Intelligent EU-TIRADS classificator for early detection of thyroid anomalies using deep learning convolutional neural network

The paper presents intelligent EU-TIRADS classificator for early detection of thyroid anomalies using a Deep Learning approach. The image data set is consisting of ultrasound images of the Thyroid glance which are classified in five classes: EU-TIRADS – 1, where 1 represents healthy individuals and EU-TIRADS 2-5, where the stage represents the severity of the disease. The Deep learning approach of choice in this experiment is a Deep learning Convolutional Neural Network. This algorithm was selected as it can provide high accuracy without explicit image processing prior to modelling. Finally, the classification performance metrics are presented.