A MLP model to generate data for DOT difference imaging from target measurements only.
Significance: “Difference imaging”, which reconstructs target optical properties using measurements with and without target information, is often used in diffuse optical tomography (DOT) in vivo imaging. However, taking additional reference measurements is time-consuming, and mismatches between the target medium and the reference medium can cause inaccurate reconstruction. Aim: We aim to simplify the data acquisition and mitigate the mismatch problems in DOT difference imaging by using a deep learning-based approach to generate data from target measurements only.
The code was tested with Python 3.7.11.
Required packages:
- pytorch 1.8.1
- numpy
- pandas
- matplotlib
- scikit-learn
The MLP model is the class TarToPert in models.py. The bottleneck is tunable by changing the parameter "neck". The MLP model structures may need to be changed according to uses' DOT system configuration and the size of the dataset.
The training & validation is done by tar_to_pert_train.py. The users need to use the dataset with their own.
The pre-trained model is saved in TarToPert_pretrained.pth.
A sample testing code is given in main_phantom_tar2pert.py. Sample phantom data was given as phantom_tar1.csv and the generated perturbation will be saved as `pert_pred_phantom.csv'.
Li, S., Zhang, M., Xue, M. and Zhu, Q., 2022. Difference imaging from single measurements in diffuse optical tomography: a deep learning approach. Journal of Biomedical Optics, 27(8), p.086003.