""" This module contains Ideal bloch operators making simplifying assumptions"""
__all__ = ["InstantaneousPulse", "PerfectSpoiling"]
import tensorflow as tf
import numpy as np
from cmrsim.bloch._base import BaseBlochOperator
# pylint: disable=abstract-method
[docs]
class InstantaneousPulse(BaseBlochOperator):
""" Module designed debug simulation by assuming instananeous excitation with a given flip angle
and slice-profile
:param complex_flip_angle: (n, ) - tf.complex64
:param slice_position:
:param slice_thickness:
:param slice_normal:
:param slice_profile:
:param device:
"""
# pylint: disable=too-many-arguments
def __init__(self, complex_flip_angle: tf.Tensor,
slice_position: tf.Tensor, slice_thickness: float,
slice_normal: tf.Tensor, slice_profile: str = None,
device: str = None):
super().__init__(name="instantaneous_pulse", device=device)
self.flip_angle = tf.Variable(complex_flip_angle, shape=(None, ), dtype=tf.complex64)
self.slice_normal = tf.Variable(slice_normal, shape=(3,), dtype=tf.float32)
self.slice_position = tf.Variable(slice_position, shape=(3,), dtype=tf.float32)
self.slice_thickness = tf.Variable(slice_thickness, shape=(), dtype=tf.float32)
self.n_steps = 1
self.slice_profile = slice_profile
self.n_repetitions = complex_flip_angle.shape[0]
[docs]
def __call__(self, trajectory_module: callable, initial_position: tf.Tensor,
magnetization: tf.Tensor, repetition_index: tf.Tensor = 0, **kwargs):
positions, _ = trajectory_module.increment_particles(initial_position, dt=0.)
with tf.device(self.device):
if self.flip_angle.shape[0] == 1: # override pulse index in case of single FA
repetition_index = 0
flip_angle, phase = self._fa_from_slice_profile(positions, pulse_index=repetition_index)
complex_alpha = tf.complex(flip_angle, 0.) * tf.exp(tf.complex(0., phase))
return self.hard_pulse(complex_alpha, magnetization)
def _fa_from_slice_profile(self, r_vectors: tf.Tensor,
pulse_index: tf.Tensor = tf.constant(0, tf.int32)):
distance_to_slice = tf.einsum("ni, i", r_vectors - self.slice_position[tf.newaxis, :],
self.slice_normal)
normalized_distance = tf.abs(distance_to_slice) / self.slice_thickness
if self.slice_profile == "Sinc1":
profile = np.sinc(normalized_distance)
else:
profile = tf.where(tf.abs(normalized_distance) < 0.5,
tf.ones_like(distance_to_slice), tf.zeros_like(distance_to_slice))
flip_angle = tf.abs(self.flip_angle[pulse_index]) * profile
phase = tf.math.angle(self.flip_angle[pulse_index]) * tf.ones_like(distance_to_slice)
return flip_angle, phase
# pylint: disable=abstract-method
[docs]
class PerfectSpoiling(BaseBlochOperator):
""" Module designed to debug simulations by making residual magnetization effects negligible"""
[docs]
def __call__(self, magnetization: tf.Tensor, **kwargs):
""" Simply sets the transverse magnetization to 0.
:param magnetization: (#batch, 3)
:return: (#batch, 3)
"""
mask_tensor = tf.reshape(tf.constant([0, 0, 1], dtype=tf.complex64), [1, 3])
return magnetization * mask_tensor
