1+ """
2+ Nifti wrapper that includes addtional meta data. The meta data is embedded into
3+ the Nifti as an extension.
4+
5+ @author: moloney
6+ """
7+ import json
8+ import numpy as np
9+ import nibabel as nb
10+ from nibabel .nifti1 import Nifti1Extension
11+
12+ dcm_meta_ecode = 19
13+
14+ class DcmMetaExtension (Nifti1Extension ):
15+ '''Subclass on Nifti1Extension. Handles conversion to and from json, checks
16+ the validity of the extension, and provides access to the "meta meta" data.
17+ '''
18+
19+ _req_base_keys = set (('dcmmeta_affine' ,
20+ 'dcmmeta_slice_dim' ,
21+ 'dcmmeta_shape' ,
22+ 'dcmmeta_version' ,
23+ 'global' ,
24+ )
25+ )
26+
27+ def _unmangle (self , value ):
28+ return json .loads (value )
29+
30+ def _mangle (self , value ):
31+ return json .dumps (value , indent = 4 )
32+
33+ def is_valid (self ):
34+ '''Check if the extension is valid.'''
35+ #Check for the required base keys in the json
36+ if not self ._req_base_keys <= set (self ._content ):
37+ return False
38+
39+ shape = self ._content ['dcmmeta_shape' ]
40+
41+ #Check the 'global' dictionary
42+ if not set (('const' , 'slices' )) == set (self ._content ['global' ]):
43+ return False
44+
45+ total_slices = [self .get_n_slices ()]
46+ for dim_size in shape [3 :]:
47+ total_slices .append (total_slices [- 1 ]* dim_size )
48+ for key , vals in self ._content ['global' ]['slices' ].iteritems ():
49+ if len (vals ) != total_slices [- 1 ]:
50+ return False
51+
52+ #Check 'time' and 'vector' dictionaries if they exist
53+ if len (shape ) > 3 :
54+ if not 'time' in self ._content :
55+ return False
56+ if not set (('samples' , 'slices' )) == set (self ._content ['time' ]):
57+ return False
58+ for key , vals in self ._content ['time' ]['samples' ].iteritems ():
59+ if len (vals ) != shape [3 ]:
60+ return False
61+ for key , vals in self ._content ['time' ]['slices' ].iteritems ():
62+ if len (vals ) != total_slices [0 ]:
63+ return False
64+ if len (shape ) > 4 :
65+ if not 'vector' in self ._content :
66+ return False
67+ if not set (('samples' , 'slices' )) == set (self ._content ['vector' ]):
68+ return False
69+ for key , vals in self ._content ['time' ]['samples' ].iteritems ():
70+ if len (vals ) != shape [4 ]:
71+ return False
72+ for key , vals in self ._content ['vector' ]['slices' ].iteritems ():
73+ if len (vals ) != total_slices [1 ]:
74+ return False
75+
76+ return True
77+
78+ def get_affine (self ):
79+ return np .array (self ._content ['dcmmeta_affine' ])
80+
81+ def get_slice_dim (self ):
82+ return self ._content ['dcmmeta_slice_dim' ]
83+
84+ def get_shape (self ):
85+ return tuple (self ._content ['dcmmeta_shape' ])
86+
87+ def get_n_slices (self ):
88+ return self .get_shape ()[self .get_slice_dim ()]
89+
90+ def get_version (self ):
91+ return self ._content ['dcmmeta_version' ]
92+
93+ def to_json_file (self , path ):
94+ '''Write out a JSON formatted text file with the extensions contents.'''
95+ if not self .is_valid ():
96+ raise ValueError ('The content dictionary is not valid.' )
97+ out_file = open (path , 'w' )
98+ out_file .write (self ._mangle (self ._content ))
99+ out_file .close ()
100+
101+ @classmethod
102+ def from_json_file (klass , path ):
103+ '''Read in a JSON formatted text file with the extensions contents.'''
104+ in_file = open (path )
105+ content = in_file .read ()
106+ in_file .close ()
107+ result = klass (dcm_meta_ecode , content )
108+ if not result .is_valid ():
109+ raise ValueError ('The JSON is not valid.' )
110+ return result
111+
112+ @classmethod
113+ def from_runtime_repr (klass , runtime_repr ):
114+ result = klass (dcm_meta_ecode , '{}' )
115+ result ._content = runtime_repr
116+ if not result .is_valid ():
117+ raise ValueError ('The runtime representation is not valid.' )
118+ return result
119+
120+ #Add our extension to nibabel
121+ nb .nifti1 .extension_codes .add_codes (((dcm_meta_ecode ,
122+ "dcmmeta" ,
123+ DcmMetaExtension ),)
124+ )
125+
126+ class NiftiWrapper (object ):
127+ '''Wraps a nibabel.Nifti1Image object containing a DcmMetaExtension header
128+ extension. Provides transparent access to the meta data through 'get_meta'.
129+ Allows the Nifti to be split into sub volumes or joined with others, while
130+ also updating the meta data appropriately.'''
131+
132+ def __init__ (self , nii_img ):
133+ self .nii_img = nii_img
134+ self ._meta_ext = None
135+ for extension in nii_img .get_header ().extensions :
136+ if extension .get_code () == dcm_meta_ecode :
137+ if self ._meta_ext :
138+ raise ValueError ("More than one DcmMetaExtension found" )
139+ self ._meta_ext = extension
140+ if not self ._meta_ext :
141+ raise ValueError ("No DcmMetaExtension found." )
142+ if not self ._meta_ext .is_valid ():
143+ raise ValueError ("The meta extension is not valid" )
144+
145+ def samples_valid (self ):
146+ '''Check if the meta data corresponding to individual time or vector
147+ samples appears to be valid for the wrapped nifti image.'''
148+ #Check if the slice/time/vector dimensions match
149+ img_shape = self .nii_img .get_shape ()
150+ meta_shape = self ._meta_ext .get_shape ()
151+ return meta_shape [2 :] == img_shape [2 :]
152+
153+ def slices_valid (self ):
154+ '''Check if the meta data corresponding to individual slices appears to
155+ be valid for the wrapped nifti image.'''
156+
157+ if self ._meta_ext .get_n_slices () != self .nii_img .get_n_slices ():
158+ return False
159+
160+ #Check that the affines match
161+ return np .allclose (self .nii_img .get_affine (),
162+ self ._meta_ext .get_affine ())
163+
164+ def get_meta (self , key , index = None , default = None ):
165+ '''Return the meta data value for the provided 'key', or 'default' if
166+ there is no such (valid) key.
167+
168+ If 'index' is not provided, only meta data values that are constant
169+ across the entire data set will be considered. If 'index' is provided it
170+ must be a valid index for the nifti voxel data, and all of the meta data
171+ that is applicable to that index will be considered. The per-slice meta
172+ data will only be considered if the object's 'is_aligned' method returns
173+ True.'''
174+
175+ #Pull out the meta dictionary
176+ meta_dict = self ._meta_ext .get_content ()
177+
178+ #First check the constant values
179+ if key in meta_dict ['global' ]['const' ]:
180+ return meta_dict ['global' ]['const' ][key ]
181+
182+ #If an index is provided check the varying values
183+ if not index is None :
184+ #Test if the index is valid
185+ shape = self .nii_img .get_shape ()
186+ if len (index ) != len (shape ):
187+ raise IndexError ('Incorrect number of indices.' )
188+ for dim , ind_val in enumerate (index ):
189+ if ind_val < 0 or ind_val >= shape [dim ]:
190+ raise IndexError ('Index is out of bounds.' )
191+
192+ #First try per time/vector sample values
193+ if self .samples_valid ():
194+ if (len (shape ) > 3 and shape [3 ] > 1 and
195+ key in meta_dict ['time' ]['samples' ]):
196+ return meta_dict ['time' ]['samples' ][key ][index [2 ]]
197+ if (len (shape ) > 4 and shape [4 ] > 1 and
198+ key in meta_dict ['vector' ]['samples' ]):
199+ return meta_dict ['vector' ]['samples' ][key ][index [3 ]]
200+
201+ #Finally, if aligned, try per-slice values
202+ if self .slices_valid ():
203+ slice_dim = self ._meta_ext .get_slice_dim ()
204+ if key in meta_dict ['global' ]['slices' ]:
205+ val_idx = index [slice_dim ]
206+ slices_per_sample = shape [slice_dim ]
207+ for count , idx_val in enumerate (index [3 :]):
208+ val_idx += idx_val * slices_per_sample
209+ slices_per_sample *= shape [count + 3 ]
210+ return meta_dict ['global' ]['slices' ][key ][val_idx ]
211+
212+ if self .samples_valid ():
213+ if (len (shape ) > 3 and shape [3 ] > 1 and
214+ key in meta_dict ['time' ]['slices' ]):
215+ val_idx = index [slice_dim ]
216+ return meta_dict ['time' ]['slices' ][key ][val_idx ]
217+ elif (len (shape ) > 4 and shape [4 ] > 1 and
218+ key in meta_dict ['vector' ]['slices' ]):
219+ val_idx = index [slice_dim ]
220+ val_idx += index [3 ]* shape [slice_dim ]
221+ return meta_dict ['vector' ]['slices' ][key ][val_idx ]
222+
223+ return default
224+
225+ def split (self , dim_idx = None ):
226+ '''Split the meta data along the index 'dim_idx', returning a list of
227+ NiftiWrapper objects. If 'dim_idx' is None it will prefer the vector,
228+ then time, then slice dimensions.
229+ '''
230+ # shape = self.nii_img.get_shape()
231+ # slice_dim = self.nii_img.get_dim_info()[2]
232+ #
233+ # #If dim_idx is None, choose the vector/time/slice dim in that order
234+ # if dim_idx is None:
235+ # dim_idx = len(shape) - 1
236+ # if dim_idx == 2:
237+ # dim_idx = slice_dim
238+ #
239+ # data = self.nii_img.get_data()
240+ # affine = self.nii_img.get_affine()
241+ # header = self.nii_img.get_header()
242+ # if dim_idx == slice_dim:
243+ # header. #Need to unset slice specific bits of the header here.
244+ # results = []
245+ # slices = [slice(None)] * len(shape)
246+ # for idx in shape[dim_idx]:
247+ # slices[dim_idx] = idx
248+ # split_data = data[slices].copy()
249+ # results.append(nb.Nifti1Image(split_data,
250+ # affine.copy(),
251+ # header.copy()
252+ # )
253+ # )
254+ #
255+ # return results
256+
257+
258+ def to_filename (self , out_path ):
259+ if not self ._meta_ext .is_valid :
260+ raise ValueError ("Meta extension is not valid." )
261+ self .nii_img .to_filename (out_path )
262+
263+ @classmethod
264+ def from_filename (klass , path ):
265+ return klass (nb .load (path ))
266+
267+ @classmethod
268+ def from_sequence (klass , others , dim_idx = None ):
269+ '''Create a NiftiWrapper from a sequence of other NiftiWrappers objects.
270+ The Nifti volumes are stacked along dim_idx in the given order.
271+ '''
272+
0 commit comments