This repository contains the source code of the paper Regression of the Rician Noise Level in 3D Magnetic Resonance Images from the Distribution of the First Significant Digit (URL doi paper https://doi.org/10.3390/axioms12121117).
This code executes the method mentioned in the paper for a set of input images. The contents of this code are provided without any warranty. They are intended for evaluation purposes only.
Pre-requisites The system operated on Ubuntu 20.04 and utilized Python 3.8, 237 along with various scientific libraries matplotlib, nibabel, scipy, and sklearn.
Set up Open GenerateTrainingDataBenfordMRI.ipynb, Open CrossValidatedModelTrainingBenfordMRI.ipynb, and FiguresForPaper.ipynb. Paths are prepared to work on any computer. Parameter values are assigned by default. But some parameters' values can be modified. Read any .ipynb file.
1.- GenerateTrainingDataBenfordMRI.ipynb you must choose: - repository: the name of the MRI dataset. - directory : the directory where the MRI is located. Run GenerateTrainingDataBenfordMRI.ipynb.
2.- CrossValidatedModelTrainingBenfordMRI.ipynb you must choose: - repository: the name of the MRI dataset. Run CrossValidatedModelTrainingBenfordMRI.ipynb.
3.- FiguresForPaper.ipynb you must choose: - dataset_path: refer a pkl file, created by GenerateTrainingDataBenfordMRI.ipynb. - result_path: a folder where save the pdf output file. - NdxSet: a integer, refering the couple of metrics. The value of NdxSet is in the table in the .doc file. - predictedValuesOfBestModel: refer the best model, its mean, the values to do the figure. The values of predictedValuesOfBestModel is in the table in the .doc file. Run FiguresForPaper.ipynb.
Create '../input/input_to_do_score' folder. Copy and paste inside in all CrossValidationResults.xls files, only the columns to do the score: metrics column and regressors columns, sort by best regressor.
4.- Score_table_score.ipynb you must choose: - sheet_name variable. - The element in newList variable. Run Score_table_score.ipynb three times, one by each sheet_name varuable: MSE, MAE and R2.
5.- Score_table_performanceMeasure.ipynb you must choose: - Define the path to CrossValidationResults.xls files in #Read dataset paragraph. - Define mse_data, the score of metrics dataset file. Run Score_table_performanceMeasure.ipynb tree times, one by performance measure: MSE, MAE and R2.
6.- Score_table_sum_performanceMeasure.ipynb: - mse, mae and r2 path. Read dataset. - Run Score_table_sum_performanceMeasure.ipynb.
To compare the model with VST model, follow the following steps for each dataset: 1.- AddingNoiseTo3DMRI.ipynb, you must choose: - repository: the name of the MRI dataset. - Run AddingNoiseTo3DMRI.ipynb.
2.- CreateFilePath.ipynb, you must choose: - base_dir: directory where is the noisy datset. Run CreateFilePath.ipynb
Run VST model: 1.- sigmaValues.m, you must choose:: - path_noiseless: file with path of the original MRI. - path_noise: file with path of the noise MRI. - outputFileName: path to output file.
2.- PerformanceMeasure.ipynb, you must choose: - input_dir1: directory with xlsx files with the sigma values (noise values) predicted by VS model. - inputFileName1: xlsx files with the sigma values (noise values) predicted by VS model. - input_dir2: directory with noise levels. - inputFileName2: file with noise levels, generated by AddingNoiseTo3DMRI.ipynb. - output_dir:
To do an stimation of the SNR: 1.- EstimateSignalToNoiseRatioMRI.ipynb, you must choose: - One dataset. - Run EstimateSignalToNoiseRatioMRI.ipynb.
Citation Please, cite this work as: https://doi.org/10.3390/axioms12121117