Thyroid Nodule Classification Model from Scintigraphy Images using Convolution Neural Networks

Technological advances have allowed the machine learning area of the field of artificial intelligence to emerge and provide new solutions to different problems. Medicine being one of the sciences that is driving other advance solutions. Using realtime data, a model has been developed for classification of images of the thyroid gland taken through scintigraphy to assist in the diagnosis of thyroid glands. The thyroid is a very important gland for the functioning of the body, since it regulates the functioning of different organs through the secretion of thyroid hormones. For the development of this work, convolutional neural networks were used with a total of 5,600 images for training and testing, the images were extracted from medical reports and in the end an accuracy level of 95.40% was reached in all cases as training and evaluation. This percentage was considered acceptable by the medical specialist.