Automatic prostate cancer detection through DCE-MRI images: all you need is a good normalization

These pipelines are related to the following article The data used for this research are publicly available on Zenodo

How to use the pipeline?

Data registration

Code compilation

Before to go to data mining and mahcine learning, there is a need to register the DCE-MRI data and the ground-truth. The registration was programed in C++ with the ITK toolbox. You need to compile the code to be able to call the executable.

Therefore, you can compile the code from the root directory as:

$ mkdir build
$ cd build
$ cmake ../src

Two executables will be created in bin/:

• reg_dce: register DCE-MRI data to remove motion during the acquisition.
• reg_gt: register the T2W-MRI, DCE-MRI, and ground-truth data.

Run the executables

You can call the executable ./reg_dce as:

./reg_dce arg1 arg2

• arg1 is the folder with the DCE-MRI data,
• arg2 is the storage folder with the registered DCE-MRI data.

You can call the executable ./reg_gt as:

./reg_gt arg1 arg2 arg3 arg4

• arg1 is the T2W-MRI ground-truth with the segmentation of the prostate.
• arg2 is the DCE-MRI ground-truth with the segmentation of the prostate.
• arg3 is the folder with the DCE-MRI with intra-modality motion correction (see reg_dce),
• arg4 is the storage folder inter-modality motion correction.

Pre-processing pipeline

The following pre-processing routines were applied:

• Standard time normalization.

Addionally, Tofts model requires an estimate of the AIF which can be derived from a population.

Data variables

In the file pipeline/feature-preprocessing/pipeline_normalization_model.py, you need to set the following variables:

• path_patients, path_dce, path_gt, label_gt: the appropriate data folder,
• path_store_model, filename_model: the path where to store the model computed.

In the file pipeline/feature-preprocessing/pipeline_normalization_patient.py, you need to set the following variables:

• path_patients, path_dce, path_gt, label_gt, pt_mdl: the appropriate data folder,
• path_store, filename: the path where to store the normalization object computed.

In the file pipeline/feature-preprocessing/pipeline_average_aif.py and pipeline/feature-preprocessing/pipeline_average_aif_normalized.py, you need to set the following variables:

• path_patients, path_dce, path_gt, label_gt: the appropriate data folder,

Algorithm variables

The variables that can be changed are:

• params: parameters of the standard time normalization. Refer to the help of protoclass.StandardTimeNormalization and the function partial_fit_model and fit.

Run the pipeline

To normalize the DCE MRI data, launch an ipython or python prompt and run from the root directory:

>> run pipeline/feature-preprocessing/pipeline_normalization_model.py
>> run pipeline/feature-preprocessing/pipeline_normalization_patient.py

To compute the population-based AIF, launch an ipython or python prompt and run from the root directory:

>> run pipeline/feature-preprocessing/pipeline_average_aif.py
>> run pipeline/feature-preprocessing/pipeline_average_normalized.py

These two scripts will provide an estimate of the AIF signal which will be used later in Tofts quantification.

Extraction pipeline

The following quantification methods were applied:

• Brix model,
• Hoffmann model,
• PUN model,
• Tofts model,
• Semi-quantitative model.

Data variables

For each extraction method, the following variables need to be set:

• path_patients, path_dce, path_gt, label_gt: the appropriate data folder,
• path_store: the path where to store the normalization object computed.

Additionally, for the normalized data, the following variable need to be set:

• path_norm: the appropriate path to the normalization objects store in the previous section.

Algorithm variables

For the normalized data, you need to set:

• shift: the variable which is the maximum offtsets found during the preprocessing.

For the scripts pipeline/feature-extraction/pipeline_tofts_extraction_population_aif***.py, you need to set:

• aif_params: the parameters fo the population-based AIF found in pre-processing.

Run the pipeline

From the root directory, launch an ipython or python prompt and run:

>> run pipeline/feature-extraction/pipeline_brix_extraction_normalized.py
>> run pipeline/feature-extraction/pipeline_brix_extraction.py
>> run pipeline/feature-extraction/pipeline_hoffmann_extraction_normalized.py
>> run pipeline/feature-extraction/pipeline_hoffmann_extraction.py
>> run pipeline/feature-extraction/pipeline_pun_extraction_normalized.py
>> run pipeline/feature-extraction/pipeline_pun_extraction.py
>> run pipeline/feature-extraction/pipeline_semi_quantitative_normalized.py
>> run pipeline/feature-extraction/pipeline_semi_quantitative.py
>> run pipeline/feature-extraction/pipeline_tofts_extraction_patient_aif_normalized.py
>> run pipeline/feature-extraction/pipeline_tofts_extraction_patient_aif.py
>> run pipeline/feature-extraction/pipeline_tofts_extraction_population_aif_normalized.py
>> run pipeline/feature-extraction/pipeline_tofts_extraction_population_aif.py

Classification pipeline

The following classification methods were applied:

• Random Forest.

Data variables

For un-normalized data, you need to set the following variables:

• path_patients, path_dce, path_gt, label_gt: the appropriate data folder,
• path_store: the path where to store the normalization object computed.

For normalized data, you need to set the following variables:

• path_patients, path_dce, path_gt, label_gt, path_norm: the appropriate data folder,
• path_store: the path where to store the normalization object computed.

Algorithm variables

The variables that can be changed are:

• config: which corresponds to the different classification methods to test.

Run the pipeline

From the root directory, launch an ipython or python prompt and run:

>> run pipeline/feature-classification/pipeline_classifier_unormalized_ese.py
>> run pipeline/feature-classification/pipeline_classifier_normalized_ese.py
>> # Brix model
>> run pipeline/feature-classification/brix/pipeline_classifier_brix_A_normalized.py
>> run pipeline/feature-classification/brix/pipeline_classifier_brix_A.py
>> run pipeline/feature-classification/brix/pipeline_classifier_brix_kel_normalized.py
>> run pipeline/feature-classification/brix/pipeline_classifier_brix_kel.py
>> run pipeline/feature-classification/brix/pipeline_classifier_brix_kep_normalized.py
>> run pipeline/feature-classification/brix/pipeline_classifier_brix_kep.py
>> run pipeline/feature-classification/brix/pipeline_classifier_brix_normalized.py
>> run pipeline/feature-classification/brix/pipeline_classifier_brix.py
>> # Hoffmann model
>> run pipeline/feature-classification/hoffmann/pipeline_classifier_hoffmann_A_normalized.py
>> run pipeline/feature-classification/hoffmann/pipeline_classifier_hoffmann_A.py
>> run pipeline/feature-classification/hoffmann/pipeline_classifier_hoffmann_kel_normalized.py
>> run pipeline/feature-classification/hoffmann/pipeline_classifier_hoffmann_kel.py
>> run pipeline/feature-classification/hoffmann/pipeline_classifier_hoffmann_kep_normalized.py
>> run pipeline/feature-classification/hoffmann/pipeline_classifier_hoffmann_kep.py
>> run pipeline/feature-classification/hoffmann/pipeline_classifier_hoffmann_normalized.py
>> run pipeline/feature-classification/hoffmann/pipeline_classifier_hoffmann.py
>> # PUN model
>> run pipeline/feature-classification/pun/pipeline_classifier_pun_a0_normalized.py
>> run pipeline/feature-classification/pun/pipeline_classifier_pun_a0.py
>> run pipeline/feature-classification/pun/pipeline_classifier_pun_beta_normalized.py
>> run pipeline/feature-classification/pun/pipeline_classifier_pun_beta.py
>> run pipeline/feature-classification/pun/pipeline_classifier_pun_normalized.py
>> run pipeline/feature-classification/pun/pipeline_classifier_pun.py
>> run pipeline/feature-classification/pun/pipeline_classifier_pun_r_normalized.py
>> run pipeline/feature-classification/pun/pipeline_classifier_pun_r.py
>> # Semi-quantitative model
>> run pipeline/feature-classification/semi/pipeline_classifier_semi_quantitative_auc_normalized.py
>> run pipeline/feature-classification/semi/pipeline_classifier_semi_quantitative_auc.py
>> run pipeline/feature-classification/semi/pipeline_classifier_semi_quantitative_normalized.py
>> run pipeline/feature-classification/semi/pipeline_classifier_semi_quantitative.py
>> run pipeline/feature-classification/semi/pipeline_classifier_semi_quantitative_rel_enh_normalized.py
>> run pipeline/feature-classification/semi/pipeline_classifier_semi_quantitative_rel_enh.py
>> run pipeline/feature-classification/semi/pipeline_classifier_semi_quantitative_tau_normalized.py
>> run pipeline/feature-classification/semi/pipeline_classifier_semi_quantitative_tau.py
>> run pipeline/feature-classification/semi/pipeline_classifier_semi_quantitative_washin_normalized.py
>> run pipeline/feature-classification/semi/pipeline_classifier_semi_quantitative_washin.py
>> run pipeline/feature-classification/semi/pipeline_classifier_semi_quantitative_washout_normalized.py
>> run pipeline/feature-classification/semi/pipeline_classifier_semi_quantitative_washout.py
>> # Tofts model
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_ktrans_patient_aif_normalized.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_ktrans_patient_aif.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_ktrans_population_aif_normalized.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_ktrans_population_aif.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_patient_aif_normalized.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_patient_aif.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_population_aif_normalized.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_population_aif.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_ve_patient_aif_normalized.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_ve_patient_aif.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_ve_population_aif_normalized.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_ve_population_aif.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_vp_patient_aif_normalized.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_vp_patient_aif.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_vp_population_aif_normalized.py
>> run pipeline/feature-classification/tofts/pipeline_classifier_tofts_vp_population_aif.py

Validation pipeline

Run the pipeline

To generate the figues in the results directory, launch an ipython or python prompt and run from the root directory:

>> run pipeline/feature-validation/feature_validation_normalisation_classifier.py
>> run pipeline/feature-validation/feature_validation_normalized_pharmaco_params.py
>> run pipeline/feature-validation/feature_validation_unormalized_pharmaco_params.py
>> run pipeline/feature-validation/feature_validation_all_methods_params.py