Method and apparatus for analyzing midbrain hyperechoic area based on neural network
By using a neural network-based approach, FasterRcnn and deeplab v3+ neural networks are used to automatically identify and segment the strong echo regions in the midbrain, solving the problems of inaccurate and cumbersome assessment in existing technologies, and achieving efficient and accurate assessment of Parkinson's disease.
Patent Information
- Application Number
- CN202310116450.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-02-15
- Publication Date
- 2026-07-24
- Estimated Expiration
- 2043-02-15
AI Technical Summary
In existing technologies, diagnostic methods for Parkinson's disease, such as transcranial substantia nigra ultrasound and electron emission computed tomography, suffer from problems such as large errors or high costs, making it difficult to accurately assess the area of hyperechoic regions in the midbrain, resulting in an insufficiently objective and convenient assessment of the condition.
A neural network-based approach was adopted, using Faster R-CNN and DeepLab v3+ neural networks to automatically identify and segment the hyperechoic regions of the midbrain. Combined with feature extraction and candidate box regression techniques, quantitative analysis of the hyperechoic regions of the midbrain was achieved.
It improves the accuracy of quantitative analysis of the area of strong echogenic regions in the midbrain, simplifies the operation process, and enables automatic and objective assessment of the progression of Parkinson's disease.
Smart Images

Figure CN116363062B_ABST
Abstract
Citation Information
Patent Citations
Transcranial ultrasound nigra high echo intensity quantitative analysis method
CN113538380A