mri.scripts.gridsearch
mri.scripts.gridsearch#
- class _TestCase(kspace_data, linear_op_class, regularizer_op_class, linear_op_kwargs, regularizer_op_kwargs, optimizer_kwargs)[source]#
Bases:
object
Internal Class to save a test case in a format and call reconstruct
- Parameters
kspace_data (numpy.ndarray) – the kspace data for reconstruction
linear_op_class (class) – linear operator initialization class
linear_op_kwargs (dict) – kwargs for initializing linear operator
regularizer_op_class (class) – regularizer operator initialization class
regularizer_op_kwargs (dict) – kwargs for initializing regularizer operator
optimizer_kwargs (dict) – kwargs for optimizer
- reconstruct_case(fourier_op, reconstructor_class, reconstructor_kwargs, fourier_params=None)[source]#
Internal Function to carry out reconstruction for a special case. This function pulls in appropriate keyword arguments from input and declares appropriate Linear, Fourier and Regularizer Operators. These operators later are used to create the image model by defining the reconstructor. Then the reconstruction is carried and results are returned
- Parameters
fourier_op (object of class FFT) – this defines the fourier operator. for NonCartesianFFT, please make fourier_op as None and pass fourier_params to allow parallel execution
reconstructor_class (class) – reconstructor class
reconstructor_kwargs (dict) – extra kwargs for reconstructor
fourier_params (dict, default None) – holds dictionary with init_class pointing to fourier class to be used and args having keyword arguments for initialization This is passed only if fourier_op is None so that fourier_op can be made on spot during reconstruction. NOTE: We declare fourier operator inside this function to allow parallel execution as NonCartesianFFT cannot be pickled.
- gather_result(metric, results, metric_direction=None)[source]#
Gather the best reconstruction result.
- launch_grid(kspace_data, reconstructor_class, reconstructor_kwargs, fourier_op=None, linear_params=None, regularizer_params=None, optimizer_params=None, compare_metric_details=None, n_jobs=1, verbose=0)[source]#
This function launches off reconstruction for a grid specified through use of kwarg dictionaries.