Pointing to the artificial intelligence assisted diagnosis system for thyroid nodules introduced by the hospital, Zhao, director of the ultrasound department of the First Affiliated Hospital of Zhejiang University, told the media.
This system is still very easy to use, as the doctor does not need additional steps, just need to take a picture as in daily examination, after the identification of the nodule to tell where, what kind of contour, and at the same time how benign or malignant. Overall it's good, but of course it's in the process of being updated and optimized.
The doctor said the artificial intelligence, is a set of ultrasound-based image of the thyroid nodule benign and malignant auxiliary diagnostic system developed by Zhejiang Deshaun Rhyme Xing Image Technology Co.
For the diagnosis of thyroid nodules, the current average accuracy rate of doctors in tertiary hospitals is 60%-70%, and grassroots hospitals will be even lower. And the auxiliary diagnostic system can reach an accuracy rate of more than 85%, said Ms. Hu Hairong, general manager of Deshaun Yunxing, because artificial intelligence is not qualified to practice medicine in China, so we position it as an auxiliary diagnosis and treatment, just to provide some reference for doctors. The future market positioning of this system is grassroots hospitals, after the promotion can help grassroots hospitals to improve the level of diagnosis, which is conducive to the implementation of the national policy of hierarchical diagnosis and treatment.
It is understood that the thyroid nodule is a lump in the thyroid gland, can move up and down with the thyroid gland with the swallowing movement, is a common clinical condition, can be caused by a variety of causes.
Most thyroid nodules are benign, but a few are malignant to begin with, and some will change from benign to malignant. Therefore, once a thyroid nodule is detected during a physical examination, the doctor will recommend further testing, mainly to distinguish whether the nodule is 'benign' or 'malignant'.
The percentage of thyroid nodules that turn into cancer is 5 percent, and if detected early and treated as soon as possible, most patients can survive for a long time.
The examination of thyroid nodules mainly relies on ultrasound, which is cheaper, more flexible, more real-time, and has no side effects than CT and MRI, making it a popular test for the general public; however, its drawbacks are that the images are unclear and the imaging is large, which makes it more demanding for doctors. The advantages of CT and MRI are clear images and the disadvantages are high testing costs and side effects.
From the point of view of the application of artificial intelligence to the recognition of medical images or assisted diagnosis, ultrasound is more difficult than CT and NMR, and we wanted to test the performance, efficiency and accuracy of the algorithm, so we chose ultrasound. At the same time, as a superficial organ, the ultrasound of the thyroid is not as complex as the abdomen and heart, so it is easier to make results on the ground.