The 9th Qilu International Conference on Cardiovascular Disease (QICCD 2020) was held virtually Sept. 25-27. Experts and scholars across China gathered online to participate in a discussion focused on cardiovascular disease diagnosis and treatment.
Dr. Kunlin Cao, President of Keya Medical R&D, participated in the Cardiovascular Disease Imaging Forum alongside Dr. Pengfei Zhang of Shangdong University. Dr. Cao and Dr. Zhang presented the latest developments in deep learning-based CT-FFR and examined how AI is bringing value to both clinicians and patients.

Clinical Applications of CT-FFR

Dr. Zhang opened the session by discussing the clinical applications of CT-FFR. In his discussion, Dr. Zhang combined academic research and clinical data to explain the value of CT-FFR in clinical practice.
Dr. Zhang shared that the core product of coronary heart disease is myocardial ischemia. Traditional coronary CT angiography (CTA) can only reflect the degree of coronary artery stenosis and cannot directly reflect the coronary blood supply function, and stenosis does not necessarily equal ischemia. Unnecessary invasive coronary angiography (ICA) examinations and percutaneous coronary interventions (PCI) for these patients will not only increase the medical risks and economic burden on patients, but it will also cause waste of medical resources.
Computed tomography fraction flow reserve (CT-FFR) is a new, non-invasive technology that has received significant clinical attention in recent years. The method evolved from the invasive measurement of Fractional Flow Reserve (FFR) in which a guidewire would be used to measure blood pressure and blood flow through a specific part of the coronary artery. CT-FFR combines anatomical and physiological factors to achieve a non-invasive and accurate diagnosis of coronary artery disease.
However, early CT-FFR technology was time-consuming, expensive, and had poor scalability. With the application of deep learning, CT-FFR technology can generate a three-dimensional reconstruction of the coronary vascular tree structure while improving the accuracy of diagnosis.
The diagnostic value of deep learning-based CT-FFR has been confirmed in a number of studies. CT-FFR can be used as a non-invasive examination method for myocardial ischemia and has demonstrated potential value in assisting the decision-making of revascularization. The efficiency has a high degree of consistency and it is expected to become a powerful tool for aiding decision-making in the treatment of coronary artery disease.
According to Dr. Zhang, studies have shown that AI-based CT-FFR can improve the diagnostic performance of functional myocardial ischemia compared to ICA and PCI; the method has shown to be consistent with FFR measured invasively using a pressure guidewire, with the added benefit of the non-invasive technology reducing patient discomfort.

Applications for AI in Cardiovascular Disease Diagnosis and Treatment

In the second section of the forum, Dr. Kunlin Cao discussed how AI can be applied to coronary CTA images to improve the process of diagnosing and treating patients with cardiovascular diseases. According to Dr. Cao, AI can be used in the early stages of screening and diagnosis to assist in the automatic detection of lesions and to provide quantitative indicators required for clinical use. In terms of surgical planning, AI can be applied to assist doctors in formulating personalized surgical plans. AI is able to comprehensively analyze multi-source information, such as images, case reports, and patient health data, providing medication guidance, risk prediction, disease tracking, follow-up management, and other services to build a full-cycle diagnosis and treatment management system for patients.
In the diagnosis of patients with coronary artery disease, the current clinical evaluation is mainly used to evaluate the vascular structure stenosis and functional ischemia based on FFR. According to the recommendations provided in China’s Coronary Intervention Guidelines, if the diameter of the lesion of stenosis is less than 90%, intervention should be carried out for the lesion with FFR less than or equal to 0.8.
Dr. Cao continued the lecture by sharing that Keya Medical is developing products using deep learning to more accurately diagnose patients suspected with coronary artery disease while reducing unnecessary invasive procedures. By minimizing the commonality and individuality of different case data, Keya Medical’s products can realize fast and reliable analysis and calculation for the early diagnosis and personalized treatment of cardiovascular diseases.

Keya Medical is currently developing AI products using coronary CTA scans to enable more accurate diagnosis and treatment for patients with cardiovascular diseases. DEEPVESSEL FFR, Keya Medical’s first CE-marked and National Medical Products Administration (NMPA) approved product, uses deep learning technology to perform a non-invasive physiological functional assessment of the coronary arteries using coronary CTA scans. DEEPVESSEL FFR enables clinicians to automatically calculate the FFR value at any location on the coronary artery tree, reducing unnecessary invasive procedures. After conducting multi-center retrospective and prospective clinical trials in 2017 and 2019, DEEPVESSEL FFR demonstrated that it can improve the coronary CTA diagnostic specificity and is consistent with invasive FFR. Keya Medical is actively recruiting collaboration partners in the United States to help bring DEEPVESSEL FFR to clinical use.

Since 2016, Keya Medical has collaborated with over 200 hospitals around the world to enable better treatment of patients with coronary artery disease. Keya Medical looks forward to continuing to collaborate with more health systems worldwide to bring the benefits of AI to both patients and providers.