# Deep-Learning-for-PET-Imaging

**Repository Path**: Heconnor/Deep-Learning-for-PET-Imaging

## Basic Information

- **Project Name**: Deep-Learning-for-PET-Imaging
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2024-04-03
- **Last Updated**: 2025-07-03

## Categories & Tags

**Categories**: Uncategorized

**Tags**: reconstruction

## README


# Deep-Learning-for-PET-Imaging

- URL: https://github.com/AlessandroGuazzo/Deep-Learning-for-PET-Imaging
- Author: Alessandro Guazzo
- Description: A collection of codes that I developed during my master thesis project at KTH

## Reference

[1] A. Guazzo and M. Colarieti-Tosti, “Learned Primal Dual Reconstruction for PET,” Journal of Imaging, vol. 7, no. 12, 2021, doi: 10.3390/jimaging7120248.


## 目录结构

### Denoising
- **Encoder-Decoder**
  - `2D-TrainED3LayersVariableNoise.py` - 用于训练具有检查点策略的3层去噪编码器-解码器的Python代码
  - `2D-ED3LayersResultsEvaluationVariableNoise.ipynb` - 用于评估3层去噪编码器-解码器结果的Jupyter Notebook

- **U-Net**
  - `2D-TrainUnet3LayersVariableNoise.py` - 用于训练具有检查点策略的3层去噪U-Net的Python代码
  - `2D-Unet3LayersResultsEvaluationVariableNoise.ipynb` - 用于评估3层去噪U-Net结果的Jupyter Notebook

### Reconstruction
- **Learned Primal-Dual**
  - `2D-TrainLPD3IterVariableNoise.py` - 根据渐进式学习策略，从2次迭代开始训练3次迭代的学习型原始对偶算法的Python代码
  - `2D-LPD3IterResultsEvaluationVariableNoise.ipynb` - 用于评估3次迭代学习型原始对偶算法结果的Jupyter Notebook

- **Learned Update with Memory**
  - `2D-TrainLUM4IterVariableNoise.py` - 从4次学习更新算法开始训练4次迭代的学习更新与内存算法的Python代码
  - `2D-LUM4IterResultsEvaluationVariableNoise.ipynb` - 用于评估4次迭代学习更新与内存算法结果的Jupyter Notebook

- **Learned Update**
  - `2D-TrainLU4IterVariableNoise.py` - 根据渐进式学习策略，从3次迭代开始训练4次迭代的学习更新算法的Python代码
  - `2D-LU4IterResultsEvaluationVariableNoise.ipynb` - 用于评估1次、2次、3次、4次迭代学习更新算法结果的Jupyter Notebook

### miniPET
- **LPD Reconstruction**
  - `2D-TestLPD3miniPET.ipynb` - 用于测试LPD在miniPET数据上的性能的Jupyter Notebook
  - `2D-Train3LPDminiPETHybrid.py` - 用于使用Hybrid方法训练LPD算法的Python代码
  - `2D-Train3LPDminiPETOnly.py` - 用于使用仅miniPET方法训练LPD算法的Python代码

- **Measurments**
  - `IQphantom.sino.mnc` - 包含从miniPET测量获得的直接正弦图的文件
  - `MeasureSimulatorMouse.xlsx` - 用于设置带有老鼠模体的测量的Excel文件
  - `MeasureSimulatorP1.xlsx` - 用于设置带有第一个模体的测量的Excel文件
  - `MeasureSimulatorP2.xlsx` - 用于设置带有第二个模体的测量的Excel文件
  - `MeasureSimulatorP3.xlsx` - 用于设置带有第三个模体的测量的Excel文件
  - `Measurements Description.txt` - 用于报告所有执行测量的有效活动水平的文件
  - `processing.sh` - 用于处理miniPET采集提供的.clm.lr5文件的Linux bash文件

## 描述
这些文件夹包含用于不同深度学习模型的训练脚本和评估Notebook，这些模型用于正电子发射断层扫描(PET)中的图像重建和去噪。此外，miniPET文件夹包含用于实际miniPET设备数据的训练脚本和数据处理工具。