Journals / Niğde Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi / 2025 / Cilt: 18

Vision transformer-based multi-class classification of aortic calcification scores

Aort kalsifikasyon skorlarının görü dönüştürücü tabanlı çok sınıflı sınıflandırması

Mervenur Çakır(KARADENİZ TEKNİK ÜNİVERSİTESİ, MÜHENDİSLİK FAKÜLTESİ)0000-0002-8845-583X
Murat Ekinci(KARADENIZ TECHNICAL UNIVERSITY)0000-0001-9326-8425
Elif Baykal Kablan(KARADENIZ TECHNICAL UNIVERSITY)0000-0003-3552-638X
Mürsel Şahin(KARADENIZ TECHNICAL UNIVERSITY)0000-0003-0245-2038
Pages
—
Publication Date
2026-07-20
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

By providing a radiation-free, reproducible, and economically accessible solution, the proposed framework offers a clinically applicable decision-support system for the evaluation of aortic stenosis.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:This study presents a fully ultrasound-based and cost-effective approach for the automatic grading of aortic valve calcification, which plays a critical role in the assessment of aortic stenosis.
📊
Key Findings & Data:A Vision Transformer (ViT) model operating on ROI frames extracted from the aortic valve region was developed to classify four calcification levels: mild, moderate, severe, and critical.
💡
Conclusion & Impact:By providing a radiation-free, reproducible, and economically accessible solution, the proposed framework offers a clinically applicable decision-support system for the evaluation of aortic stenosis.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Aort kapağı · Ekokardiyografi · Kalsiyum skoru

Abstract

This study presents a fully ultrasound-based and cost-effective approach for the automatic grading of aortic valve calcification, which plays a critical role in the assessment of aortic stenosis. To eliminate radiation exposure associated with computed tomography, the proposed method relies exclusively on echocardiographic images and was trained on a dedicated dataset constructed for this study. A Vision Transformer (ViT) model operating on ROI frames extracted from the aortic valve region was developed to classify four calcification levels: mild, moderate, severe, and critical. During testing, the model achieved 85% accuracy and a macro-averaged precision of 0.74, demonstrating a reliable decision mechanism with a low false-positive tendency. To further evaluate the architectural choice, ImageNet pre¬trained ResNet50 and EfficientNet-B3 models were trained and tested under identical conditions. Although CNN-based architectures produced competitive results, the ViT model demonstrated more balanced performance, particularly in intermediate and advanced grades, and achieved the highest macro-averaged metrics among the evaluated models. By providing a radiation-free, reproducible, and economically accessible solution, the proposed framework offers a clinically applicable decision-support system for the evaluation of aortic stenosis.

Özet

Bu çalışma, aort darlığının değerlendirilmesinde kritik öneme sahip aort kapak kalsifikasyonunun otomatik derecelendirilmesi için tamamen ultrason tabanlı ve düşük maliyetli bir yaklaşım sunmaktadır. Bilgisayarlı tomografi yöntemlerinde maruz kalınan radyasyonu ortadan kaldırmak amacıyla yöntem yalnızca ekokardiyografi görsellerini kullanmakta ve çalışmaya özel oluşturulan bir veri seti ile eğitilmiştir. Aort kapak bölgesinden elde edilen ROI frameleri üzerinden çalışan Vision Transformer (ViT) modeli, düşük, orta, yüksek ve kritik olmak üzere dört kalsifikasyon seviyesini sınıflandırmak üzere geliştirilmiştir. Test aşamasında model %85 doğruluk ve 0.74 ortalama precision elde ederek güvenilir ve düşük yanlış pozitif oranına sahip bir karar mekanizması ortaya koymuştur. Mimarinin etkinliğini değerlendirmek amacıyla aynı veri seti üzerinde ImageNet ön-eğitimli ResNet50 ve EfficientNet-B3 modelleriyle karşılaştırmalı deneyler gerçekleştirilmiştir. CNN tabanlı modeller rekabetçi sonuçlar üretmekle birlikte, ViT modeli özellikle ara ve ileri derecelerde daha dengeli performans göstermiş ve makro ortalama metriklerde en yüksek başarıyı elde etmiştir. Ultrason temelli olması sayesinde radyasyona maruz bırakmayan, tekrarlanabilir ve ekonomik bir çözüm sunan bu yaklaşım, aort darlığının değerlendirilmesinde klinik olarak uygulanabilir bir karar destek sistemi sağlamaktadır.

🔖

Cite This Article

Çakır, M., Ekinci, M., Kablan, E. B., Şahin, M., *, M. Ç. (2026). Vision transformer-based multi-class classification of aortic calcification scores. Niğde Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi, Cilt: 18. https://doi.org/10.28948/ngumuh.1830650

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (11)

  1. R. L. J. Osnabrugge et al., "Aortic stenosis in the elderly: disease prevalence and number of candidates for transcatheter aortic valve replacement: a meta-analysis and modeling study," Journal of the American College of Cardiology, vol. 62, no. 11, pp. 1002-1012, 2013. https://doi.org/10.1016/j.jacc.2013.05.015
  2. L. Tastet et al., "Impact of aortic valve calcification and sex on hemodynamic progression and clinical outcomes in AS," Journal of the American College of Cardiology, vol. 69, no. 16, pp. 2096-2098, 2017. https://doi.org/10.1016/j.jacc.2017.02.037
  3. T. A. Pawade, D. E. Newby, and M. R. Dweck, "Calcification in aortic stenosis: the skeleton key," Journal of the American College of Cardiology, vol. 66, no. 5, pp. 561-577, 2015. https://doi.org/10.1016/j.jacc.2015.05.066
  4. V. Falk et al., "2017 ESC/EACTS Guidelines for the management of valvular heart disease," European Journal of Cardio-Thoracic Surgery, vol. 52, no. 4, pp. 616-664, 2017. https://doi.org/10.1093/ejcts/ezx324
  5. A. S. Agatston et al., "Quantification of coronary artery calcium using ultrafast computed tomography," Journal of the American College of Cardiology, vol. 15, no. 4, pp. 827-832, 1990. https://doi.org/10.1093/ejcts/ezx324
  6. L. Tang et al., "DLFFNet: A new dynamical local feature fusion network for automatic aortic valve calcification recognition using echocardiography," Computer Methods and Programs in Biomedicine, vol. 243, p. 107882, 2024. https://doi.org/10.1016/j.cmpb.2023.107882
  7. M. Çakır et al., "AVD-YOLOv5: a new lightweight network architecture for high-speed aortic valve detection from a new and large echocardiography dataset," Medical & Biological Engineering & Computing, vol. 62, no. 8, pp. 2511-2528, 2024. https://doi.org/10.1007/s11517-024-03090-3
  8. M. Cakir et al., "A New Aortic Valve Calcium Scoring Framework for Automatic Calcification Detection in Echocardiography," Journal of Imaging Informatics in Medicine, pp. 1-19, 2025. https://doi.org/10.1007/s10278-025-01576-6
  9. L. B. Elvas et al., "Deep learning for automatic calcium detection in echocardiography," BioData Mining, vol. 17, no. 1, p. 27, 2024. https://doi.org/10.1186/s13040-024-00381-1
  10. L. B. Elvas et al., "AI-based aortic stenosis classification in MRI scans," Electronics, vol. 12, no. 23, p. 4835, 2023. https://doi.org/10.3390/electronics12234835
  11. H. Vaseli et al., "Protoasnet: Dynamic prototypes for inherently interpretable and uncertainty-aware aortic stenosis classification in echocardiography," in Proc. International Conference on Medical Image Computing and Computer-Assisted Intervention, 2023. https://doi.org/10.48550/arXiv.2307.14433