consistent classification and reduce inter-observer variability. While the tool provides robust diagnostic support, it's important to remember that it's an assistive technology, and clinical judgment remains paramount, especially when considering the broader patient context and differential diagnoses. Integrating MediAI-OA into existing imaging review workflows can streamline the diagnostic process and potentially reduce radiologist workload.
MediAI-OA is a deep learning software designed for analyzing knee radiographs. Its primary integration would be within radiology information systems (RIS) and picture archiving and communication systems (PACS) to allow for seamless analysis of X-ray images. The output, including KL grade, joint space narrowing quantification, and osteophyte detection, could then be integrated into electronic health records (EHRs) for comprehensive patient management. Future integrations could explore connections with other clinical data for a more holistic view of OA progression, potentially through federated learning frameworks to maintain data privacy.
Details
| Category | Clinical Decision Support & Reference, Radiology & Imaging AI |
| Pricing | Unknown — unknown |
| Deployment | Web-based |
| Compliance | |
| BAA Available | Unknown AI-estimated |
| HIPAA Compliant | Unknown AI-estimated |
| FDA Status |
Unknown AI-estimated MediAI-OA has received approval from Korea's Ministry of Food and Drug Safety (MFDS) as a Class II medical device. Crescom Co., Ltd. is also listed on the FDA's AI/ML-Enabled Medical Device List. |
| Integrations | |
| EHR | Not specified |
| Specialties | Orthopedics, Radiology, Rheumatology |
Ratings & Reviews
No reviews yet. Be the first to review this tool!
Rate MediAI-OA
No press coverage or publications on file yet.
Know of some? Suggest an edit →
Videos
Product demos, reviews, and walkthroughs for MediAI-OA.
No videos found. Search YouTube directly
Frequently Asked Questions
MediAI-OA is a deep learning-based software designed to analyze knee X-ray images for features of osteoarthritis (OA), including joint space narrowing and osteophyte detection, and to classify the severity using the Kellgren-Lawrence (KL) grading system. It has demonstrated satisfactory performance comparable to experienced orthopedic surgeons and radiologists, with an OA diagnosis accuracy of 92% and KL grading accuracy of 83%.
While effective, MediAI-OA's accuracy for KL grading is 83%, which some newer deep learning software claim to surpass. Additionally, the initial training data for MediAI-OA primarily used the Osteoarthritis Initiative (OAI) dataset, which did not include lateral radiographs, potentially limiting its comprehensive analysis compared to approaches that incorporate more diverse imaging views.
MediAI-OA has received manufacturing approval as a Class II medical device from the Korea Ministry of Food and Drug Safety in 2024. Physicians should ensure that any AI tool, including MediAI-OA, adheres to high standards of privacy and data security and is used to support, not replace, clinical judgment, in line with evolving national guidelines for AI in healthcare.
Yes, other AI software like MedKnee and KOALAu2122 also assist in knee OA diagnosis. MedKnee, for instance, claims a higher accuracy of 97.20% in classifying knee OA severity compared to MediAI-OA's 83% for KL grading. KOALAu2122 focuses on metric evaluations of knee joint imaging to reduce diagnostic errors.
Specific pricing for MediAI-OA is not publicly detailed, but similar medical AI imaging solutions often employ per-study pricing, with costs varying based on study volume. Some providers offer tiered plans, custom quotes, and may include setup fees for integration and training.
MediAI-OA is designed to preprocess knee radiographs, detect knee positions and regions of interest, and then perform quantitative analysis and KL grading. Integration with existing PACS is a common feature for medical AI imaging software, often covered by a setup fee that includes initial configuration and team training.
MediAI-OA was trained and validated using data from the Osteoarthritis Initiative (OAI), specifically 44,193 radiographs for training and 810 for validation. The quality and completeness of this training data are critical, as incomplete or skewed data can lead to inequitable outcomes or disparities in diagnosis.
Suggest an Edit →
| Last Verified: 2026-09-23 | First Added: 2026-09-23