Helper Module for Deep Learning.

Module that provides functions to prepare the IMPAC dataset. IMPAC stands for IMaging-PsychiAtry Challenge: predicting autism which is a data challenge on Autism Spectrum Disorder detection: https://paris-saclay-cds.github.io/autism_challenge.

class pynet.datasets.impac.FeatureExtractor[source]¶

Make a transformer which will load the time series and compute the connectome matrix.

__init__()[source]¶

Initialize self. See help(type(self)) for accurate signature.

fit(X_df, y, datadir)[source]¶

transform(X_df, datadir)[source]¶

class pynet.datasets.impac.Item(input_path, output_path, metadata_path, labels, nb_features)¶

property input_path¶

Alias for field number 0

property labels¶

Alias for field number 3

property metadata_path¶

Alias for field number 2

property nb_features¶

Alias for field number 4

property output_path¶

Alias for field number 1

pynet.datasets.impac.fetch_fmri_time_series(outdir, atlas='all')[source]¶

Fetch the time-series extracted from the fMRI data using a specific atlas.

Parameters

outdir: str

the detination folder.

atlas : string, default=’all’

The name of the atlas used during the extraction. The possibilities are: * ‘basc064, ‘basc122’, ‘basc197’: BASC parcellations with 64, 122, and 197 regions [R1]; * ‘craddock_scorr_mean’: Ncuts parcellations [R2]; * ‘harvard_oxford_cort_prob_2mm’: Harvard-Oxford anatomical parcellations; * ‘msdl’: MSDL functional atlas [R3]; * ‘power_2011’: Power atlas [R4].

References

R1(1,2)

Bellec, Pierre, et al. “Multi-level bootstrap analysis of stable clusters in resting-state fMRI.” Neuroimage 51.3 (2010): 1126-1139.

R2(1,2)

Craddock, R. Cameron, et al. “A whole brain fMRI atlas generated via spatially constrained spectral clustering.” Human brain mapping 33.8 (2012): 1914-1928.

R3(1,2)

Varoquaux, Gaël, et al. “Multi-subject dictionary learning to segment an atlas of brain spontaneous activity.” Biennial International Conference on Information Processing in Medical Imaging. Springer, Berlin, Heidelberg, 2011.

R4(1,2)

Power, Jonathan D., et al. “Functional network organization of the human brain.” Neuron 72.4 (2011): 665-678.

pynet.datasets.impac.fetch_impac(datasetdir, mode='train', dtype='all')[source]¶

Fetch/prepare the IMPAC dataset for pynet.

To compute the functional connectivity using the rfMRI data, we use the BASC atlas with 122 ROIs.

Parameters

datasetdir: str

the dataset destination folder.

mode: str

ask the ‘train’ or ‘test’ dataset.

dtype: str, default ‘all’

the features type: ‘anatomy’, ‘fmri’, or ‘all’.

Returns

item: namedtuple

a named tuple containing ‘input_path’, ‘output_path’, and ‘metadata_path’.