Related Experiment Video
Updated: Nov 12, 2025
04:23
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
2.1K
Automatic Deep Learning Semantic Segmentation of Ultrasound Thyroid Cineclips Using Recurrent Fully Convolutional
Jeremy M Webb1, Duane D Meixner1, Shaheeda A Adusei2
1Department of Radiology, Mayo Clinic College of Medicine and Science, Rochester, MN 55905, USA.
IEEE Access : Practical Innovations, Open Solutions
|March 22, 2021
Summary
This study introduces a novel deep learning network for segmenting thyroid glands in ultrasound videos. The model accurately identifies thyroid tissue, nodules, and cysts, improving medical imaging analysis.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Deep Learning
Background:
- Medical image segmentation is crucial for applications like standardized reporting and remote diagnostics.
- Automating thyroid gland segmentation in ultrasound cineclips presents significant challenges.
Purpose of the Study:
- To develop and evaluate a novel spatio-temporal recurrent deep learning network for automatic thyroid gland segmentation in ultrasound cineclips.
- To leverage time-sequence information and spatial context for improved segmentation accuracy.
Main Methods:
- A DeepLabv3+ based convolutional Long Short-Term Memory (LSTM) model was trained in four stages.
- The model utilized an atrous spatial pyramid pooling configuration with replicated DeepLabv3+ backbones and convolutional LSTM output layers.
Main Results:
- The proposed model achieved mean intersection over union (IoU) scores of 0.427 for cysts, 0.533 for nodules, and 0.739 for the thyroid gland.
- Demonstrated the effectiveness of convolutional LSTM models in thyroid ultrasound segmentation.
Conclusions:
- The developed spatio-temporal recurrent deep learning network shows significant potential for accurate thyroid ultrasound segmentation.
- This approach can aid in standardized report generation, remote medicine, and potentially reduce medical examination costs.
