Thyroid Disease Classification Using Transfer Learning with EfficientNetB2
Seerat Singla, Rupesh Gupta · 2025
A deep learning-based approach for the classification of thyroid disease using medical images is focused upon in this research work. The EfficientNetB2 model, which was then fine-tuned on a custom dataset containing images labeled as "Normal" and "Thyroid," and ensured that the class imbalance issue was mitigated, and fine adjustments were done on the image preprocessing methods to improve the performance of the model. The model was trained using the training and validation sets combined, and performance was evaluated based on accuracy, precision, recall, F1-score, and confusion matrix. The experimental results gave an impressive 90% classification accuracy for the detected thyroid abnormalities with a high precision of 0.94 for the thyroid class and 0.86 for the normal class. The model was cross-validated to check its ability to generalize, showing robust results in finding thyroid abnormalities. This research explores the application of deep learning models in the classification of medical images, which represents a promising solution for early detection of disease.