Can deep learning effectively diagnose cardiac amyloidosis with 99mTc-PYP scintigraphy?

This study investigates the effectiveness of deep learning models in diagnosing cardiac amyloidosis using 99mTc-PYP scintigraphy. We evaluated more than 40 deep learning models, including both convolutional neural networks (CNNs) and Vision Transformer (ViT) models. The highest-performing model achieved 89.80% accuracy. The study highlights the potential of deep learning methods to improve diagnostic accuracy and reduce patient wait times. These results demonstrate the clinical value of deep learning models in early and accurate cardiac amyloidosis diagnosis, contributing to better patient outcomes and timely interventions.