preprocess.py

import os
from pathlib import Path

from nipype.interfaces.base import (
    InputMultiPath,
    TraitedSpec,
    traits,
    isdefined,
    File,
    Str,
)
from nipype.interfaces.cat12.base import Cell

from nipype.interfaces.spm import SPMCommand
from nipype.interfaces.spm.base import (
    SPMCommandInputSpec,
    ImageFileSPM,
    scans_for_fnames,
    scans_for_fname,
)
from nipype.utils.filemanip import split_filename, fname_presuffix


class CAT12SegmentInputSpec(SPMCommandInputSpec):
    in_files = InputMultiPath(
        ImageFileSPM(exists=True),
        field="data",
        desc="file to segment",
        mandatory=True,
        copyfile=False,
    )

    _help_tpm = (
        "Tissue Probability Map. Select the tissue probability image that includes 6 tissue probability "
        "classes for (1) grey matter, (2) white matter, (3) cerebrospinal fluid, (4) bone, (5) non-brain  "
        "soft tissue, and (6) the background.  CAT uses the TPM only for the initial SPM segmentation."
    )
    tpm = InputMultiPath(
        ImageFileSPM(exists=True),
        field="tpm",
        desc=_help_tpm,
        mandatory=False,
        copyfile=False,
    )

    _help_shoots_tpm = (
        "Shooting Template %d.  The Shooting template must be in multi-volume nifti format and should contain GM,"
        " WM, and background segmentations and have to be saved with at least 16 bit. "
    )

    shooting_tpm = ImageFileSPM(
        exists=True,
        field="extopts.registration.shooting.shootingtpm",
        desc=_help_shoots_tpm % 0,
        mandatory=False,
        copyfile=False,
    )

    shooting_tpm_template_1 = ImageFileSPM(
        exists=True, desc=_help_shoots_tpm % 1, mandatory=False, copyfile=False
    )
    shooting_tpm_template_2 = ImageFileSPM(
        exists=True, desc=_help_shoots_tpm % 2, mandatory=False, copyfile=False
    )
    shooting_tpm_template_3 = ImageFileSPM(
        exists=True, desc=_help_shoots_tpm % 3, mandatory=False, copyfile=False
    )
    shooting_tpm_template_4 = ImageFileSPM(
        exists=True, desc=_help_shoots_tpm % 4, mandatory=False, copyfile=False
    )

    n_jobs = traits.Int(
        1, usedefault=True, mandatory=True, field="nproc", desc="Number of threads"
    )

    _help_affine_reg = (
        "Affine Regularization. The procedure is a local optimisation, so it needs reasonable initial "
        "starting estimates. Images should be placed in approximate alignment using the Display "
        "function of SPM before beginning.  A Mutual Information affine registration with the tissue "
        "probability maps (D"
        "Agostino et al, 2004) is used to achieve approximate alignment."
    )
    affine_regularization = Str(
        default_value="mni", field="opts.affreg", usedefault=True, desc=_help_affine_reg
    )

    _help_bias_acc = (
        "Strength of the SPM inhomogeneity (bias) correction that simultaneously controls the SPM "
        "biasreg, biasfwhm, samp (resolution), and tol (iteration) parameter."
    )
    power_spm_inhomogeneity_correction = traits.Float(
        default_value=0.5, field="opts.biasacc", usedefault=True, desc=_help_bias_acc
    )
    # Extended options for CAT12 preprocessing
    _help_app = (
        "Affine registration and SPM preprocessing can fail in some subjects with deviating anatomy (e.g. "
        "other species/neonates) or in images with strong signal inhomogeneities, or untypical intensities "
        "(e.g. synthetic images). An initial bias correction can help to reduce such problems (see details "
        'below). Recommended are the "default" and "full" option.'
    )
    affine_preprocessing = traits.Int(
        1070, field="extopts.APP", desc=_help_app, usedefault=True
    )

    _help_initial_seg = (
        "In rare cases the Unified Segmentation can fail in highly abnormal brains, where e.g. the "
        "cerebrospinal fluid of superlarge ventricles (hydrocephalus) were classified as white "
        "matter. However, if the affine registration is correct, the AMAP segmentation with an "
        "prior-independent k-means initialization can be used to replace the SPM brain tissue "
        "classification. Moreover, if the default Dartel and Shooting registrations will fail then "
        'rhe "Optimized Shooting - superlarge ventricles" option for "Spatial registration" is ! '
        "required Values: \nnone: 0;\nlight: 1;\nfull: 2;\ndefault: 1070."
    )
    initial_segmentation = traits.Int(
        0, field="extopts.spm_kamap", desc=_help_initial_seg, usedefault=True
    )

    _help_las = (
        "Additionally to WM-inhomogeneities, GM intensity can vary across different regions such as the motor"
        " cortex, the basal ganglia, or the occipital lobe. These changes have an anatomical background "
        "(e.g. iron content, myelinization), but are dependent on the MR-protocol and often lead to "
        "underestimation of GM at higher intensities and overestimation of CSF at lower intensities. "
        "Therefore, a local intensity transformation of all tissue classes is used to reduce these effects in"
        " the image. This local adaptive segmentation (LAS) is applied before the final AMAP segmentation."
        "Possible Values: \nSPM Unified Segmentation: 0 \nk-means AMAP: 2"
    )
    local_adaptive_seg = traits.Float(
        0.5, field="extopts.LASstr", usedefault=True, desc=_help_las
    )

    _help_gcutstr = (
        "Method of initial skull-stripping before AMAP segmentation. The SPM approach works quite stable "
        "for the majority of data. However, in some rare cases parts of GM (i.e. in frontal lobe) might "
        "be cut. If this happens the GCUT approach is a good alternative. GCUT is a graph-cut/region-"
        "growing approach starting from the WM area. APRG (adaptive probability region-growing) is a new"
        " method that refines the probability maps of the SPM approach by region-growing techniques of "
        "the gcut approach with a final surface-based optimization strategy. This is currently the method"
        " with the most accurate and reliable results. If you use already skull-stripped data you can "
        "turn off skull-stripping although this is automaticaly detected in most cases. Please note that "
        "the choice of the skull-stripping method will also influence the estimation of TIV, because the"
        " methods mainly differ in the handling of the outer CSF around the cortical surface. "
        "\nPossible Values:\n - none (already skull-stripped): -1;\n - SPM approach: 0; "
        "\n - GCUT approach: 0.50; \n - APRG approach: 2"
    )
    skull_strip = traits.Float(
        2, field="extopts.gcutstr", desc=_help_gcutstr, usedefault=True
    )

    _help_wmhc = (
        "WARNING: Please note that the detection of WM hyperintensies is still under development and does "
        "not have the same accuracy as approaches that additionally consider FLAIR images (e.g. Lesion "
        "Segmentation Toolbox)! In aging or (neurodegenerative) diseases WM intensity can be reduced "
        "locally in T1 or increased in T2/PD images. These so-called WM hyperintensies (WMHs) can lead to "
        "preprocessing errors. Large GM areas next to the ventricle can cause normalization problems. "
        "Therefore, a temporary correction for normalization is useful if WMHs are expected. CAT allows "
        "different ways to handle WMHs: "
        "\n0) No Correction (handled as GM). \n1) Temporary (internal) correction as WM for spatial "
        "normalization and estimation of cortical thickness. \n2) Permanent correction to WM. "
    )
    wm_hyper_intensity_correction = traits.Int(
        1, field="extopts.WMHC", desc=_help_wmhc, usedefault=True
    )

    _help_vox = (
        "The (isotropic) voxel sizes of any spatially normalised written images. A non-finite value will be "
        "replaced by the average voxel size of the tissue probability maps used by the segmentation."
    )
    voxel_size = traits.Float(1.5, field="extopts.vox", desc=_help_vox, usedefault=True)

    _help_resampling = (
        "Internal resampling for preprocessing.\n The default fixed image resolution offers a good "
        "trade-off between optimal quality and preprocessing time and memory demands. Standard "
        "structural data with a voxel resolution around 1mm or even data with high in-plane resolution"
        " and large slice thickness (e.g. 0.5x0.5x1.5 mm) will benefit from this setting. If you have"
        ' higher native resolutions the highres option "Fixed 0.8 mm" will sometimes offer slightly'
        " better preprocessing quality with an increase of preprocessing time and memory demands. In"
        " case of even higher resolutions and high signal-to-noise ratio (e.g. for 7 T data) the "
        '"Best native" option will process the data on the highest native resolution. A resolution'
        " of 0.4x0.7x1.0 mm will be interpolated to 0.4x0.4x0.4 mm. A tolerance range of 0.1 mm is "
        "used to avoid interpolation artifacts, i.e. a resolution of 0.95x1.01x1.08 mm will not be "
        'interpolated in case of the "Fixed 1.0 mm"! This "optimal" option prefers an isotropic voxel '
        "size with at least 1.1 mm that is controlled by the median voxel size and a volume term that "
        "penalizes highly anisotropic voxels."
        "Values:\nOptimal: [1.0 0.1]\nFixed 1.0 mm: [1.0 0.1];\nFixed 0.8 mm:[0.8 0.1]"
        "\nBest native: [0.5 0.1]"
    )
    internal_resampling_process = traits.Tuple(
        traits.Float(1),
        traits.Float(0.1),
        minlen=2,
        maxlen=2,
        usedefault=True,
        field="extopts.restypes.optimal",
        desc="help_resampling",
    )
    _errors_help = (
        "Error handling.\nTry to catch preprocessing errors and continue with the next data set or ignore "
        "all warnings (e.g., bad intensities) and use an experimental pipeline which is still in "
        "development. In case of errors, CAT continues with the next subject if this option is enabled. If "
        "the experimental option with backup functions is selected and warnings occur, CAT will try to use"
        " backup routines and skip some processing steps which require good T1 contrasts (e.g., LAS). If "
        "you want to avoid processing of critical data and ensure that only the main pipeline is used then"
        ' select the option "Ignore errors (continue with the next subject)". It is strongly recommended '
        "to check for preprocessing problems, especially with non-T1 contrasts. "
        "\nValues:\nnone: 0,\ndefault: 1,\ndetails: 2."
    )
    ignore_errors = traits.Int(
        1, field="extopts.ignoreErrors", desc=_errors_help, usedefault=True
    )

    # Writing options
    _help_surf = (
        "Surface and thickness estimation. \nUse projection-based thickness (PBT) (Dahnke et al. 2012) to"
        " estimate cortical thickness and to create the central cortical surface for left and right "
        "hemisphere. Surface reconstruction includes topology correction (Yotter et al. 2011), spherical "
        "inflation (Yotter et al.) and spherical registration. Additionally you can also estimate surface "
        "parameters such as gyrification, cortical complexity or sulcal depth that can be subsequently "
        "analyzed at each vertex of the surface. Please note, that surface reconstruction and spherical "
        "registration additionally requires about 20-60 min of computation time. A fast (1-3 min) surface "
        "pipeline is available for visual preview (e.g., to check preprocessing quality) in the "
        "cross-sectional, but not in the longitudinal pipeline.  Only the initial surfaces are created with "
        "a lower resolution and without topology correction, spherical mapping and surface registration. "
        "Please note that the files with the estimated surface thickness can therefore not be used for "
        'further analysis!  For distinction, these files contain "preview" in their filename and they'
        " are not available as batch dependencies objects. "
    )
    surface_and_thickness_estimation = traits.Int(
        1, field="surface", desc=_help_surf, usedefault=True
    )
    surface_measures = traits.Int(
        1,
        field="output.surf_measures",
        usedefault=True,
        desc="Extract surface measures",
    )

    # Templates
    neuromorphometrics = traits.Bool(
        True,
        field="output.ROImenu.atlases.neuromorphometrics",
        usedefault=True,
        desc="Extract brain measures for Neuromorphometrics template",
    )
    lpba40 = traits.Bool(
        True,
        field="output.ROImenu.atlases.lpba40",
        usedefault=True,
        desc="Extract brain measures for LPBA40 template",
    )
    cobra = traits.Bool(
        True,
        field="output.ROImenu.atlases.hammers",
        usedefault=True,
        desc="Extract brain measures for COBRA template",
    )
    hammers = traits.Bool(
        True,
        field="output.ROImenu.atlases.cobra",
        usedefault=True,
        desc="Extract brain measures for Hammers template",
    )
    own_atlas = InputMultiPath(
        ImageFileSPM(exists=True),
        field="output.ROImenu.atlases.ownatlas",
        desc="Extract brain measures for a given template",
        mandatory=False,
        copyfile=False,
    )

    # Grey matter
    gm_output_native = traits.Bool(
        False,
        field="output.GM.native",
        usedefault=True,
        desc="Save modulated grey matter images.",
    )
    gm_output_modulated = traits.Bool(
        True,
        field="output.GM.mod",
        usedefault=True,
        desc="Save native grey matter images.",
    )
    gm_output_dartel = traits.Bool(
        False,
        field="output.GM.dartel",
        usedefault=True,
        desc="Save dartel grey matter images.",
    )

    # White matter
    _wm_desc = "Options to save white matter images."
    wm_output_native = traits.Bool(
        False,
        field="output.WM.native",
        usedefault=True,
        desc="Save dartel white matter images.",
    )
    wm_output_modulated = traits.Bool(
        True,
        field="output.WM.mod",
        usedefault=True,
        desc="Save dartel white matter images.",
    )
    wm_output_dartel = traits.Bool(
        False,
        field="output.WM.dartel",
        usedefault=True,
        desc="Save dartel white matter images.",
    )

    # CSF matter
    _csf_desc = "Options to save CSF images."
    csf_output_native = traits.Bool(
        False,
        field="output.CSF.native",
        usedefault=True,
        desc="Save dartel CSF images.",
    )
    csf_output_modulated = traits.Bool(
        True, field="output.CSF.mod", usedefault=True, desc="Save dartel CSF images."
    )
    csf_output_dartel = traits.Bool(
        False,
        field="output.CSF.dartel",
        usedefault=True,
        desc="Save dartel CSF images.",
    )

    # Labels
    _help_label_desc = (
        "This is the option to save a labeled version of your segmentations in the %s space for fast visual "
        "comparision. Labels are saved as Partial Volume Estimation (PVE) values with different mix "
        "classes for GM-WM (2.5) and GM-CSF (1.5). BG=0, CSF=1, GM=2, WM=3, WMH=4 (if WMHC=3), "
        "SL=1.5 (if SLC)"
    )
    label_native = traits.Bool(
        False,
        field="output.label.native",
        usedefault=True,
        desc=_help_label_desc % "native",
    )
    label_warped = traits.Bool(
        True,
        field="output.label.warped",
        usedefault=True,
        desc=_help_label_desc % "warped",
    )
    label_dartel = traits.Bool(
        False,
        field="output.label.dartel",
        usedefault=True,
        desc=_help_label_desc % "dartel",
    )
    output_labelnative = traits.Bool(
        False,
        field="output.labelnative",
        usedefault=True,
        desc=_help_label_desc % "native",
    )

    # Bias
    save_bias_corrected = traits.Bool(
        True,
        field="output.bias.warped",
        usedefault=True,
        desc="Save bias corrected image",
    )

    # las
    _las_desc = (
        "This is the option to save a bias, noise, and local intensity corrected version of the original T1"
        " image in the %s space. MR images are usually corrupted by a smooth, spatially varying artifact that modulates the"
        " intensity of the image (bias). These artifacts, although not usually a problem for visual "
        "inspection, can impede automated processing of the images. The bias corrected version should have "
        "more uniform intensities within the different types of tissues and can be saved in native space "
        "and/or normalised. Noise is corrected by an adaptive non-local mean (NLM) filter (Manjon 2008, "
        "Medical Image Analysis 12)."
    )
    las_native = traits.Bool(
        False, field="output.las.native", usedefault=True, desc=_las_desc % "native"
    )
    las_warped = traits.Bool(
        True, field="output.las.warped", usedefault=True, desc=_las_desc % "warped"
    )
    las_dartel = traits.Bool(
        False, field="output.las.dartel", usedefault=True, desc=_las_desc % "dartel"
    )

    # Jacobian Warped
    _help_jacobian = (
        "This is the option to save the Jacobian determinant, which expresses local volume changes. This"
        " image can be used in a pure deformation based morphometry (DBM) design. Please note that the"
        " affine part of the deformation field is ignored. Thus, there is no need for any additional"
        " correction for different brain sizes using ICV."
    )
    jacobianwarped = traits.Bool(
        True, field="output.jacobianwarped", usedefault=True, desc=_help_jacobian
    )

    # Deformation Fields
    _help_warp = (
        "Deformation fields can be saved to disk, and used by the Deformations Utility and/or applied to "
        "coregistered data from other modalities (e.g. fMRI). For spatially normalising images to MNI space,"
        " you will need the forward deformation, whereas for spatially normalising (eg) GIFTI surface files,"
        " you"
        "ll need the inverse. It is also possible to transform data in MNI space on to the individual"
        " subject, which also requires the inverse transform. Deformations are saved as .nii files, which"
        " contain three volumes to encode the x, y and z coordinates."
        "\nValues: No:[0 0];\nImage->Template (forward): [1 0];\nTemplate->Image (inverse): [0 1]; "
        "\ninverse + forward: [1 1]"
    )
    warps = traits.Tuple(
        traits.Int(1),
        traits.Int(0),
        minlen=2,
        maxlen=2,
        field="output.warps",
        usedefault=True,
        desc=_help_warp,
    )


class CAT12SegmentOutputSpec(TraitedSpec):
    ##########################################
    # Label XML files
    ##########################################
    label_files = traits.List(
        File(exists=True), desc="Files with the measures extracted for OI ands ROIs"
    )

    label_rois = File(exists=True, desc="Files with thickness values of ROIs.")
    label_roi = File(exists=True, desc="Files with thickness values of ROI.")

    ##########################################
    # MRI .nii files
    ##########################################

    mri_images = traits.List(File(exists=True), desc="Different segmented images.")

    # Grey Matter
    gm_modulated_image = File(exists=True, desc="Grey matter modulated image.")
    gm_dartel_image = File(exists=True, desc="Grey matter dartel image.")
    gm_native_image = File(exists=True, desc="Grey matter native space.")

    # White Matter
    wm_modulated_image = File(exists=True, desc="White matter modulated image.")
    wm_dartel_image = File(exists=True, desc="White matter dartel image.")
    wm_native_image = File(exists=True, desc="White matter in native space.")

    # CSF
    csf_modulated_image = File(exists=True, desc="CSF modulated image.")
    csf_dartel_image = File(exists=True, desc="CSF dartel image.")
    csf_native_image = File(exists=True, desc="CSF in native space.")

    bias_corrected_image = File(exists=True, desc="Bias corrected image")
    ##########################################
    # Surface files
    ##########################################

    surface_files = traits.List(File(exists=True), desc="Surface files")

    # Right hemisphere
    rh_central_surface = File(exists=True, desc="Central right hemisphere files")
    rh_sphere_surface = File(exists=True, desc="Sphere right hemisphere files")

    # Left hemisphere
    lh_central_surface = File(exists=True, desc="Central left hemisphere files")
    lh_sphere_surface = File(exists=True, desc="Sphere left hemisphere files")

    # Report files
    report_files = traits.List(File(exists=True), desc="Report files.")
    report = File(exists=True, desc="Report file.")


class CAT12Segment(SPMCommand):
    """
    CAT12: Segmentation

    This toolbox is an extension to the default segmentation in SPM12, but uses a completely different segmentation
    approach.
    The segmentation approach is based on an Adaptive Maximum A Posterior (MAP) technique without the need for a priori
    information about tissue probabilities. That is, the Tissue Probability Maps (TPM) are not used constantly in the
    sense of the classical Unified Segmentation approach (Ashburner et. al. 2005), but just for spatial normalization.
    The following AMAP estimation is adaptive in the sense that local variations of the parameters (i.e., means and
    variance) are modeled as slowly varying spatial functions (Rajapakse et al. 1997). This not only accounts for
    intensity inhomogeneities but also for other local variations of intensity.
    Additionally, the segmentation approach uses a Partial Volume Estimation (PVE) with a simplified mixed model of at
    most two tissue types (Tohka et al. 2004). We start with an initial segmentation into three pure classes: gray
    matter (GM), white matter (WM), and cerebrospinal fluid (CSF) based on the above described AMAP estimation. The
    initial segmentation is followed by a PVE of two additional mixed classes: GM-WM and GM-CSF. This results in an
    estimation of the amount (or fraction) of each pure tissue type present in every voxel (as single voxels - given by
    Another important extension to the SPM12 segmentation is the integration of the Dartel or Geodesic Shooting
    registration into the toolbox by an already existing Dartel/Shooting template in MNI space. This template was
    derived from 555 healthy control subjects of the IXI-database (http://www.brain-development.org) and provides the
    several Dartel or Shooting iterations. Thus, for the majority of studies the creation of sample-specific templates
    is not necessary anymore and is mainly recommended for children data.'};

    http://www.neuro.uni-jena.de/cat12/CAT12-Manual.pdf#page=15

    Examples
    --------
    >>> path_mr = 'structural.nii'
    >>> cat = CAT12Segment(in_files=path_mr)
    >>> cat.run() # doctest: +SKIP
    """

    input_spec = CAT12SegmentInputSpec
    output_spec = CAT12SegmentOutputSpec

    def __init__(self, **inputs):
        _local_version = SPMCommand().version
        if _local_version and "12." in _local_version:
            self._jobtype = "tools"
            self._jobname = "cat.estwrite"

        SPMCommand.__init__(self, **inputs)

    def _format_arg(self, opt, spec, val):
        """Convert input to appropriate format for spm"""
        if opt == "in_files":
            if isinstance(val, list):
                return scans_for_fnames(val)
            else:
                return scans_for_fname(val)
        elif opt in ["tpm", "shooting_tpm"]:
            return Cell2Str(val)

        return super(CAT12Segment, self)._format_arg(opt, spec, val)

    def _list_outputs(self):
        outputs = self._outputs().get()
        f = self.inputs.in_files[0]
        pth, base, ext = split_filename(f)

        outputs["mri_images"] = [
            str(mri) for mri in Path(pth).glob("mri/*") if mri.is_file()
        ]

        for tidx, tissue in enumerate(["gm", "wm", "csf"]):

            for image, prefix in [("modulated", "mw"), ("dartel", "r"), ("native", "")]:
                outtype = f"{tissue}_output_{image}"
                if isdefined(getattr(self.inputs, outtype)) and getattr(
                    self.inputs, outtype
                ):
                    outfield = f"{tissue}_{image}_image"
                    prefix = os.path.join("mri", f"{prefix}p{tidx + 1}")
                    if image != "dartel":
                        outputs[outfield] = fname_presuffix(f, prefix=prefix)
                    else:
                        outputs[outfield] = fname_presuffix(
                            f, prefix=prefix, suffix="_rigid"
                        )

        if self.inputs.save_bias_corrected:
            outputs["bias_corrected_image"] = fname_presuffix(
                f, prefix=os.path.join("mri", "wmi")
            )

        outputs["surface_files"] = [
            str(surf) for surf in Path(pth).glob("surf/*") if surf.is_file()
        ]

        for hemisphere in ["rh", "lh"]:
            for suffix in ["central", "sphere"]:
                outfield = f"{hemisphere}_{suffix}_surface"
                outputs[outfield] = fname_presuffix(
                    f,
                    prefix=os.path.join("surf", f"{hemisphere}.{suffix}."),
                    suffix=".gii",
                    use_ext=False,
                )

        outputs["report_files"] = outputs["report_files"] = [
            str(report) for report in Path(pth).glob("report/*") if report.is_file()
        ]

        outputs["report"] = fname_presuffix(
            f, prefix=os.path.join("report", f"cat_"), suffix=".xml", use_ext=False
        )

        outputs["label_files"] = [
            str(label) for label in Path(pth).glob("label/*") if label.is_file()
        ]

        outputs["label_rois"] = fname_presuffix(
            f, prefix=os.path.join("label", "catROIs_"), suffix=".xml", use_ext=False
        )
        outputs["label_roi"] = fname_presuffix(
            f, prefix=os.path.join("label", "catROI_"), suffix=".xml", use_ext=False
        )

        return outputs


class CAT12SANLMDenoisingInputSpec(SPMCommandInputSpec):

    in_files = InputMultiPath(
        ImageFileSPM(exists=True),
        field="data",
        desc="Images for filtering.",
        mandatory=True,
        copyfile=False,
    )

    spm_type = traits.Enum(
        "float32",
        "uint16",
        "uint8",
        "same",
        field="spm_type",
        usedefault=True,
        desc="Data type of the output images. 'same' matches the input image type.",
    )

    intlim = traits.Int(
        field="intlim",
        default_value=100,
        usedefault=True,
        desc="intensity limitation (default = 100)",
    )

    filename_prefix = traits.Str(
        field="prefix",
        default_value="sanlm_",
        usedefault=True,
        desc="Filename prefix. Specify  the  string  to be prepended to the filenames of the filtered image file(s).",
    )

    filename_suffix = traits.Str(
        field="suffix",
        default_value="",
        usedefault=True,
        desc="Filename suffix. Specify  the  string  to  be  appended  to the filenames of the filtered image file(s).",
    )

    addnoise = traits.Float(
        default_value=0.5,
        usedefault=True,
        field="addnoise",
        desc="""Strength of additional noise in noise-free regions.
        Add  minimal  amount  of noise in regions without any noise to avoid image segmentation problems.
        This parameter defines the strength of additional noise as percentage of the average signal intensity.""",
    )

    rician = traits.Bool(
        True,
        field="rician",
        usedefault=True,
        desc="""Rician noise
        MRIs  can  have  Gaussian  or  Rician  distributed  noise with uniform or nonuniform variance across the image.
        If SNR is high enough (>3)  noise  can  be  well  approximated by Gaussian noise in the foreground. However, for
        SENSE reconstruction or DTI data a Rician distribution is expected. Please note that the Rician noise estimation
        is sensitive for large signals in the neighbourhood and can lead to artefacts, e.g. cortex can be affected by
        very high values in the scalp or in blood vessels.""",
    )

    replace_nan_and_inf = traits.Bool(
        True,
        field="replaceNANandINF",
        usedefault=True,
        desc="Replace NAN by 0, -INF by the minimum and INF by the maximum of the image.",
    )

    noisecorr_strength = traits.Enum(
        "-Inf",
        2,
        4,
        field="nlmfilter.optimized.NCstr",
        usedefault=True,
        desc="""Strength of Noise Corrections
        Strength  of  the  (sub-resolution)  spatial  adaptive    non local means (SANLM) noise correction. Please note
        that the filter strength is automatically  estimated.  Change this parameter only for specific conditions. The
        "light" option applies half of the filter strength of the adaptive  "medium"  cases,  whereas  the  "strong"
        option  uses  the  full  filter  strength,  force sub-resolution filtering and applies an additional  iteration.
        Sub-resolution  filtering  is  only  used  in  case  of  high image resolution below 0.8 mm or in case of the
        "strong" option. light = 2, medium = -Inf, strong = 4""",
    )


class CAT12SANLMDenoisingOutputSpec(TraitedSpec):

    out_file = File(desc="out file")


class CAT12SANLMDenoising(SPMCommand):
    """
    Spatially adaptive non-local means (SANLM) denoising filter

    This  function  applies  an spatial adaptive (sub-resolution) non-local means denoising filter
    to  the  data.  This  filter  will  remove  noise  while  preserving  edges. The filter strength is
    automatically estimated based on the standard deviation of the noise.

    This   filter   is  internally  used  in  the  segmentation  procedure  anyway.  Thus,  it  is  not
    necessary (and not recommended) to apply the filter before segmentation.
    ______________________________________________________________________
    Christian Gaser, Robert Dahnke
    Structural Brain Mapping Group (http://www.neuro.uni-jena.de)
    Departments of Neurology and Psychiatry
    Jena University Hospital
    ______________________________________________________________________

    Examples
    --------
    >>> from nipype.interfaces import cat12
    >>> c = cat12.CAT12SANLMDenoising()
    >>> c.inputs.in_files = 'anatomical.nii'
    >>> c.run() # doctest: +SKIP
    """

    input_spec = CAT12SANLMDenoisingInputSpec
    output_spec = CAT12SANLMDenoisingOutputSpec

    def __init__(self, **inputs):
        _local_version = SPMCommand().version
        if _local_version and "12." in _local_version:
            self._jobtype = "tools"
            self._jobname = "cat.tools.sanlm"

        SPMCommand.__init__(self, **inputs)

    def _format_arg(self, opt, spec, val):
        """Convert input to appropriate format for spm"""
        if opt == "in_files":
            if isinstance(val, list):
                return scans_for_fnames(val)
            else:
                return scans_for_fname(val)
        if opt == "spm_type":
            type_map = {"same": 0, "uint8": 2, "uint16": 512, "float32": 16}
            val = type_map[val]
        return super(CAT12SANLMDenoising, self)._format_arg(opt, spec, val)

    def _list_outputs(self):
        outputs = self._outputs().get()
        outputs["out_file"] = fname_presuffix(
            self.inputs.in_files[0],
            newpath=os.getcwd(),
            prefix=self.inputs.filename_prefix,
            suffix=self.inputs.filename_suffix,
        )
        return outputs


class Cell2Str(Cell):
    def __str__(self):
        """Convert input to appropriate format for cat12"""
        return "{'%s'}" % self.to_string()