Abstract
Objective: In recent years, the application of deep learning (DL) technology in the thyroid field has expanded rapidly, driving substantial innovation in thyroid disease research. This review aims to provide clinicians with the latest research advances in the application of DL to the diagnosis and treatment of thyroid cancer.
Methods: A systematic review was conducted of studies published in the past five years in the PubMed database on the application of deep learning in the diagnosis, treatment, and prognosis of thyroid cancer.
Results: DL has made substantial advances in the diagnosis and treatment of thyroid cancer, particularly through the application of advanced models such as convolutional neural networks, long short-term memory networks, and generative adversarial networks. These models have delivered breakthrough performance in key areas, including ultrasound image analysis of thyroid nodules, automated classification of pathological images, and assessment of extrathyroidal extension. DL also shows considerable promise for individualized treatment planning and prognosis prediction. Nonetheless, its widespread clinical adoption is hindered by substantial technical, clinical, and ethical challenges. Addressing these barriers is crucial to achieving meaningful improvements in thyroid cancer care and realizing the full potential of DL in precision medicine.
Conclusion: DL techniques are advancing the precision diagnosis and treatment of thyroid cancer and hold the potential to enhance diagnostic accuracy and improve therapeutic outcomes for patients.
Methods: A systematic review was conducted of studies published in the past five years in the PubMed database on the application of deep learning in the diagnosis, treatment, and prognosis of thyroid cancer.
Results: DL has made substantial advances in the diagnosis and treatment of thyroid cancer, particularly through the application of advanced models such as convolutional neural networks, long short-term memory networks, and generative adversarial networks. These models have delivered breakthrough performance in key areas, including ultrasound image analysis of thyroid nodules, automated classification of pathological images, and assessment of extrathyroidal extension. DL also shows considerable promise for individualized treatment planning and prognosis prediction. Nonetheless, its widespread clinical adoption is hindered by substantial technical, clinical, and ethical challenges. Addressing these barriers is crucial to achieving meaningful improvements in thyroid cancer care and realizing the full potential of DL in precision medicine.
Conclusion: DL techniques are advancing the precision diagnosis and treatment of thyroid cancer and hold the potential to enhance diagnostic accuracy and improve therapeutic outcomes for patients.
| Original language | English |
|---|---|
| Pages (from-to) | 1608-1614 |
| Number of pages | 7 |
| Journal | Endocrine Practice |
| Volume | 31 |
| Issue number | 12 |
| Early online date | 30 Jul 2025 |
| DOIs | |
| Publication status | Published - Dec 2025 |
| Externally published | Yes |
Bibliographical note
Publisher Copyright:© 2025
Funding
This study was supported by grant from the Noncommunicable Chronic Diseases- National Science and Technology Major Project (No. 2024ZD0525600), National Natural Science Foundation of China (No. 82372600, 82170803), and GDPH Supporting Fund for Talent Program (No. KJ012020629).
Keywords
- deep learning
- thyroid cancer
- convolutional neural network
- thyroid ultrasound
- precision medicine
