run_dipy_gpu.py

#!/usr/bin/env python

# Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
#    list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
#    this list of conditions and the following disclaimer in the documentation
#    and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
#    contributors may be used to endorse or promote products derived from
#    this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

import os
import argparse
import random
import time
from tqdm import trange, tqdm

import numpy as np

import dipy.reconst.dti as dti
from dipy.io import read_bvals_bvecs
from dipy.io.stateful_tractogram import Origin, Space, StatefulTractogram
from dipy.io.streamline import save_tractogram
from dipy.tracking import utils
from dipy.core.gradients import gradient_table
from dipy.data import small_sphere
#from dipy.direction import BootDirectionGetter
from dipy.reconst.shm import OpdtModel
#from dipy.tracking.local import LocalTracking, ThresholdTissueClassifier
from dipy.reconst import shm

import nibabel as nib
from nibabel.orientations import aff2axcodes

# Import custom module
import cuslines.cuslines as cuslines

t0 = time.time()

# set seed to get deterministic streamlines
np.random.seed(0)
random.seed(0)

#Get Gradient values
def get_gtab(fbval, fbvec):
    bvals, bvecs = read_bvals_bvecs(fbval, fbvec)
    gtab = gradient_table(bvals, bvecs)
    return gtab

def get_img(ep2_seq):
    img = nib.load(ep2_seq)
    return img

print("parsing arguments")
parser = argparse.ArgumentParser()
parser.add_argument("nifti_file", help="path to the ep2multiband sequence nifti file")
parser.add_argument("bvals", help="path to the bvals")
parser.add_argument("bvecs", help="path to the bvecs")
parser.add_argument("mask_nifti", help="path to the mask file")
parser.add_argument("--fa_numpy", default=None, help="path to the FA numpy file")
parser.add_argument("--roi_nifti", default=None, help="path to the ROI file")
parser.add_argument("--output-prefix", type=str, default ='', help="path to the output file")
parser.add_argument("--chunk-size", type=int, required=True, help="how many seeds to process per sweep, per GPU")
parser.add_argument("--sampling-density", type=int, default=3, help="sampling density for seed generation")
parser.add_argument("--ngpus", type=int, required=True, help="number of GPUs to use")
parser.add_argument("--use-fast-write", action='store_true', help="use fast file write")
parser.add_argument("--max-angle", type=float, default=1.0471975511965976, help="default: 60 deg (in rad)")
parser.add_argument("--min-signal", type=float, default=1.0, help="default: 1.0")
parser.add_argument("--tc-threshold", type=float, default=0.1, help="default: 0.1")
parser.add_argument("--step-size", type=float, default=0.5, help="default: 0.5")
args = parser.parse_args()

img = get_img(args.nifti_file)
voxel_order = "".join(aff2axcodes(img.affine))
gtab = get_gtab(args.bvals, args.bvecs)
mask = get_img(args.mask_nifti)
data = img.get_fdata()

# resample mask if necessary
if mask.shape != data.shape:
    from dipy.align.imaffine import AffineMap
    identity = np.eye(4)
    affine_map = AffineMap(identity,
                           img.shape[:3], img.affine,
                           mask.shape[:3], mask.affine)
    mask = affine_map.transform(mask.get_fdata())
    #mask = np.round(mask)
else:
    mask = mask.get_fdata()

# load or compute and save FA file
if (args.fa_numpy is not None) and os.path.isfile(args.fa_numpy):
    FA = np.load(args.fa_numpy, allow_pickle=True)
else:
    # Fit
    tenmodel = dti.TensorModel(gtab, fit_method='WLS')
    print('Fitting Tensor')
    tenfit = tenmodel.fit(data, mask)
    print('Computing anisotropy measures (FA,MD,RGB)')
    FA = tenfit.fa
    FA[np.isnan(FA)] = 0

    if args.fa_numpy is not None:
        np.save(args.fa_numpy, FA)

# Setup tissue_classifier args
#tissue_classifier = ThresholdTissueClassifier(FA, 0.1)
metric_map = np.asarray(FA, 'float64')

# resample roi if necessary
if args.roi_nifti is not None:
    roi = get_img(args.roi_nifti)
    roi_data = roi.get_fdata()
else:
    roi_data = np.zeros(data.shape[:3])
    roi_data[ FA > 0.1 ] = 1.

# Create seeds for ROI
seed_mask = utils.seeds_from_mask(roi_data, density=args.sampling_density, affine=np.eye(4))

# Setup model
print('slowadcodf')
sh_order = 6
model = OpdtModel(gtab, sh_order=sh_order, min_signal=1)

# Setup direction getter args
print('Bootstrap direction getter')
#boot_dg = BootDirectionGetter.from_data(data, model, max_angle=60., sphere=small_sphere)
b0s_mask = gtab.b0s_mask
dwi_mask = ~b0s_mask

# get fit_matrix from model
fit_matrix = model._fit_matrix
delta_b, delta_q = fit_matrix

# setup sampling matrix
sphere = small_sphere
theta = sphere.theta
phi = sphere.phi
sampling_matrix, _, _ = shm.real_sym_sh_basis(sh_order, theta, phi)

## from BootPmfGen __init__
# setup H and R matrices
# TODO: figure out how to get H, R matrices from direction getter object
x, y, z = model.gtab.gradients[dwi_mask].T
r, theta, phi = shm.cart2sphere(x, y, z)
B, _, _ = shm.real_sym_sh_basis(sh_order, theta, phi)
H = shm.hat(B)
R = shm.lcr_matrix(H)

# create floating point copy of data
dataf = np.asarray(data, dtype=float)

print('streamline gen')
global_chunk_size = args.chunk_size * args.ngpus
nchunks = (seed_mask.shape[0] + global_chunk_size - 1) // global_chunk_size

#streamline_generator = LocalTracking(boot_dg, tissue_classifier, seed_mask, affine=np.eye(4), step_size=.5)
gpu_tracker = cuslines.GPUTracker(args.max_angle,
                                  args.min_signal,
                                  args.tc_threshold,
                                  args.step_size,
                                  dataf, H, R, delta_b, delta_q,
                                  b0s_mask.astype(np.int32), metric_map, sampling_matrix,
                                  sphere.vertices, sphere.edges.astype(np.int32),
                                  ngpus=args.ngpus, rng_seed=0)
t1 = time.time()
streamline_time = 0
io_time = 0

# debug start val to fast forward
with tqdm(total=seed_mask.shape[0]) as pbar:
    for idx in range(int(nchunks)):
        # Main streamline computation
        ts = time.time()
        streamlines = gpu_tracker.generate_streamlines(seed_mask[idx*global_chunk_size:(idx+1)*global_chunk_size])
        te = time.time()
        streamline_time += (te-ts)
        pbar.update(seed_mask[idx*global_chunk_size:(idx+1)*global_chunk_size].shape[0])
        #print("Generated {} streamlines from {} seeds, time: {} s".format(len(streamlines),
        #                                                                  seed_mask[idx*global_chunk_size:(idx+1)*global_chunk_size].shape[0],
        #                                                                  te-ts))

        # Save tracklines file
        if args.output_prefix:
            if args.use_fast_write:
                prefix = "{}.{}_{}".format(args.output_prefix, idx+1, nchunks)
                ts = time.time()
                gpu_tracker.dump_streamlines(prefix, voxel_order, roi_data.shape, img.header.get_zooms(), img.affine)
                te = time.time()
                print("Saved streamlines to {}_*.trk, time {} s".format(prefix, te-ts))
            else:
                fname = "{}.{}_{}.trk".format(args.output_prefix, idx+1, nchunks)
                ts = time.time()
                #save_tractogram(fname, streamlines, img.affine, vox_size=roi.header.get_zooms(), shape=roi_data.shape)
                sft = StatefulTractogram(streamlines, args.nifti_file, Space.VOX)
                save_tractogram(sft, fname)
                te = time.time()
                print("Saved streamlines to {}, time {} s".format(fname, te-ts))
            io_time += (te-ts)

pbar.close()
t2 = time.time()

print("Completed processing {} seeds.".format(seed_mask.shape[0]))
print("Initialization time: {} sec".format(t1-t0))
print("Streamline generation total time: {} sec".format(t2-t1))
print("\tStreamline processing: {} sec".format(streamline_time))
if args.output_prefix:
  print("\tFile writing: {} sec".format(io_time))