Source code for cortex.database

"""
Contains a singleton object `db` of type `Database` which allows easy access to surface files, anatomical images, and transforms that are stored in the pycortex filestore.
"""
import os
import re
import copy
import glob
import json
import shutil
import warnings
import tempfile
import functools
import numpy as np
from hashlib import sha1
from builtins import input

from . import options

default_filestore = options.config.get('basic', 'filestore')


def _memo(fn):
    @functools.wraps(fn)
    def memofn(self, *args, **kwargs):
        if not hasattr(self, "_memocache"):
            setattr(self, "_memocache", dict())
        #h = sha1(str((id(fn), args, kwargs))).hexdigest()
        h = str((id(fn), args, kwargs))
        if h not in self._memocache:
            self._memocache[h] = fn(self, *args, **kwargs)
        return copy.deepcopy(self._memocache[h])

    return memofn

class SubjectDB(object):
    def __init__(self, subj, filestore=default_filestore):
        self.subject = subj
        self._warning = None
        self._transforms = None
        self._surfaces = None
        self.filestore = filestore

        try:
            with open(os.path.join(filestore, subj, "warning.txt")) as fp:
                self._warning = fp.read()
        except IOError:
            pass

    @property
    def transforms(self):
        if self._transforms is not None:
            return self._transforms
        self._transforms = XfmDB(self.subject, filestore=self.filestore)
        return self._transforms

    @property
    def surfaces(self):
        if self._surfaces is not None:
            return self._surfaces
        self._surfaces = SurfaceDB(self.subject, filestore=self.filestore)
        return self._surfaces

class SurfaceDB(object):
    def __init__(self, subj, filestore=default_filestore):
        self.subject = subj
        self.types = {}
        db = Database(filestore)
        for name in db.get_paths(subj)['surfs'].keys():
            self.types[name] = Surf(subj, name, filestore=filestore)
                
    def __repr__(self):
        return "Surfaces: [{surfs}]".format(surfs=', '.join(list(self.types.keys())))
    
    def __dir__(self):
        return list(self.types.keys())

    def __getattr__(self, attr):
        if attr in self.types:
            return self.types[attr]
        raise AttributeError(attr)

class Surf(object):
    def __init__(self, subject, surftype, filestore=default_filestore):
        self.subject, self.surftype = subject, surftype
        self.db = Database(filestore)

    def get(self, hemisphere="both"):
        return self.db.get_surf(self.subject, self.surftype, hemisphere)
    
    def show(self, hemisphere="both"):
        from mayavi import mlab
        pts, polys = self.db.get_surf(self.subject, self.surftype, hemisphere, merge=True, nudge=True)
        return mlab.triangular_mesh(pts[:,0], pts[:,1], pts[:,2], polys)

class XfmDB(object):
    def __init__(self, subj, filestore=default_filestore):
        self.subject = subj
        self.filestore = filestore
        self.xfms = Database(self.filestore).get_paths(subj)['xfms']

    def __getitem__(self, name):
        if name in self.xfms:
            return XfmSet(self.subject, name, filestore=self.filestore)
        raise AttributeError
    
    def __repr__(self):
        return "Transforms: [{xfms}]".format(xfms=",".join(self.xfms))

class XfmSet(object):
    def __init__(self, subj, name, filestore=default_filestore):
        self.subject = subj
        self.name = name
        jspath = os.path.join(filestore, subj, 'transforms', name, 'matrices.xfm')
        with open(jspath) as fp:
            self._jsdat = json.load(fp)
        self.masks = MaskSet(subj, name, filestore=filestore)
        self.db = Database(filestore)
    
    def __getattr__(self, attr):
        if attr in self._jsdat:
            return self.db.get_xfm(self.subject, self.name, attr)
        raise AttributeError
    
    def __repr__(self):
        return "Types: {types}".format(types=", ".join(self._jsdat.keys()))

class MaskSet(object):
    def __init__(self, subj, name, filestore=default_filestore):
        self.subject = subj
        self.xfmname = name
        maskform = Database(filestore).get_paths(subj)['masks']
        maskpath = maskform.format(xfmname=name, type='*')
        self._masks = dict((os.path.split(path)[1][5:-7], path) for path in glob.glob(maskpath))

    def __getitem__(self, item):
        import nibabel
        return nibabel.load(self._masks[item]).get_data().T

    def __repr__(self):
        return "Masks: [{types}]".format(types=', '.join(self._masks.keys()))

[docs]class Database(object): """ Database() Surface database Attributes ---------- This database object dynamically generates handles to all subjects within the filestore. """
[docs] def __init__(self, filestore=default_filestore): self.filestore = filestore self._subjects = None self.auxfile = None
def __repr__(self): subjs = "\n ".join(sorted(self.subjects.keys())) return """Pycortex database\n Subjects:\n {subjs}""".format(subjs=subjs) def __getattr__(self, attr): if attr in self.subjects: if self.subjects[attr]._warning is not None: warnings.warn(self.subjects[attr]._warning) return self.subjects[attr] else: raise AttributeError def __dir__(self): return ["save_xfm","get_xfm", "get_surf", "get_anat", "get_surfinfo", "subjects", # "get_paths", # Add? "get_mask", "get_overlay","get_cache", "get_view", "save_view", "get_mnixfm", 'get_mri_surf2surf_matrix'] + list(self.subjects.keys()) @property def subjects(self): if self._subjects is not None: return self._subjects subjs = os.listdir(os.path.join(self.filestore)) subjs = [s for s in subjs if os.path.isdir(os.path.join(self.filestore, s))] self._subjects = dict([(sname, SubjectDB(sname, filestore=self.filestore)) for sname in subjs]) return self._subjects
[docs] def get_anat(self, subject, type='raw', xfmname=None, recache=False, order=1, **kwargs): """Return anatomical information from the filestore. Anatomical information is defined as any volume-space anatomical information pertaining to the subject, such as T1 image, white matter masks, etc. Volumes not found in the database will be automatically generated. Parameters ---------- subject : str Name of the subject type : str Type of anatomical volume to return. This should be the name of one of the recache : bool Regenerate the information Returns ------- volume : nibabel object Volume containing """ opts = "" if len(kwargs) > 0: opts = "[%s]"%','.join(["%s=%s"%i for i in kwargs.items()]) anatform = self.get_paths(subject)['anats'] anatfile = anatform.format(type=type, opts=opts, ext="nii.gz") if not os.path.exists(anatfile) or recache: print("Generating %s anatomical..."%type) from . import anat getattr(anat, type)(anatfile, subject, **kwargs) import nibabel anatnib = nibabel.load(anatfile) if xfmname is None: return anatnib from . import volume return volume.anat2epispace(anatnib.get_data().T.astype(np.float), subject, xfmname, order=order)
[docs] def get_surfinfo(self, subject, type="curvature", recache=False, **kwargs): """Return auxillary surface information from the filestore. Surface info is defined as anatomical information specific to a subject in surface space. A Vertex class will be returned as necessary. Info not found in the filestore will be automatically generated. See documentation in cortex.surfinfo for auto-generation code Parameters ---------- subject: str Subject name for which to return info type: str Type of surface info returned, IE. curvature, distortion, sulcaldepth, etc. recache: bool Regenerate the information Returns ------- verts : Vertex class If the surface information has "left" and "right" entries, a Vertex class is returned - OR - npz : npzfile Otherwise, an npz object is returned. Remember to close it! """ opts = "" if len(kwargs) > 0: opts = "[%s]"%','.join(["%s=%s"%i for i in kwargs.items()]) try: self.auxfile.get_surf(subject, "fiducial") surfifile = os.path.join(self.get_cache(subject),"%s%s.npz"%(type, opts)) except (AttributeError, IOError): surfiform = self.get_paths(subject)['surfinfo'] surfifile = surfiform.format(type=type, opts=opts) if not os.path.exists(os.path.join(self.filestore, subject, "surface-info")): os.makedirs(os.path.join(self.filestore, subject, "surface-info")) if not os.path.exists(surfifile) or recache: print ("Generating %s surface info..."%type) from . import surfinfo getattr(surfinfo, type)(surfifile, subject, **kwargs) npz = np.load(surfifile) if "left" in npz and "right" in npz: from .dataset import Vertex verts = np.hstack([npz['left'], npz['right']]) npz.close() return Vertex(verts, subject) return npz
[docs] def get_mri_surf2surf_matrix(self, subject, surface_type, hemi='both', fs_subj=None, target_subj='fsaverage', **kwargs): """Get matrix generated by surf2surf to map one subject's surface to another's Parameters ---------- subject : pycortex pycortex subject ID surface_type : str type of surface; one of ['fiducial', 'inflated', 'white', 'pial'] ... more? hemi : str or list one of: 'both', 'lh', 'rh' fs_subj : str string ID for freesurfer subject (if different from pycortex subject ID; None assumes they are the same) target_subj : str string ID for freesurfer subject to which to map surface from `subject` (or `fs_subj`) Other Parameters ---------------- subjects_dir : str Custom path to freesurfer subjects dir can be specified in the keyword args n_neighbors : int ... random_state : scalar ... n_test_images : int ... coef_threshold : scalar or None ... renormalize : bool ... """ from .freesurfer import get_mri_surf2surf_matrix as mri_s2s from .utils import load_sparse_array, save_sparse_array if fs_subj is None: fs_subj = subject fpath = self.get_paths(subject)['surf2surf'].format(source=fs_subj, target=target_subj) # Backward compatiblity fdir, _ = os.path.split(fpath) if not os.path.exists(fdir): print("Creating surf2surf directory for subject %s"%(subject)) os.makedirs(fdir) if hemi == 'both': hemis = ['lh', 'rh'] else: hemis = [hemi] if os.path.exists(fpath): mats = [load_sparse_array(fpath, h) for h in hemis] else: mats = [] for h in hemis: tmp = mri_s2s(fs_subj, h, surface_type, target_subj=target_subj, **kwargs) mats.append(tmp) save_sparse_array(fpath, tmp, h, mode='a') return mats
[docs] def get_overlay(self, subject, overlay_file=None, **kwargs): from . import svgoverlay pts, polys = self.get_surf(subject, "flat", merge=True, nudge=True) paths = self.get_paths(subject) # NOTE: This try loop is broken, in that it does nothing for the inteded # use case (loading an overlay from a packed subject) - needs fixing. # This hasn't come up yet due to very rare use of packed subjects. if self.auxfile is not None: try: # Prob should be something like: #self.auxfile.get_overlay(subject, **kwargs) tf = self.auxfile.get_overlay(subject) # kwargs?? svgfile = tf.name except (AttributeError, IOError): # I think the rest of the code should be here (as in other functions...) pass if 'pts' in kwargs: pts = kwargs['pts'] del kwargs['pts'] # ... and `svgfile` variable should be used here... if os.path.exists(paths['rois']) and not os.path.exists(paths['overlays']): svgoverlay.import_roi(paths['rois'], paths['overlays']) if overlay_file is None: overlay_file = paths['overlays'] return svgoverlay.get_overlay(subject, overlay_file, pts, polys, **kwargs)
[docs] def save_xfm(self, subject, name, xfm, xfmtype="magnet", reference=None): """ Load a transform into the surface database. If the transform exists already, update it If it does not exist, copy the reference epi into the filestore and insert. Parameters ---------- subject : str Name of the subject name : str Name to identify the transform xfm : (4,4) array The affine transformation matrix xfmtype : str, optional Type of the provided transform, either magnet space or coord space. Defaults to 'magnet'. reference : str, optional The nibabel-compatible reference image associated with this transform. Required if name not in database """ if xfmtype not in ["magnet", "coord"]: raise TypeError("Unknown transform type") import nibabel path = os.path.join(self.filestore, subject, "transforms", name) fname = os.path.join(path, "matrices.xfm") if os.path.exists(fname): with open(fname) as fp: jsdict = json.load(fp) else: os.mkdir(path) if reference is None: raise ValueError("Please specify a reference") fpath = os.path.join(path, "reference.nii.gz") nib = nibabel.load(reference) data = nib.get_data() if len(data.shape) > 3: import warnings warnings.warn('You are importing a 4D dataset, automatically selecting the first volume as reference') data = data[...,0] out = nibabel.Nifti1Image(data, nib.get_affine(), header=nib.get_header()) nibabel.save(out, fpath) jsdict = dict() nib = nibabel.load(os.path.join(path, "reference.nii.gz")) if xfmtype == "magnet": jsdict['magnet'] = np.array(xfm).tolist() jsdict['coord'] = np.dot(np.linalg.inv(nib.get_affine()), xfm).tolist() elif xfmtype == "coord": jsdict['coord'] = np.array(xfm).tolist() jsdict['magnet'] = np.dot(nib.get_affine(), xfm).tolist() files = self.get_paths(subject) if len(glob.glob(files['masks'].format(xfmname=name, type="*"))) > 0: raise ValueError('Refusing to change a transform with masks') with open(fname, "w") as fp: json.dump(jsdict, fp, sort_keys=True, indent=4)
[docs] def get_xfm(self, subject, name, xfmtype="coord"): """Retrieves a transform from the filestore Parameters ---------- subject : str Name of the subject name : str Name of the transform xfmtype : str, optional Type of transform to return. Defaults to coord. """ from .xfm import Transform if xfmtype == 'coord': try: return self.auxfile.get_xfm(subject, name) except (AttributeError, IOError): pass if name == "identity": nib = self.get_anat(subject, 'raw') return Transform(np.linalg.inv(nib.get_affine()), nib) fname = os.path.join(self.filestore, subject, "transforms", name, "matrices.xfm") reference = os.path.join(self.filestore, subject, "transforms", name, "reference.nii.gz") with open(fname) as f: xfmdict = json.load(f) return Transform(xfmdict[xfmtype], reference)
[docs] @_memo def get_surf(self, subject, type, hemisphere="both", merge=False, nudge=False): '''Return the surface pair for the given subject, surface type, and hemisphere. Parameters ---------- subject : str Name of the subject type : str Type of surface to return, probably in (fiducial, inflated, veryinflated, hyperinflated, superinflated, flat) hemisphere : "lh", "rh" Which hemisphere to return merge : bool Vstack the hemispheres, if requesting both nudge : bool Nudge the hemispheres apart from each other, for overlapping surfaces (inflated, etc) Returns ------- left, right : If request is for both hemispheres, otherwise: pts, polys, norms : ((p,3) array, (f,3) array, (p,3) array or None) For single hemisphere ''' try: return self.auxfile.get_surf(subject, type, hemisphere, merge=merge, nudge=nudge) except (AttributeError, IOError): pass files = self.get_paths(subject)['surfs'] if hemisphere.lower() == "both": left, right = [ self.get_surf(subject, type, hemisphere=h) for h in ["lh", "rh"]] if type != "fiducial" and nudge: left[0][:,0] -= left[0].max(0)[0] right[0][:,0] -= right[0].min(0)[0] if merge: pts = np.vstack([left[0], right[0]]) polys = np.vstack([left[1], right[1]+len(left[0])]) return pts, polys return left, right elif hemisphere.lower() in ("lh", "left"): hemi = "lh" elif hemisphere.lower() in ("rh", "right"): hemi = "rh" else: raise TypeError("Not a valid hemisphere name") if type == 'fiducial' and 'fiducial' not in files: wpts, polys = self.get_surf(subject, 'wm', hemi) ppts, _ = self.get_surf(subject, 'pia', hemi) return (wpts + ppts) / 2, polys try: from . import formats fnm = str(os.path.splitext(files[type][hemi])[0]) return formats.read(fnm) except KeyError: raise IOError
[docs] def save_mask(self, subject, xfmname, type, mask): fname = self.get_paths(subject)['masks'].format(xfmname=xfmname, type=type) if os.path.exists(fname): raise IOError('Refusing to overwrite existing mask') import nibabel xfm = self.get_xfm(subject, xfmname) if xfm.shape != mask.shape: raise ValueError("Invalid mask shape: must match shape of reference image") affine = xfm.reference.get_affine() nib = nibabel.Nifti1Image(mask.astype(np.uint8).T, affine) nib.to_filename(fname)
[docs] def get_mask(self, subject, xfmname, type='thick'): if hasattr(type, 'decode'): type = type.decode('utf8') try: self.auxfile.get_mask(subject, xfmname, type) except (AttributeError, IOError): pass fname = self.get_paths(subject)['masks'].format(xfmname=xfmname, type=type) try: import nibabel nib = nibabel.load(fname) return nib.get_data().T != 0 except IOError: print('Mask not found, generating...') from .utils import get_cortical_mask mask = get_cortical_mask(subject, xfmname, type) self.save_mask(subject, xfmname, type, mask) return mask
[docs] def get_shared_voxels(self, subject, xfmname, hemi="both", merge=True, use_astar=True, recache=False): """Get an array indicating which vertices are inappropriately mapped to the same voxel. For a given transform and surface, returns an array containing a list of vertices which are spatially distant on the cortical surface but that map to the same voxels (this occurs at sulcal crossings) """ # Test for packed subjects try: voxels = self.auxfile.get_shared_voxels(subject, xfmname, hemi=hemi, merge=merge, use_astar=use_astar) return voxels except (AttributeError, IOError): pass # Proceed w/ potential load shared_voxel_file = os.path.join(self.get_cache(subject), 'shared_vertices_{xfmname}_{hemi}.npy'.format(xfmname=xfmname, hemi=hemi)) if not os.path.exists(shared_voxel_file) or recache: print('Shared voxel array not found, generating...') from .utils import get_shared_voxels as _get_shared_voxels voxels = _get_shared_voxels(subject, xfmname, hemi=hemi, merge=merge, use_astar=use_astar) np.save(shared_voxel_file, voxels) return voxels else: voxels = np.load(shared_voxel_file) return voxels
[docs] def get_coords(self, subject, xfmname, hemisphere="both", magnet=None): """Calculate the coordinates of each vertex in the epi space by transforming the fiducial to the coordinate space Parameters ---------- subject : str Name of the subject xfmname : str Name of the transform hemisphere : str, optional Which hemisphere to return. If "both", return concatenated. Defaults to "both". """ import warnings warnings.warn('Please use a Mapper object instead', DeprecationWarning) if magnet is None: xfm = self.get_xfm(subject, xfmname, xfmtype="coord") else: xfm = self.get_xfm(subject, xfmname, xfmtype="magnet") xfm = np.linalg.inv(magnet) * xfm coords = [] vtkTmp = self.get_surf(subject, "fiducial", hemisphere=hemisphere, nudge=False) if not isinstance(vtkTmp,(tuple,list)): vtkTmp = [vtkTmp] for pts, polys in vtkTmp: wpts = np.vstack([pts.T, np.ones(len(pts))]) coords.append(np.dot(xfm.xfm, wpts)[:3].round().astype(int).T) return coords
[docs] def get_cache(self, subject): try: self.auxfile.get_surf(subject, "fiducial") #generate the hashed name of the filename and subject as the directory name import hashlib hashname = "pycx_%s"%hashlib.md5(self.auxfile.h5.filename).hexdigest()[-8:] cachedir = os.path.join(tempfile.gettempdir(), hashname, subject) except (AttributeError, IOError): try: # Get cache dir from config file cachedir = os.path.join(options.config.get('basic', 'cache'), subject, 'cache') except options.configparser.NoOptionError: # If not defined, go with default cache cachedir = os.path.join(self.filestore, subject, "cache") cachedir = os.path.expanduser(cachedir) if not os.path.exists(cachedir): os.makedirs(cachedir) return cachedir
[docs] def clear_cache(self, subject, clear_all_caches=True): """Clears config-specified and default file caches for a subject. """ local_cachedir = self.get_cache(subject) shutil.rmtree(local_cachedir) os.makedirs(local_cachedir) # Check on default cache for subject as well, on the off chance # that other people have cached files here for this subject default_cachedir = os.path.join(self.filestore, subject, "cache") if clear_all_caches is True: # Just delete them. shutil.rmtree(default_cachedir) os.makedirs(default_cachedir) if (len(os.listdir(default_cachedir)) > 0): # Make sure you didn't want to delete them. You probably should. proceed = input("Files exist in %s too! Delete them? [y]/n: "%default_cachedir) if proceed.lower() in ('', 'y'): shutil.rmtree(default_cachedir) os.makedirs(default_cachedir)
[docs] def get_paths(self, subject): """Get a dictionary with a list of all candidate filenames for associated data, such as roi overlays, flatmap caches, and ctm caches. """ surfpath = os.path.join(self.filestore, subject, "surfaces") if self.subjects[subject]._warning is not None: warnings.warn(self.subjects[subject]._warning) surfs = dict() for surf in os.listdir(surfpath): ssurf = os.path.splitext(surf)[0].split('_') name = '_'.join(ssurf[:-1]) hemi = ssurf[-1] if name not in surfs: surfs[name] = dict() surfs[name][hemi] = os.path.abspath(os.path.join(surfpath,surf)) viewsdir = os.path.join(self.filestore, subject, "views") if not os.path.exists(viewsdir): os.makedirs(viewsdir) views = os.listdir(viewsdir) filenames = dict( surfs=surfs, xfms=sorted(os.listdir(os.path.join(self.filestore, subject, "transforms"))), xfmdir=os.path.join(self.filestore, subject, "transforms", "{xfmname}", "matrices.xfm"), anats=os.path.join(self.filestore, subject, "anatomicals", '{type}{opts}.{ext}'), surfinfo=os.path.join(self.filestore, subject, "surface-info", '{type}{opts}.npz'), masks=os.path.join(self.filestore, subject, 'transforms', '{xfmname}', 'mask_{type}.nii.gz'), rois=os.path.join(self.filestore, subject, "rois.svg").format(subj=subject), overlays=os.path.join(self.filestore, subject, "overlays.svg").format(subj=subject), views=sorted([os.path.splitext(f)[0] for f in views]), surf2surf=os.path.join(self.filestore, subject, "surf2surf", "{source}_to_{target}", "matrices.hdf"), ) return filenames
[docs] def make_subj(self, subject): if os.path.exists(os.path.join(self.filestore, subject)): if input("Are you sure you want to overwrite this existing subject?\n" "This will delete all files for this subject in the filestore, " "including all blender cuts. Type YES\n") == "YES": shutil.rmtree(os.path.join(self.filestore, subject)) else: raise ValueError('Do not overwrite') for dirname in ['transforms', 'anatomicals', 'cache', 'surfaces', 'surface-info','views']: try: path = os.path.join(self.filestore, subject, dirname) os.makedirs(path) except OSError: print("Error making directory %s"%path)
[docs] def save_view(self,vw,subject,name,is_overwrite=False): """Set the view for an open webshow instance from a saved view Sets the view in a currently-open cortex.webshow instance (with handle `vw`) to the saved view named `name` Parameters ---------- vw : handle for pycortex webgl viewer Handle for open webgl session (returned by cortex.webshow) subject : string pycortex subject id name : string Name of stored view to re-load Notes ----- Equivalent to call to vw.save_view(subject,name) For a list of the view parameters saved, see viewer._capture_view """ view = vw._capture_view() sName = os.path.join(self.filestore, subject, "views", name+'.json') if os.path.exists(sName): if not is_overwrite: raise IOError('Refusing to over-write extant view If you want to do this, set is_overwrite=True!') with open(sName,'w') as fp: json.dump(view, fp)
[docs] def get_view(self,vw,subject,name): """Set the view for an open webshow instance from a saved view Sets the view in a currently-open cortex.webshow instance (with handle `vw`) to the saved view named `name` Parameters ---------- vw : handle for cortex.webshow Handle for open webshow session (returned by cortex.webshow) subject : string, subject name name : string Name of stored view to re-load Notes ----- Equivalent to call to vw.get_view(subject,name) For a list of the view parameters saved, see viewer._capture_view """ sName = os.path.join(self.filestore, subject, "views", name+'.json') with open(sName) as fp: view = json.load(fp) vw._set_view(**view)
[docs] def get_mnixfm(self, subject, xfm, template=None): """Get transform from the space specified by `xfm` to MNI space. Parameters ---------- subject : str Subject identifier xfm : str Name of functional space transform. Can be 'identity' for anat space. template : str or None, optional Path to MNI template volume. If None, uses default specified in cortex.mni Returns ------- mnixfm : numpy.ndarray Transformation matrix from the space specified by `xfm` to MNI space. Notes ----- Equivalent to cortex.mni.compute_mni_transform, but this function also caches the result (which is nice because computing it can be slow). See Also -------- cortex.mni.compute_mni_transform cortex.mni.transform_to_mni cortex.mni.transform_mni_to_subject """ from . import mni if template is None: templatehash = "default" else: templatehash = sha1(template).hexdigest() # Check cache first mnixfmfile = os.path.join(self.get_cache(subject), "mni_xfm-%s-%s.txt"%(xfm, templatehash)) if os.path.exists(mnixfmfile): mnixfm = np.loadtxt(mnixfmfile) else: # Run the transform if template is None: mnixfm = mni.compute_mni_transform(subject, xfm) else: mnixfm = mni.compute_mni_transform(subject, xfm, template) # Cache the result mni._save_fsl_xfm(mnixfmfile, mnixfm) return mnixfm
db = Database()