Automated Thyroid Nodule Classification in Ultrasound Imaging Using a Hybrid Vision Transformer and Wasserstein GAN with Gradient Penalty

Abstract This study introduces an innovative hybrid model that combines the Vision Transformer (ViT) and Wasserstein Generative Adversarial Networks with Gradient Penalty (WGAN-GP) to enhance the accuracy and robustness of thyroid nodule detection. The Vision Transformer model leverages its powerful attention mechanism to capture global contextual information from ultrasound images. At the same time, WGAN-GP generates high-quality synthetic images to augment the training dataset and address class imbalance issues. The proposed hybrid model is evaluated on a comprehensive dataset of thyroid ultrasonography images, demonstrating significant improvements in classification accuracy, sensitivity, and specificity compared to traditional Convolutional Neural Network (CNN) approaches. The experimental results highlight the potential of integrating ViT-WGAN-GP for automated, reliable thyroid nodule classification, providing a promising tool for medical professionals in diagnostic radiology. This proposed framework surpasses current state-of-the-art diagnostic methods for thyroid-related abnormalities in ultrasound and histopathological datasets and can significantly assist medical professionals by alleviating the excessive burden on the medical fraternity.