The paper proposed an approach for thyroid cancer detection based on artificial intelligence in Internet of Medical Imaging Things (IoMIT) ecosystem. Ultrasonic imaging collected in IoMIT ecosystem is the best way for thyroid cancer diagnosis. Image segmentation and detection of benign and malignant thyroid nodules is an important part of the proposed approach. It is implemented in Apache Spark using MLlib based on Convolutional Neural Networks (CNNs). Finally, the results of medical imaging analytics are discussed.