AI applications are blurring the fine line in thyroid diagnosis.

Thyroid disorders are increasingly prevalent and affecting younger people, posing significant challenges for early detection and treatment. The application of artificial intelligence to aid diagnosis is a breakthrough, helping to improve the accuracy and effectiveness of treatment.

Challenges in early diagnosis of thyroid tumors

In clinical practice, thyroid lesions are not difficult to detect on ultrasound.

However, the classification and assessment of tumor malignancy, particularly through the TIRADS classification system, remains controversial. This is because the boundary between benign and malignant lesions in the diagnosis of thyroid tumors is extremely delicate and heavily dependent on the experience of the sonographer.

According to Dr. Trinh Tu Tam, Head of the Department of Diagnostic Imaging and Interventional Radiology at Hong Ngoc - Phuc Truong Minh General Hospital, to accurately assess a thyroid lesion, doctors need to evaluate several factors simultaneously, such as the tumor structure (solid or cystic); echogenicity (hypoechoic, isoechoic, or hyperechoic); shape (higher than wider); margins (regular or irregular); and the presence of microcalcifications or coarse calcifications. Classification becomes particularly difficult with small lesions or those with atypical manifestations.

Diagnosing thyroid nodules requires a comprehensive evaluation of many factors.

In fact, many thyroid nodules are classified as low-risk because the ultrasound images are not clear enough. This makes early detection and intervention at the "golden time" more difficult and carries a higher risk of the tumor progressing to malignancy.

Therefore, accurate diagnosis and assessment of thyroid nodules play a crucial role in the early detection of thyroid cancer. When the disease is detected in its early stages, patients can be treated with less invasive methods such as endoscopic thyroidectomy, preserving thyroid function, avoiding extensive lymph node dissection, and even eliminating the need for post-operative radioactive iodine treatment. Furthermore, in cases where the lesion appears benign, this reduces the need for unnecessary fine-needle aspiration biopsy, thereby reducing unnecessary anxiety and costs for the patient.

Artificial intelligence is helping to blur the fine line in thyroid tumor diagnosis.

Given this reality, integrating artificial intelligence into thyroid imaging diagnostic processes will help doctors screen more effectively, diagnose more accurately, and ultimately provide the most optimal and effective treatment methods.

This content was also presented at the "Application of Artificial Intelligence in Screening and Early Detection of Breast and Thyroid Cancer" workshop, held on January 10th at Hong Ngoc - Phuc Truong Minh General Hospital. The workshop brought together more than 200 experts from Vietnam and abroad to discuss, share experiences, and update on the role of AI in diagnosing breast and thyroid cancer.

Also at the event, Dr. Trinh Tu Tam presented "Important Diagnostic Methods in Thyroid Diseases," helping doctors gain a more comprehensive and objective view of the patient's condition through diagnostic image analysis. This allows for earlier detection of suspected lesions, supporting doctors in making accurate and timely diagnostic decisions.

Dr. Trinh Tu Tam presented his report at the conference.

Continuing on the same topic, Assoc. Prof. Dr. Jin Chung - Department of Diagnostic Imaging, Mokdong Hospital, Ewha Women's University (South Korea), shared the latest updates in the diagnosis of thyroid diseases with the integration of artificial intelligence. The report emphasized that modern AI systems act as a reliable reference tool to improve diagnostic reliability and bridge the expertise gap among diagnostic imaging physicians.

Associate Professor, Doctor Jin Chung shared the latest advancements in the field at the conference.

Currently, Hong Ngoc - Phuc Truong Minh General Hospital is applying artificial intelligence (AI) technology in the diagnosis of thyroid tumors, with very high effectiveness observed.

A typical case is that of patient Thanh Thúy (31 years old, Hanoi ), whose thyroid ultrasound revealed a small thyroid nodule of about 5mm in size on the left thyroid lobe. The nodule had many imaging characteristics that made it difficult to clearly distinguish between benign and malignant.

Following a specialist consultation and analysis of the thyroid nodule's condition using an AI system, the doctors determined that the nodule belonged to a high-risk malignant group, according to TIRADS classification group V, suspected to be papillary carcinoma.

Thyroid ultrasound integrated with artificial intelligence (AI) at Hong Ngoc General Hospital

The patient was scheduled for surgery, and an intraoperative frozen section biopsy revealed papillary thyroid cancer, confirming the preoperative diagnosis. Thanks to early detection, Ms. Thuy was scheduled for endoscopic left thyroidectomy – a modern, minimally invasive surgical method that minimizes scarring and ensures post-operative aesthetics.

After surgery, the patient's thyroid lobe is preserved, ensuring stable thyroid function without the need for hormone replacement therapy and adjuvant radioactive iodine treatment.

While artificial intelligence cannot completely replace the role of doctors, it is contributing to reshaping the approach to diagnosing and managing thyroid diseases. When properly integrated into clinical practice, built on a foundation of solid professional training and standardized data, AI helps improve diagnostic accuracy, reduce missed early lesions, and support doctors in making more appropriate treatment decisions for each patient.

Hong Ngoc General Hospital

Source: https://suckhoedoisong.vn/ung-dung-ai-xoa-nhoa-ranh-gioi-mong-manh-trong-chan-doan-tuyen-giap-169260113163357706.htm