enh: write lta

Author: mgxd

nitransforms/io.py

@@ -1,6 +1,7 @@
 import numpy as np
 from nibabel.wrapstruct import LabeledWrapStruct
 from nibabel.volumeutils import Recoder
+from nibabel.affines import voxel_sizes
 
 transform_codes = Recoder((
     (0, 'LINEAR_VOX_TO_VOX'),
@@ -69,10 +70,10 @@ def from_image(klass, img):
         sa = volgeom.structarr
         sa['valid'] = 1
         sa['volume'][:, 0] = img.shape[:3]    # Assumes xyzt-ordered image
-        sa['voxelsize'][:, 0] = img.header.get_zooms()[:3]
+        sa['voxelsize'][:, 0] = voxel_sizes(img.affine)[:3]
         A = img.affine[:3, :3]
         b = img.affine[:3, [3]]
-        cols = A * (1 / sa['voxelsize'])
+        cols = A / sa['voxelsize']
         sa['xras'] = cols[:, [0]]
         sa['yras'] = cols[:, [1]]
         sa['zras'] = cols[:, [2]]
@@ -208,6 +209,10 @@ def from_string(klass, string):
 
             val = np.genfromtxt([valstring.encode()],
                                 dtype=klass.dtype[key])
+            if val not in (0, 1):
+                raise NotImplementedError(
+                    "LTA type {0} is not currently supported".format(transform_codes.label[val])
+                )
             sa[key] = val.reshape(sa[key].shape) if val.size else ''
         for _ in range(sa['nxforms']):
             lta._xforms.append(

nitransforms/linear.py

@@ -17,7 +17,7 @@
 from nibabel.affines import from_matvec, voxel_sizes, obliquity
 from .base import TransformBase
 from .patched import shape_zoom_affine
-from .io import LinearTransformArray, VolumeGeometry
+from .io import LinearTransformArray, LinearTransform, VolumeGeometry
 
 
 LPS = np.diag([-1, -1, 1, 1])
@@ -300,12 +300,20 @@ def to_filename(self, filename, fmt='X5', moving=None):
                 np.savetxt(filename, mat[0], delimiter=' ', fmt='%g')
             return filename
         elif fmt.lower() in ('fs', 'lta'):
-            # linear tranform info
-            src = VolumeGeometry.from_image(moving)
-            dst = VolumeGeometry.from_image(self.reference)
-
-
+            # xform info
+            lt = LinearTransform()
+            lt['sigma'] = 1.
+            lt['m_L'] = self.matrix
+            lt['src'] = VolumeGeometry.from_image(moving)
+            lt['dst'] = VolumeGeometry.from_image(self.reference)
+            # to make LTA file format
+            lta = LinearTransformArray()
+            lta['type'] = 1  # RAS2RAS
+            lta['xforms'] = [lt]
 
+            with open(filename, 'w') as f:
+                f.write(lta.to_string())
+            return filename
 
         return super(Affine, self).to_filename(filename, fmt=fmt)
 
@@ -338,8 +346,7 @@ def load(filename, fmt='X5', reference=None):
     elif fmt.lower() in ('fs', 'lta'):
         with open(filename) as ltafile:
             lta = LinearTransformArray.from_fileobj(ltafile)
-        # TODO: multiple transforms?
-        assert lta['nxforms'] == 1
+        assert lta['nxforms'] == 1  # ever have multiple transforms?
         matrix = lta._xforms[0]['m_L']
     elif fmt.lower() in ('x5', 'bids'):
         raise NotImplementedError