__all__ = ["OffResonanceProperty", ]
from typing import Union, TYPE_CHECKING
import tensorflow as tf
from cmrsim.analytic.contrast.base import BaseSignalModel
if TYPE_CHECKING:
import numpy as np
[docs]
class OffResonanceProperty(BaseSignalModel):
""" Initializes a module to calculate the phase contribution of an offresonance according to:
.. math::
M_{xy}^{i, \\prime} = M_{xy}^{i} exp(-j\gamma \\Delta B \\delta t),
where :math:`\\delta t` is the duration from the RF-excitation center for GRE sequences
while it is the temporal distance to the echo-center for SE sequences.
.. note::
This module does not expand the repetition axis, but it always broad-casts the internal
calculations to work for all #repetions > 0
:param sampling_times: timing of acquisition (ADC) events in milliseconds
of shape (#reps, #k-space-points)
:param gamma: Gyromagnetic ratio in rad/ms/mT
"""
#: Tuple of names specifying the input quantities that must be passed as keyword-arguments to
#: the __call__ method of the operator instance.
required_quantities = ('deltaB0', )
#: timing of acquisition (ADC) events in milliseconds
#: Assumes the duration from the RF-excitation center for GRE sequences and the temporal
#: distance to the echo-center for SE sequences.
sampling_times: tf.Variable
def __init__(self, sampling_times: Union[tf.Tensor, 'np.ndarray'], gamma: float,
expand_repetitions: bool, **kwargs):
"""
:param sampling_times: timing of acquisition (ADC) events in milliseconds
of shape (#reps, #k-space-points)
:param gamma: Gyromagnetic ratio in rad/ms/mT
"""
super().__init__(expand_repetitions=expand_repetitions, name="offres", **kwargs)
if not(len(sampling_times.shape) == 1 or len(sampling_times.shape) == 2):
raise ValueError("Invalid shape for specified sampling_times. "
"Must either be only (#k-space-points) or "
f"(#repetitions, #k-space-points), but is: {sampling_times.shape}")
if len(sampling_times.shape) == 1:
sampling_times = tf.reshape(sampling_times, [1, -1])
with tf.device(self.device):
self.gamma = tf.constant(gamma, dtype=tf.float32)
self.sampling_times = tf.Variable(tf.constant(sampling_times, tf.float32),
shape=(None, None), dtype=tf.float32)
self.expansion_factor = sampling_times.shape[0]
[docs]
def update(self):
super().update()
self.expansion_factor == tf.shape(self.sampling_times.read_value())[0]
# pylint: disable=signature-differs
[docs]
@tf.function(jit_compile=True)
def __call__(self, signal_tensor: tf.Tensor, deltaB0: tf.Tensor, **kwargs):
""" Evaluates the off-resonance operator under the assumption
:param signal_tensor: (#voxel, #repetitions, #k-space-samples)
#repetitions and #k-space-samples can be1
:param deltaB0: tf.Tensor (#voxel, 1, 1) in mT.
:param segment_index: int, if k-space is segmented for Memory reasons. defaults to 0
:return: signal_tensor (#voxel, #repetition, #k-space-samples) with #k-space-samples > 1
"""
with tf.device(self.device):
input_shape = tf.shape(signal_tensor)
n_samples = tf.shape(self.sampling_times)[-1]
tf.Assert(tf.shape(deltaB0)[1] == 1 or tf.shape(deltaB0)[1] == self.expansion_factor,
["Shape missmatch for input argument T2star, repetitions axis assumed to be 1 or ",
self.expansion_factor, "but revceived ", tf.shape(deltaB0)])
tf.Assert(input_shape[2] == 1 or input_shape[2] == n_samples,
["Shape missmatch for input argument signal_tensor and self.sampling_times",
input_shape, tf.shape(self.sampling_times)])
off_res_phase = self.sampling_times * deltaB0 * self.gamma
phase_factor = tf.exp(-tf.complex(0., off_res_phase)) # (p, self.exp, k)
if self.expand_repetitions or self.expansion_factor == 1:
result = signal_tensor[:, :, tf.newaxis, :] * phase_factor[:, tf.newaxis, :, :]
new_shape = tf.stack([input_shape[0], input_shape[1] * self.expansion_factor,
n_samples], axis=0)
result = tf.reshape(result, new_shape)
else:
tf.Assert(input_shape[1] == self.expansion_factor,
["Shape missmatch for input argument signal_tensor for case no-expand"])
result = signal_tensor * phase_factor
return result
