According to Associate Professor Huong, there have been previous studies in Vietnam on developing AI diagnostics for thyroid cancer, but they rely heavily on ultrasound images.

In this project, K Hospital used all three data sources: ultrasound image (Ultrasound Image), test results of fine needle aspiration cytology of thyroid nodules (FNA result data), and results Surgical pathology data of cancer patients to provide input for the development of a more accurate AI model.

The model will be trained on a dataset of 1,800 Vietnamese patients. K Hospital's doctors are directly involved in data labeling, while KT's engineers are in charge of developing AI algorithms and software.

Once training is complete, the AI model will be able to read ultrasound images to detect thyroid nodules in the images, predict benign/malignant grade for the screened area, and characterize of thyroid nodules according to the ACR TI-RADS thyroid imaging and data classification system.

Thyroid nodules appear so frequently that they often lead to suspicion of disease and need to be examined. In fact, only 1 in 10 thyroid nodules is malignant.

Overdiagnosing cancer not only causes unnecessary surgeries and complications, but also emotional and financial suffering. This can be partially overcome with powerful AI-powered diagnostic tools.

The project is expected to have results by the end of August 2023.