Advantages
- It has been confirmed that the Agatston score measured from general body CT images has an evaluation accuracy equivalent to that of cardiac CT.
- This leads to the early detection of AS patients by utilizing CT imaging data from health checkups and examinations for other diseases.
Technology Overview & Background
Aortic stenosis (AS) is a disease where blood flow to the body is obstructed by the narrowing of the aortic valve, primarily caused by age-related valve calcification. It often progresses without symptoms, but the prognosis is poor once symptoms appear. Recently, the effectiveness of early surgical treatment for asymptomatic severe AS has been demonstrated, creating a need for improved early detection and diagnostic techniques. However, there is no established method for early diagnosis; echocardiography, used for severity assessment, is typically performed only after symptoms emerge. Furthermore, while the Agatston score (cardiac CT AVAS) is used to evaluate the degree of aortic valve calcification, it requires an ECG-gated cardiac CT, making it unsuitable for widespread screening.
To address this, the inventor focused on a method to calculate the Agatston score from more common body CT scans (body CT AVAS) and developed an application that uses a deep learning-based AI to automatically quantify the score. The body CT AVAS calculated by this application showed a high correlation with the conventional cardiac CT AVAS, confirming its usefulness as an alternative to cardiac CT. This achievement suggests that it is possible to evaluate calcification in AS patients by utilizing CT image data from health checkups and examinations for other diseases. This will enable the comprehensive detection of asymptomatic severe AS even from CT scans not originally intended for AS assessment, with the potential to reduce medical costs by expanding opportunities for early therapeutic intervention.
 |
Data
- The method was applied to 264 cases of body CT data (165 for training, 99 for testing) after a preprocessing step to segment the chest data.
- An evaluation of the concordance between the AI’s automated quantitative values and the ground truth values manually quantified by experts confirmed a good agreement for the Agatston score (AVAS) with an ICC = 0.841. Similarly, high agreement was shown for calcium volume (AVCV) with an ICC=0.835 and calcium mass (AVCM) with an ICC=0.835.
Patent and Publication
Patent pending (unpublished)
Kisohara M,et al. https://doi.org/10.1016/j.jcct.2025.03.008
Principal Investigator & Academic Institution
Dr. Masaya Kisohara (Nagoya City University)
Current Stage & Future Research Plans
- Current Stage: The calcium amount was quantified and the Agatston score was calculated from body CT data, achieving a high correlation with the results from coronary artery CT calcium scans.
- Next Stage: The developer has demonstrated that the AVAS derived from non-ECG-gated standard CT shows high correlation with the AVAS from cardiac CT, suggesting it can be used as a substitute or as a gateway for detecting severe AS. To apply this finding more generally, we plan to conduct research to screen patients with severe AS by applying the newly developed AI to CT scans taken in general clinical practice to automatically quantify the AVAS.
Expectations
Nagoya City University is seeking companies to jointly develop diagnostic imaging software based on this invention. Disclosure of unpublished data under a non-disclosure agreement with Nagoya City University is possible, as are direct meetings and discussions with the researcher.
Project ID:WL-05189