Source code for AFQ.tasks.segmentation

import nibabel as nib
import os
import os.path as op
from time import time
import numpy as np
import pandas as pd
import logging

import pimms

from AFQ.tasks.decorators import as_file
from AFQ.tasks.utils import get_fname, with_name
import AFQ.segmentation as seg
from AFQ.utils.path import drop_extension
import AFQ.utils.streamlines as aus
from AFQ.tasks.utils import get_default_args
from AFQ.data.s3bids import write_json
import AFQ.api.bundle_dict as abd
import AFQ.utils.streamlines as aus

from dipy.io.streamline import load_tractogram, save_tractogram
from dipy.io.stateful_tractogram import Space
from dipy.stats.analysis import afq_profile, gaussian_weights
from dipy.tracking.streamline import set_number_of_points, values_from_volume


[docs]logger = logging.getLogger('AFQ.api.seg')
@pimms.calc("bundles") @as_file('_tractography.trk', include_track=True, include_seg=True)
[docs]def segment(dwi, data_imap, mapping_imap, tractography_imap, segmentation_params): """ full path to a trk file containing containting segmented streamlines, labeled by bundle Parameters ---------- segmentation_params : dict, optional The parameters for segmentation. Default: use the default behavior of the seg.Segmentation object. """ bundle_dict = data_imap["bundle_dict"] reg_template = data_imap["reg_template"] streamlines = tractography_imap["streamlines"] # We pass `clean_params` here, but do not use it, so we have the # same signature as `_clean_bundles`. img = nib.load(dwi) tg = load_tractogram( streamlines, img, Space.VOX, bbox_valid_check=False) tg.remove_invalid_streamlines() start_time = time() segmentation = seg.Segmentation(**segmentation_params) bundles = segmentation.segment( bundle_dict, tg, dwi, data_imap["bval"], data_imap["bvec"], reg_template=reg_template, mapping=mapping_imap["mapping"]) seg_sft = aus.SegmentedSFT(bundles, Space.VOX) tgram, meta = seg_sft.get_sft_and_sidecar() segmentation_params_out = { arg_name: value if isinstance(value, (int, float, bool, str)) or ( value is None) else str(value) for arg_name, value in segmentation_params.items()} meta["source"] = streamlines meta["Parameters"] = segmentation_params_out meta["Timing"] = time() - start_time return tgram, meta
@pimms.calc("clean_bundles") @as_file('-clean_tractography.trk', include_track=True, include_seg=True)
[docs]def clean_bundles(bundles, data_imap, clean_params=None): """ full path to a trk file containting segmented streamlines, cleaned using the Mahalanobis distance, and labeled by bundle Parameters ---------- clean_params: dict, optional The parameters for cleaning. Default: use the default behavior of the seg.clean_bundle function. """ bundle_dict = data_imap["bundle_dict"] default_clean_params = get_default_args(seg.clean_bundle) if clean_params is not None: for k in clean_params: default_clean_params[k] = clean_params[k] clean_params = default_clean_params seg_sft = aus.SegmentedSFT.fromfile(bundles) start_time = time() bundles = {} for b in bundle_dict.keys(): if b != "whole_brain": idx = seg_sft.bundle_idxs[b] this_tg = seg_sft.get_bundle(b) this_tg = seg.clean_bundle(this_tg, **clean_params) if clean_params['return_idx']: bundles[b] = {} bundles[b]['sl'], bundles[b]['idx'] = this_tg bundles[b]['idx'] = np.array( idx)[bundles[b]['idx']].tolist() else: bundles[b] = this_tg sft, meta = aus.SegmentedSFT( bundles, Space.RASMM).get_sft_and_sidecar() seg_args = get_default_args(seg.clean_bundle) for k in seg_args: if callable(seg_args[k]): seg_args[k] = seg_args[k].__name__ meta["source"] = bundles meta["Parameters"] = seg_args meta["Timing"] = time() - start_time return sft, meta
@pimms.calc("indiv_bundles")
[docs]def export_bundles(base_fname, results_dir, clean_bundles, bundles, data_imap, tracking_params, segmentation_params): """ dictionary of paths, where each path is a full path to a trk file containing the streamlines of a given bundle, cleaned or uncleaned """ bundle_dict = data_imap["bundle_dict"] reg_template = data_imap["reg_template"] if "presegment_bundle_dict" in segmentation_params and\ segmentation_params["presegment_bundle_dict"] is not None\ and not isinstance( segmentation_params["presegment_bundle_dict"], abd.BundleDict): segmentation_params["presegment_bundle_dict"] =\ abd.BundleDict( segmentation_params["presegment_bundle_dict"], seg_algo="afq", resample_to=reg_template) for this_bundles_file, folder in zip([clean_bundles, bundles], ['clean_bundles', 'bundles']): bundles_dir = op.join(results_dir, folder) os.makedirs(bundles_dir, exist_ok=True) seg_sft = aus.SegmentedSFT.fromfile(this_bundles_file) for bundle in bundle_dict: if bundle != "whole_brain": fname = op.split( get_fname( base_fname, f'-{bundle}' f'_tractography.trk', tracking_params=tracking_params, segmentation_params=segmentation_params)) fname = op.join(bundles_dir, fname[1]) logger.info(f"Saving {fname}") save_tractogram( seg_sft.get_bundle(bundle), fname, bbox_valid_check=False) meta = dict(source=this_bundles_file) meta_fname = drop_extension(fname) + '.json' write_json(meta_fname, meta) return True
@pimms.calc("sl_counts") @as_file('_sl_count.csv', include_track=True, include_seg=True)
[docs]def export_sl_counts(data_imap, clean_bundles, bundles): """ full path to a JSON file containing streamline counts """ bundle_dict = data_imap["bundle_dict"] sl_counts_clean = [] sl_counts = [] bundle_names = list(bundle_dict.keys()) if "whole_brain" not in bundle_names: bundle_names.append("whole_brain") bundles_files = [clean_bundles, bundles] lists = [sl_counts_clean, sl_counts] for bundles_file, count in zip(bundles_files, lists): seg_sft = aus.SegmentedSFT.fromfile(bundles_file) for bundle in bundle_names: if bundle == "whole_brain": count.append(len(seg_sft.sft.streamlines)) else: count.append(len( seg_sft.get_bundle(bundle).streamlines)) counts_df = pd.DataFrame( data=dict( n_streamlines=sl_counts, n_streamlines_clean=sl_counts_clean), index=bundle_names) return counts_df, dict(sources=bundles_files)
@pimms.calc("profiles") @as_file('_profiles.csv', include_track=True, include_seg=True)
[docs]def tract_profiles(clean_bundles, data_imap, scalar_dict, dwi_affine, profile_weights="gauss"): """ full path to a CSV file containing tract profiles Parameters ---------- profile_weights : str, 1D array, 2D array callable, optional How to weight each streamline (1D) or each node (2D) when calculating the tract-profiles. If callable, this is a function that calculates weights. If None, no weighting will be applied. If "gauss", gaussian weights will be used. If "median", the median of values at each node will be used instead of a mean or weighted mean. Default: "gauss" """ bundle_dict = data_imap["bundle_dict"] if not (profile_weights is None or isinstance(profile_weights, str) or callable(profile_weights) or hasattr(profile_weights, "__len__")): raise TypeError( "profile_weights must be string, None, callable, or" + "a 1D or 2D array") if isinstance(profile_weights, str): profile_weights = profile_weights.lower() if isinstance(profile_weights, str) and\ profile_weights != "gauss" and profile_weights != "median": raise TypeError( "if profile_weights is a string," + " it must be 'gauss' or 'median'") bundle_names = [] node_numbers = [] profiles = np.empty((len(scalar_dict), 0)).tolist() this_profile = np.zeros((len(scalar_dict), 100)) seg_sft = aus.SegmentedSFT.fromfile( clean_bundles) seg_sft.sft.to_rasmm() for bundle_name in bundle_dict.keys(): this_sl = seg_sft.get_bundle(bundle_name).streamlines if len(this_sl) == 0: continue for ii, (scalar, scalar_file) in enumerate(scalar_dict.items()): scalar_data = nib.load(scalar_file).get_fdata() if isinstance(profile_weights, str): if profile_weights == "gauss": this_prof_weights = gaussian_weights(this_sl) elif profile_weights == "median": # weights bundle to only return the mean def _median_weight(bundle): fgarray = set_number_of_points(bundle, 100) values = np.array( values_from_volume( scalar_data, fgarray, dwi_affine)) weights = np.zeros(values.shape) for ii, jj in enumerate( np.argsort(values, axis=0)[ len(values) // 2, :]): weights[jj, ii] = 1 return weights this_prof_weights = _median_weight else: this_prof_weights = profile_weights this_profile[ii] = afq_profile( scalar_data, this_sl, dwi_affine, weights=this_prof_weights) profiles[ii].extend(list(this_profile[ii])) nodes = list(np.arange(this_profile[0].shape[0])) bundle_names.extend([bundle_name] * len(nodes)) node_numbers.extend(nodes) profile_dict = dict() profile_dict["tractID"] = bundle_names profile_dict["nodeID"] = node_numbers for ii, scalar in enumerate(scalar_dict.keys()): profile_dict[scalar] = profiles[ii] profile_dframe = pd.DataFrame(profile_dict) meta = dict(source=clean_bundles, parameters=get_default_args(afq_profile)) return profile_dframe, meta
@pimms.calc("scalar_dict")
[docs]def get_scalar_dict(data_imap, mapping_imap, scalars=["dti_fa", "dti_md"]): """ dicionary mapping scalar names to their respective file paths Parameters ---------- scalars : list of strings and/or scalar definitions, optional List of scalars to use. Can be any of: "dti_fa", "dti_md", "dki_fa", "dki_md", "dki_awf", "dki_mk". Can also be a scalar from AFQ.definitions.image. Default: ["dti_fa", "dti_md"] """ # Note: some scalars preprocessing done in plans, before this step scalar_dict = {} for scalar in scalars: if isinstance(scalar, str): sc = scalar.lower() scalar_dict[sc] = data_imap[f"{sc}"] else: scalar_dict[scalar.get_name()] = mapping_imap[ f"{scalar.get_name()}"] return {"scalar_dict": scalar_dict}
[docs]def get_segmentation_plan(kwargs): if "segmentation_params" in kwargs\ and not isinstance(kwargs["segmentation_params"], dict): raise TypeError( "segmentation_params a dict") segmentation_tasks = with_name([ get_scalar_dict, export_sl_counts, export_bundles, clean_bundles, segment, tract_profiles]) default_seg_params = get_default_args(seg.Segmentation.__init__) if "segmentation_params" in kwargs: for k in kwargs["segmentation_params"]: default_seg_params[k] = kwargs["segmentation_params"][k] kwargs["segmentation_params"] = default_seg_params return pimms.plan(**segmentation_tasks)