Detection of Thyroid Nodule Disease by using Deep Convolution Neural Network
Nowadays thyroid disease is more common in nature. In India, one out of ten people gets easily affected by this disease. The endocrine gland present in the neck stimulates two hormones namely triiodothyronine and thyroxine that regularize the metabolism in our body. It is a very essential hormone for the digestion of food by which it conserves energy for fulfilling our daily activities. It causes diseases like hypothyroidism, hyperthyroidism, thyroid nodule, and thyroiditis. Detection of these diseases helps to gradually minimize their potential risks sometimes it protects our life with timely treatment. The traditional method of identifying these diseases is highly dependent upon a doctor's skill and experience. The development of radiological imaging and Magnetic Resonance Imaging (MRI) techniques facilitates the detection process even then it requires an expert's knowledge. The evolution of deep learning techniques and its progress towards computer vision, image processing, and natural language processing initiates us to build an effective computer-assisted diagnostic system that serves as a screening tool for doctor’s analysis. Our proposed diagnostic system employs a deep convolution neural network for detecting the thyroid nodule disease and its associated class types viz. benign and malignant. We have experimented with our proposed system using open-access image data of 99 cases available in an online repository. The proposed diagnostic system gives a remarkable performance with an accuracy of 75.73%, a precision of 72.73%, recall 100%, and f1score 84.21%.