Thyroid Detection and Recognition Based on Multi-Layer Recursive Neural Network (ML-RNN) Using in Deep Learning

One of the most prevalent illnesses in the world is thyroid disease. Over 42 million individuals in India suffer from thyroid gland disorders. All of our metabolic processes are regulated by the thyroid, a little gland in the neck that secretes thyroid hormones. All facets of human health are impacted by thyroid gland dysfunction. Thyroid illness testing is crucial to healing because it necessitates a thorough study. Hormone secretion can be balanced and neutralising problems can be overcome with early diagnosis and treatment of the disease. Numerous patients and a shortage of doctors are posing a threat to conventional diagnosis methods for thyroid disorders. To overcome the issues, deep learning based Multi-Layer Recursive Neural Network (ML-RNN) mainly focuses on pre-processing the input, selecting features based on standard datasets, extracting features from the selected features and classifying thyroid diseases into normal, hyperthyroid and hypothyroidism. First step is preprocessing to improve the obtained values based on clearing, data splitting and handling the missing values. The second stage is feature selection is based on the Fisher score method used for selecting the suboptimal features set. Data information is computed based on Region-of-Interest (ROI) volumes. Classification is based on Multi-Layer Recursive Neural Network (ML-RNN) for analysis of the classification performance based on improving the accuracy and predicting the risk getting thyroid disease. The accuracy, recall, predicted positive value, and predicted negative value of the classification results were high respectively.