The Virtual Brain Project

Source code for tvb.datatypes.surfaces

# -*- coding: utf-8 -*-
#
#
#  TheVirtualBrain-Scientific Package. This package holds all simulators, and 
# analysers necessary to run brain-simulations. You can use it stand alone or
# in conjunction with TheVirtualBrain-Framework Package. See content of the
# documentation-folder for more details. See also http://www.thevirtualbrain.org
#
# (c) 2012-2017, Baycrest Centre for Geriatric Care ("Baycrest") and others
#
# This program is free software: you can redistribute it and/or modify it under the
# terms of the GNU General Public License as published by the Free Software Foundation,
# either version 3 of the License, or (at your option) any later version.
# This program is distributed in the hope that it will be useful, but WITHOUT ANY
# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A
# PARTICULAR PURPOSE.  See the GNU General Public License for more details.
# You should have received a copy of the GNU General Public License along with this
# program.  If not, see <http://www.gnu.org/licenses/>.
#
#
#   CITATION:
# When using The Virtual Brain for scientific publications, please cite it as follows:
#
#   Paula Sanz Leon, Stuart A. Knock, M. Marmaduke Woodman, Lia Domide,
#   Jochen Mersmann, Anthony R. McIntosh, Viktor Jirsa (2013)
#       The Virtual Brain: a simulator of primate brain network dynamics.
#   Frontiers in Neuroinformatics (7:10. doi: 10.3389/fninf.2013.00010)
#
#

"""
Surface relates DataTypes. This brings together the scientific and framework 
methods that are associated with the surfaces data.

.. moduleauthor:: Ionel Ortelecan <ionel.ortelecan@codemart.ro>
.. moduleauthor:: Bogdan Neacsa <bogdan.neacsa@codemart.ro>
.. moduleauthor:: Lia Domide <lia.domide@codemart.ro>
.. moduleauthor:: Stuart A. Knock <stuart.knock@gmail.com>
.. moduleauthor:: Marmaduke Woodman <mmwoodman@gmail.com>

"""

import warnings
import json
import numpy
import tvb.basic.traits.types_basic as basic
import tvb.datatypes.arrays as arrays
from tvb.basic.traits import util, exceptions
from tvb.basic.logger.builder import get_logger
from tvb.basic.traits.types_mapped import MappedType, SparseMatrix
from tvb.basic.traits.core import FILE_STORAGE_NONE
from tvb.basic.profile import TvbProfile
from tvb.basic.readers import ZipReader, try_get_absolute_path


try:
    import gdist
except ImportError:
    class ExceptionRaisingGdistModule:
        msg = "Geodesic distance module is unavailable, cannot compute gdist matrix."

        def local_gdist_matrix(self, *args, **kwds):
            raise RuntimeError(self.msg)

        def compute_gdist(self, *args, **kwds):
            raise RuntimeError(self.msg)

    gdist = ExceptionRaisingGdistModule()
    msg = "Geodesic distance module is unavailable; some functionality for surfaces will be unavailable."
    warnings.warn(msg)


LOG = get_logger(__name__)


OUTER_SKIN = "Skin Air"
OUTER_SKULL = "Skull Skin"
INNER_SKULL = "Brain Skull"
CORTICAL = "Cortical Surface"
WHITE_MATTER = "White Matter"
EEG_CAP = "EEG Cap"
FACE = "Face"


# TODO: This is just a temporary solution placed here to remove dependency from tvb.basic to tvb framework.
# As soon as we implement a better solution to the datatype framework diamond problem this should be removed.
[docs]def paths2url(datatype_entity, attribute_name, flatten=False, parameter=None, datatype_kwargs=None): """Prepare a File System Path for passing into an URL.""" jskw = json.dumps(datatype_kwargs) url = '%s%s/%s/%s/%s' url %= TvbProfile.current.web.VISUALIZERS_URL_PREFIX, datatype_entity.gid, attribute_name, flatten, jskw if parameter is None: return url return '%s?%s' % (url, parameter) # TODO document magic numbers here # Slices are for vertices [0.....SPLIT_MAX_SIZE + SPLIT_BUFFER_SIZE] # [SPLIT_MAX_SIZE ..... 2 * SPLIT_BUFFER_SIZE + SPLIT_BUFFER_SIZE] # Triangles [0, 1, 2], [3, 4, 5], [6, 7, 8]..... # Vertices - no of triangles * 3
SPLIT_MAX_SIZE = 40000 SPLIT_BUFFER_SIZE = 15000 SPLIT_PICK_MAX_TRIANGLE = 20000 KEY_TRIANGLES = "triangles" KEY_VERTICES = "vertices" KEY_HEMISPHERE = "hemisphere" KEY_START = "start_idx" KEY_END = "end_idx" HEMISPHERE_LEFT = "LEFT" HEMISPHERE_RIGHT = "RIGHT" HEMISPHERE_UNKNOWN = "NONE"
[docs]class ValidationResult(object): """ Used by surface validate methods to report non-fatal failed validations """ def __init__(self): self.warnings = []
[docs] def add_warning(self, msg, data): self.warnings.append((msg, data)) self._log(msg, data)
def _log(self, msg, data): LOG.warn(msg) if data: LOG.debug(data)
[docs] def merge(self, other): r = ValidationResult() r.warnings = self.warnings + other.warnings return r
[docs] def summary(self): return ' | '.join(msg for msg, _ in self.warnings)
[docs]class Surface(MappedType): """A base class for other surfaces.""" vertices = arrays.PositionArray( label="Vertex positions", order=-1, doc="""An array specifying coordinates for the surface vertices.""") triangles = arrays.IndexArray( label="Triangles", order=-1, target=vertices, doc="""Array of indices into the vertices, specifying the triangles which define the surface.""") vertex_normals = arrays.OrientationArray( label="Vertex normal vectors", order=-1, doc="""An array of unit normal vectors for the surfaces vertices.""") triangle_normals = arrays.OrientationArray( label="Triangle normal vectors", order=-1, doc="""An array of unit normal vectors for the surfaces triangles.""") geodesic_distance_matrix = SparseMatrix( label="Geodesic distance matrix", order=-1, required=False, file_storage=FILE_STORAGE_NONE, doc="""A sparse matrix of truncated geodesic distances""") # 'CS' number_of_vertices = basic.Integer( label="Number of vertices", order=-1, doc="""The number of vertices making up this surface.""") number_of_triangles = basic.Integer( label="Number of triangles", order=-1, doc="""The number of triangles making up this surface.""") edge_mean_length = basic.Float(order=-1) edge_min_length = basic.Float(order=-1) edge_max_length = basic.Float(order=-1) ##--------------------- FRAMEWORK ATTRIBUTES -----------------------------## hemisphere_mask = arrays.BoolArray( label="An array specifying if a vertex belongs to the right hemisphere", file_storage=FILE_STORAGE_NONE, required=False, order=-1) zero_based_triangles = basic.Bool(order=-1) split_triangles = arrays.IndexArray(order=-1, required=False) number_of_split_slices = basic.Integer(order=-1) split_slices = basic.Dict(order=-1) bi_hemispheric = basic.Bool(order=-1) surface_type = basic.String valid_for_simulations = basic.Bool(order=-1) __mapper_args__ = {'polymorphic_on': 'surface_type'} @classmethod def from_file(cls, source_file="cortex_16384.zip", instance=None): """Construct a Surface from source_file.""" if instance is None: result = cls() else: result = instance source_full_path = try_get_absolute_path("tvb_data.surfaceData", source_file) reader = ZipReader(source_full_path) result.vertices = reader.read_array_from_file("vertices.txt") result.vertex_normals = reader.read_array_from_file("normals.txt") result.triangles = reader.read_array_from_file("triangles.txt", dtype=numpy.int32) return result def configure(self): "Compute additional attributes on surface data required for full functionality." self.number_of_vertices = self.vertices.shape[0] self.number_of_triangles = self.triangles.shape[0] if self.triangle_normals.size == 0: LOG.debug("Triangle normals not available. Start to compute them.") self.compute_triangle_normals() LOG.debug("End computing triangles normals") if self.vertex_normals.size == 0: LOG.debug("Vertex normals not available. Start to compute them.") self.compute_vertex_normals() LOG.debug("End computing vertex normals") if self._edge_lengths is None: self._find_edge_lengths() self.framework_configure() def validate(self): """ Combines scientific and framework surface validations. """ result_sci = self.scientific_validate() result_fr = self.framework_validate() validation_result = result_sci.merge(result_fr) self.user_tag_3 = validation_result.summary() return validation_result # from scientific surfaces _vertex_neighbours = None _vertex_triangles = None _triangle_centres = None _triangle_angles = None _triangle_areas = None _edges = None _number_of_edges = None _edge_lengths = None _edge_length_mean = None _edge_length_min = None _edge_length_max = None _edge_triangles = None def _find_summary_info(self): """ Gather scientifically interesting summary information from an instance of this datatype. """ summary = {"Surface type": self.__class__.__name__, "Valid for simulations": self.valid_for_simulations, "Number of vertices": self.number_of_vertices, "Number of triangles": self.number_of_triangles, "Number of edges": self.number_of_edges, "Has two hemispheres": self.bi_hemispheric, "Edge lengths, mean (mm)": self.edge_length_mean, "Edge lengths, shortest (mm)": self.edge_length_min, "Edge lengths, longest (mm)": self.edge_length_max} return summary def geodesic_distance(self, sources, max_dist=None, targets=None): """ Calculate the geodesic distance between vertices of the surface, ``sources``: one or more indices into vertices, these are required, they specify the vertices from which the distance is calculated. NOTE: if multiple sources are provided then the distance returned is the shortest from the closest source. ``max_dist``: find the distance to vertices out as far as max_dist. ``targets``: one or more indices into vertices,. NOTE: Either ``targets`` or ``max_dist`` should be specified, but not both, specifying neither is equivalent to max_dist=1e100. NOTE: when max_dist is specifed, distances > max_dist are returned as numpy.inf If end_vertex is omitted the distance from the starting vertex to all vertices within max_dist will be returned, if max_dist is also omitted the distance to all vertices on the surface will be returned. """ # TODO: Probably should check that targets and start_vertex are less than # number of vertices, etc... # if NO_GEODESIC_DISTANCE: # LOG.error("%s: The geodesic distance library didn't load" % repr(self)) # return if max_dist is not None and targets is not None: raise ValueError("Specifying both targets and max_dist doesn't work.") # Cython expects data with specific dtype verts = self.vertices.astype(numpy.float64) tris = self.triangles.astype(numpy.int32) srcs = sources.astype(numpy.int32) kwd = {} # handle custom args if targets is not None: kwd['target_indices'] = targets.astype(numpy.int32) if max_dist is not None: kwd['max_distance'] = max_dist dist = gdist.compute_gdist(verts, tris, source_indices=srcs, **kwd) return dist # TODO why two methods for this? def compute_geodesic_distance_matrix(self, max_dist): """ Calculate a sparse matrix of the geodesic distance from each vertex to all vertices within max_dist of them on the surface, ``max_dist``: find the distance to vertices out as far as max_dist. NOTE: Compute time increases rapidly with max_dist and the memory efficiency of the sparse matrices decreases, so, don't use too large a value for max_dist... """ # TODO: Probably should check that max_dist isn't "too" large or too # small, min should probably be max edge length... # if NO_GEODESIC_DISTANCE: # LOG.error("%s: The geodesic distance library didn't load" % repr(self)) # return dist = gdist.local_gdist_matrix(self.vertices.astype(numpy.float64), self.triangles.astype(numpy.int32), max_distance=max_dist) self.geodesic_distance_matrix = dist @property def vertex_neighbours(self): """ List of the set of neighbours for each vertex. """ if self._vertex_neighbours is None: self._vertex_neighbours = self._find_vertex_neighbours() return self._vertex_neighbours def _find_vertex_neighbours(self): """ . """ neighbours = [[] for _ in range(self.number_of_vertices)] for k in range(self.number_of_triangles): neighbours[self.triangles[k, 0]].append(self.triangles[k, 1]) neighbours[self.triangles[k, 0]].append(self.triangles[k, 2]) neighbours[self.triangles[k, 1]].append(self.triangles[k, 0]) neighbours[self.triangles[k, 1]].append(self.triangles[k, 2]) neighbours[self.triangles[k, 2]].append(self.triangles[k, 0]) neighbours[self.triangles[k, 2]].append(self.triangles[k, 1]) neighbours = map(frozenset, neighbours) return neighbours @property def vertex_triangles(self): """ List of the set of triangles surrounding each vertex. """ if self._vertex_triangles is None: self._vertex_triangles = self._find_vertex_triangles() return self._vertex_triangles def _find_vertex_triangles(self): # self.attr calls __get__@type_mapped which is performance sensitive here triangles = self.triangles vertex_triangles = [[] for _ in range(self.number_of_vertices)] for k in range(self.number_of_triangles): vertex_triangles[triangles[k, 0]].append(k) vertex_triangles[triangles[k, 1]].append(k) vertex_triangles[triangles[k, 2]].append(k) vertex_triangles = map(frozenset, vertex_triangles) return vertex_triangles def nth_ring(self, vertex, neighbourhood=2, contains=False): """ Return the vertices of the nth ring around a given vertex, defaults to neighbourhood=2. NOTE: if you want neighbourhood=1 then you should directly access the property vertex_neighbours, ie use surf_obj.vertex_neighbours[vertex] setting contains=True returns all vertices from rings 1 to n inclusive. """ ring = set([vertex]) local_vertices = set([vertex]) for _ in range(neighbourhood): neighbours = [self.vertex_neighbours[indx] for indx in ring] neighbours = set(vert for subset in neighbours for vert in subset) ring = neighbours.difference(local_vertices) local_vertices.update(ring) if contains: local_vertices.discard(vertex) return frozenset(local_vertices) return frozenset(ring) def compute_triangle_normals(self): """Calculates triangle normals.""" tri_u = self.vertices[self.triangles[:, 1], :] - self.vertices[self.triangles[:, 0], :] tri_v = self.vertices[self.triangles[:, 2], :] - self.vertices[self.triangles[:, 0], :] tri_norm = numpy.cross(tri_u, tri_v) try: self.triangle_normals = tri_norm / numpy.sqrt(numpy.sum(tri_norm ** 2, axis=1))[:, numpy.newaxis] except FloatingPointError: # TODO: NaN generation would stop execution, however for normals this case could maybe be # handled in a better way. self.triangle_normals = tri_norm util.log_debug_array(LOG, self.triangle_normals, "triangle_normals", owner=self.__class__.__name__) def compute_vertex_normals(self): """ Estimates vertex normals, based on triangle normals weighted by the angle they subtend at each vertex... """ vert_norms = numpy.zeros((self.number_of_vertices, 3)) bad_normal_count = 0 # todo: vectorize this for k in range(self.number_of_vertices): try: tri_list = list(self.vertex_triangles[k]) angle_mask = self.triangles[tri_list, :] == k angles = self.triangle_angles[tri_list, :] angles = angles[angle_mask][:, numpy.newaxis] angle_scaling = angles / numpy.sum(angles, axis=0) vert_norms[k, :] = numpy.mean(angle_scaling * self.triangle_normals[tri_list, :], axis=0) # Scale by angle subtended. vert_norms[k, :] = vert_norms[k, :] / numpy.sqrt(numpy.sum(vert_norms[k, :] ** 2, axis=0)) # Normalise to unit vectors. except (ValueError, FloatingPointError): # If normals are bad, default to position vector # A nicer solution would be to detect degenerate triangles and ignore their # contribution to the vertex normal vert_norms[k, :] = self.vertices[k] / numpy.sqrt(self.vertices[k].dot(self.vertices[k])) bad_normal_count += 1 if bad_normal_count: self.logger.warn(" %d vertices have bad normals" % bad_normal_count) util.log_debug_array(LOG, vert_norms, "vertex_normals", owner=self.__class__.__name__) self.vertex_normals = vert_norms @property def triangle_areas(self): """An array specifying the area of the triangles making up a surface.""" if self._triangle_areas is None: self._triangle_areas = self._find_triangle_areas() return self._triangle_areas def _find_triangle_areas(self): """Calculates the area of triangles making up a surface.""" tri_u = self.vertices[self.triangles[:, 1], :] - self.vertices[self.triangles[:, 0], :] tri_v = self.vertices[self.triangles[:, 2], :] - self.vertices[self.triangles[:, 0], :] tri_norm = numpy.cross(tri_u, tri_v) triangle_areas = numpy.sqrt(numpy.sum(tri_norm ** 2, axis=1)) / 2.0 triangle_areas = triangle_areas[:, numpy.newaxis] util.log_debug_array(LOG, triangle_areas, "triangle_areas", owner=self.__class__.__name__) return triangle_areas @property def triangle_centres(self): """ An array specifying the location of triangle centres. """ if self._triangle_centres is None: self._triangle_centres = self._find_triangle_centres() return self._triangle_centres def _find_triangle_centres(self): """ Calculate the location of the centre of all triangles comprising the mesh surface. """ tri_verts = self.vertices[self.triangles, :] tri_centres = numpy.mean(tri_verts, axis=1) util.log_debug_array(LOG, tri_centres, "tri_centres") return tri_centres @property def triangle_angles(self): """ An array containing the inner angles for each triangle, same shape as triangles. """ if self._triangle_angles is None: self._triangle_angles = self._find_triangle_angles() return self._triangle_angles def _normalized_edge_vectors(self): """ for triangle abc computes the normalized vector edges b-a c-a c-b """ tri_verts = self.vertices[self.triangles] tri_verts[:, 2, :] -= tri_verts[:, 0, :] tri_verts[:, 1, :] -= tri_verts[:, 0, :] tri_verts[:, 0, :] = tri_verts[:, 2, :] - tri_verts[:, 1, :] # normalize tri_verts /= numpy.sqrt(numpy.sum(tri_verts ** 2, axis=2, keepdims=True)) return tri_verts def _find_triangle_angles(self): """ Calculates the inner angles of all the triangles which make up a surface """ def _angle(a, b): """ Angle between normalized vectors. <a|b> = cos(alpha)""" return numpy.arccos(numpy.sum(a * b, axis=1, keepdims=True)) edges = self._normalized_edge_vectors() a0 = _angle(edges[:, 1, :], edges[:, 2, :]) a1 = _angle(edges[:, 0, :], - edges[:, 1, :]) a2 = 2 * numpy.pi - a0 - a1 angles = numpy.hstack([a0, a1, a2]) util.log_debug_array(LOG, angles, "triangle_angles", owner=self.__class__.__name__) return angles @property def edges(self): """ A sorted list of the two element tuples(vertex_0, vertex_1) representing the edges of the mesh. """ if self._edges is None: self._edges = self._find_edges() return self._edges def _find_edges(self): """ Find all the edges of the mesh surface, return them sorted as a list of two element tuple, where the elements are vertex indices. """ v0 = numpy.vstack((self.triangles[:, 0][:, numpy.newaxis], self.triangles[:, 0][:, numpy.newaxis], self.triangles[:, 1][:, numpy.newaxis])) v1 = numpy.vstack((self.triangles[:, 1][:, numpy.newaxis], self.triangles[:, 2][:, numpy.newaxis], self.triangles[:, 2][:, numpy.newaxis])) edges = numpy.hstack((v0, v1)) edges.sort(axis=1) edges = set(tuple(edges[k]) for k in range(edges.shape[0])) edges = sorted(edges) return edges @property def number_of_edges(self): """ The number of edges making up the mesh surface. """ if self._number_of_edges is None: self._number_of_edges = len(self.edges) return self._number_of_edges @property def edge_lengths(self): """ The length of the edges defined in the ``edges`` attribute. """ if self._edge_lengths is None: self._edge_lengths = self._find_edge_lengths() return self._edge_lengths def _find_edge_lengths(self): """ Calculate the Euclidean distance between the pair of vertices that define the edges in the ``edges`` attribute. """ # TODO: Would a Sparse matrix be a more useful data structure for these??? elem = numpy.sqrt(((self.vertices[self.edges, :][:, 0, :] - self.vertices[self.edges, :][:, 1, :]) ** 2).sum(axis=1)) self.edge_mean_length = float(elem.mean()) self.edge_min_length = float(elem.min()) self.edge_max_length = float(elem.max()) return elem @property def edge_length_mean(self): """The mean length of the edges of the mesh.""" if self.edge_mean_length is None: self._find_edge_lengths() return self.edge_mean_length @property def edge_length_min(self): """The length of the shortest edge in the mesh.""" if self.edge_min_length is None: self._find_edge_lengths() return self.edge_min_length @property def edge_length_max(self): """The length of the longest edge in the mesh.""" if self.edge_max_length is None: self._find_edge_lengths() return self.edge_max_length @property def edge_triangles(self): """ List of the pairs of triangles sharing an edge. """ if self._edge_triangles is None: self._edge_triangles = self._find_edge_triangles() return self._edge_triangles def _find_edge_triangles(self): triangles = [None] * self.number_of_edges for k in range(self.number_of_edges): triangles[k] = (frozenset(self.vertex_triangles[self.edges[k][0]]) & frozenset(self.vertex_triangles[self.edges[k][1]])) return triangles def compute_topological_constants(self): """ Returns a 4 tuple: * the Euler characteristic number * indices for any isolated vertices * indices of edges where the surface is pinched * indices of edges that border holes in the surface We call isolated vertices those who do not belong to at least 3 triangles. """ euler = self.number_of_vertices + self.number_of_triangles - self.number_of_edges triangles_per_vertex = numpy.array(map(len, self.vertex_triangles)) isolated = numpy.nonzero(triangles_per_vertex < 3) triangles_per_edge = numpy.array(map(len, self.edge_triangles)) pinched_off = numpy.nonzero(triangles_per_edge > 2) holes = numpy.nonzero(triangles_per_edge < 2) return euler, isolated[0], pinched_off[0], holes[0] def validate_topology_for_simulations(self): """ Validates if this surface can be used in simulations. The surface should be topologically equivalent to one or two closed spheres. It should not contain isolated vertices. It should not be pinched or have holes: all edges must belong to 2 triangles. The allowance for one or two closed surfaces is because the skull/etc should be represented by a single closed surface and we typically represent the cortex as one closed surface per hemisphere. :return: a ValidationResult """ r = ValidationResult() euler, isolated, pinched_off, holes = self.compute_topological_constants() # The Euler characteristic for a 2D sphere embedded in a 3D space is 2. # This should be 2 or 4 -- meaning one or two closed topologically spherical surfaces if euler not in (2, 4): r.add_warning("Topologically not 1 or 2 spheres.", "Euler characteristic: " + str(euler)) if len(isolated): r.add_warning("Has isolated vertices.", "Offending indices: \n" + str(isolated)) if len(pinched_off): r.add_warning("Surface is pinched off.", "These are edges with more than 2 triangles: \n" + str(pinched_off)) if len(holes): r.add_warning("Has holes.", "Free boundaries: \n" + str(holes)) return r def laplace_beltrami(self, fv, h=1.0): """ Evaluates the discrete Laplace-Beltrami operator for a given vertex-wise function and geodesic distance matrix. From Belkin 2008: Let K be a mesh in R^3, with V its set of vertices. For face t in K, the number of vertices in t is #t, and V(t) is the set of vertices in t. For a function f : V -> R, this produces another function L_K^h f : V -> R, and L_K^h is computed for any w in V, L_K^h f (w) = 1 / (4 pi h^2) sum_{t in K} area(t) / #t sum_{p in V(t)} exp(-||p - w||^2/(4*h)) (f(p) - f(w)) :param fv: a function evaluated on each vertex, shape (n, ) :return: matrix of evaluated L-B operator """ from math import exp, pi assert fv.shape[0] == self.vertices.shape[0] assert hasattr(self, 'geodesic_distance_matrix') lbo = numpy.zeros_like(fv) gd = self.geodesic_distance_matrix for w in range(len(self.vertices)): mesh_sum = 0.0 for t in range(len(self.triangles)): face_sum = 0.0 for p in self.triangles[t]: face_sum += exp(-gd[w, p] ** 2 / (4 * h)) * (fv[p] - fv[w]) face_sum *= self.triangle_areas[t] / 3.0 mesh_sum += face_sum lbo[w] = mesh_sum / (4.0 * pi * h ** 2) return lbo # TODO def scientific_validate(self): self.number_of_vertices = self.vertices.shape[0] self.number_of_triangles = self.triangles.shape[0] if self.triangles.max() >= self.number_of_vertices: raise exceptions.ValidationException("There are triangles that index nonexistent vertices.") validation_result = self.validate_topology_for_simulations() self.valid_for_simulations = len(validation_result.warnings) == 0 return validation_result def compute_equation(self, focal_points, equation): """ focal_points - a list of focal points. Used for specifying the vertices from which the distance is calculated. equation - the equation which should be evaluated """ focal_points = numpy.array(focal_points, dtype=numpy.int32) dist = self.geodesic_distance(focal_points) equation.pattern = dist return equation.pattern # framework methods def load_from_metadata(self, meta_dictionary): self.edge_mean_length = 0 self.edge_min_length = 0 self.edge_max_length = 0 self.valid_for_simulations = True super(Surface, self).load_from_metadata(meta_dictionary) def get_vertices_slice(self, slice_number=0): """ Read vertices slice, to be used by WebGL visualizer. """ slice_number = int(slice_number) start_idx, end_idx = self._get_slice_vertex_boundaries(slice_number) return self.get_data('vertices', slice(start_idx, end_idx, 1)) def get_vertex_normals_slice(self, slice_number=0): """ Read vertex-normal slice, to be used by WebGL visualizer. """ slice_number = int(slice_number) start_idx, end_idx = self._get_slice_vertex_boundaries(slice_number) return self.get_data('vertex_normals', slice(start_idx, end_idx, 1)) def get_triangles_slice(self, slice_number=0): """ Read split-triangles slice, to be used by WebGL visualizer. """ if self.number_of_split_slices == 1: return self.triangles slice_number = int(slice_number) start_idx, end_idx = self._get_slice_triangle_boundaries(slice_number) return self.get_data('split_triangles', slice(start_idx, end_idx, 1)) def get_lines_slice(self, slice_number=0): """ Read the gl lines values for the current slice number. """ return self._triangles_to_lines(self.get_triangles_slice(slice_number)) def get_slices_to_hemisphere_mask(self): """ :return: a vector af length number_of_slices, with 1 when current chunk belongs to the Right hemisphere """ if not self.bi_hemispheric or self.split_slices is None: return None result = [1] * self.number_of_split_slices for key, value in self.split_slices.iteritems(): if value[KEY_HEMISPHERE] == HEMISPHERE_LEFT: result[int(key)] = 0 return result @staticmethod def _triangles_to_lines(triangles): lines_array = [] for a, b, c in triangles: lines_array.extend([a, b, b, c, c, a]) return numpy.array(lines_array) def framework_configure(self): """ Before storing Surface in DB, make sure vertices/triangles are split in slices that are readable by WebGL. WebGL only supports triangle indices in interval [0.... 2^16] """ # super(SurfaceFramework, self).configure() self.number_of_vertices = self.vertices.shape[0] self.number_of_triangles = self.triangles.shape[0] ### Do not split again, if split-data is already computed: if 1 < self.number_of_split_slices == len(self.split_slices): return ### Do not split when size is conveniently small: self.bi_hemispheric = self.hemisphere_mask is not None and numpy.unique(self.hemisphere_mask).size > 1 if self.number_of_vertices <= SPLIT_MAX_SIZE + SPLIT_BUFFER_SIZE and not self.bi_hemispheric: self.number_of_split_slices = 1 self.split_slices = {0: {KEY_TRIANGLES: {KEY_START: 0, KEY_END: self.number_of_triangles}, KEY_VERTICES: {KEY_START: 0, KEY_END: self.number_of_vertices}, KEY_HEMISPHERE: HEMISPHERE_UNKNOWN}} return ### Compute the number of split slices: left_hemisphere_slices = 0 left_hemisphere_vertices_no = 0 if self.bi_hemispheric: ## when more than one hemisphere right_hemisphere_vertices_no = numpy.count_nonzero(self.hemisphere_mask) left_hemisphere_vertices_no = self.number_of_vertices - right_hemisphere_vertices_no LOG.debug("Right %d Left %d" % (right_hemisphere_vertices_no, left_hemisphere_vertices_no)) left_hemisphere_slices = self._get_slices_number(left_hemisphere_vertices_no) self.number_of_split_slices = left_hemisphere_slices self.number_of_split_slices += self._get_slices_number(right_hemisphere_vertices_no) LOG.debug("Hemispheres Total %d Left %d" % (self.number_of_split_slices, left_hemisphere_slices)) else: ## when a single hemisphere self.number_of_split_slices = self._get_slices_number(self.number_of_vertices) LOG.debug("Start to compute surface split triangles and vertices") split_triangles = [] ignored_triangles_counter = 0 self.split_slices = {} for i in range(self.number_of_split_slices): split_triangles.append([]) if not self.bi_hemispheric: self.split_slices[i] = {KEY_VERTICES: {KEY_START: i * SPLIT_MAX_SIZE, KEY_END: min(self.number_of_vertices, (i + 1) * SPLIT_MAX_SIZE + SPLIT_BUFFER_SIZE)}, KEY_HEMISPHERE: HEMISPHERE_UNKNOWN} else: if i < left_hemisphere_slices: self.split_slices[i] = {KEY_VERTICES: {KEY_START: i * SPLIT_MAX_SIZE, KEY_END: min(left_hemisphere_vertices_no, (i + 1) * SPLIT_MAX_SIZE + SPLIT_BUFFER_SIZE)}, KEY_HEMISPHERE: HEMISPHERE_LEFT} else: self.split_slices[i] = {KEY_VERTICES: {KEY_START: left_hemisphere_vertices_no + (i - left_hemisphere_slices) * SPLIT_MAX_SIZE, KEY_END: min(self.number_of_vertices, left_hemisphere_vertices_no + SPLIT_MAX_SIZE * (i + 1 - left_hemisphere_slices) + SPLIT_BUFFER_SIZE)}, KEY_HEMISPHERE: HEMISPHERE_RIGHT} ### Iterate Triangles and find the slice where it fits best, based on its vertices indexes: for i in range(self.number_of_triangles): current_triangle = [self.triangles[i][j] for j in range(3)] fit_slice, transformed_triangle = self._find_slice(current_triangle) if fit_slice is not None: split_triangles[fit_slice].append(transformed_triangle) else: # triangle ignored, as it has vertices over multiple slices. ignored_triangles_counter += 1 continue final_split_triangles = [] last_triangles_idx = 0 ### Concatenate triangles, to be stored in a single HDF5 array. for slice_idx, split_ in enumerate(split_triangles): self.split_slices[slice_idx][KEY_TRIANGLES] = {KEY_START: last_triangles_idx, KEY_END: last_triangles_idx + len(split_)} final_split_triangles.extend(split_) last_triangles_idx += len(split_) self.split_triangles = numpy.array(final_split_triangles, dtype=numpy.int32) if ignored_triangles_counter > 0: LOG.warning("Ignored triangles from multiple hemispheres: " + str(ignored_triangles_counter)) LOG.debug("End compute surface split triangles and vertices " + str(self.split_slices)) def framework_validate(self): # First check if the surface has a valid number of vertices self.number_of_vertices = self.vertices.shape[0] self.number_of_triangles = self.triangles.shape[0] if self.number_of_vertices > TvbProfile.current.MAX_SURFACE_VERTICES_NUMBER: msg = "This surface has too many vertices (max: %d)." % TvbProfile.current.MAX_SURFACE_VERTICES_NUMBER msg += " Please upload a new surface or change max number in application settings." raise exceptions.ValidationException(msg) return ValidationResult() def get_urls_for_rendering(self, include_region_map=False, region_mapping=None): """ Compose URLs for the JS code to retrieve a surface from the UI for rendering. """ url_vertices = [] url_triangles = [] url_normals = [] url_lines = [] url_region_map = [] for i in range(self.number_of_split_slices): param = "slice_number=" + str(i) url_vertices.append(paths2url(self, 'get_vertices_slice', parameter=param, flatten=True)) url_triangles.append(paths2url(self, 'get_triangles_slice', parameter=param, flatten=True)) url_lines.append(paths2url(self, 'get_lines_slice', parameter=param, flatten=True)) url_normals.append(paths2url(self, 'get_vertex_normals_slice', parameter=param, flatten=True)) if not include_region_map or region_mapping is None: continue start_idx, end_idx = self._get_slice_vertex_boundaries(i) url_region_map.append(paths2url(region_mapping, "get_region_mapping_slice", flatten=True, parameter="start_idx=" + str(start_idx) + " ;end_idx=" + str(end_idx))) if include_region_map: return url_vertices, url_normals, url_lines, url_triangles, url_region_map return url_vertices, url_normals, url_lines, url_triangles def _get_slice_vertex_boundaries(self, slice_idx): if str(slice_idx) in self.split_slices: start_idx = max(0, self.split_slices[str(slice_idx)][KEY_VERTICES][KEY_START]) end_idx = min(self.split_slices[str(slice_idx)][KEY_VERTICES][KEY_END], self.number_of_vertices) return start_idx, end_idx else: LOG.warn("Could not access slice indices, possibly due to an incompatibility with code update!") return 0, min(SPLIT_BUFFER_SIZE, self.number_of_vertices) def _get_slice_triangle_boundaries(self, slice_idx): if str(slice_idx) in self.split_slices: start_idx = max(0, self.split_slices[str(slice_idx)][KEY_TRIANGLES][KEY_START]) end_idx = min(self.split_slices[str(slice_idx)][KEY_TRIANGLES][KEY_END], self.number_of_triangles) return start_idx, end_idx else: LOG.warn("Could not access slice indices, possibly due to an incompatibility with code update!") return 0, self.number_of_triangles @staticmethod def _get_slices_number(vertices_number): """ Slices are for vertices [SPLIT_MAX_SIZE * i ... SPLIT_MAX_SIZE * (i + 1) + SPLIT_BUFFER_SIZE] Slices will overlap : |........SPLIT_MAX_SIZE|...SPLIT_BUFFER_SIZE| <-- split 1 |......... SPLIT_MAX_SIZE|...SPLIT_BUFFER_SIZE| <-- split 2 If we have trailing data smaller than the SPLIT_BUFFER_SIZE, then we no longer split but we need to have at least 1 slice. """ slices_number, trailing = divmod(vertices_number, SPLIT_MAX_SIZE) if trailing > SPLIT_BUFFER_SIZE or (slices_number == 0 and trailing > 0): slices_number += 1 return slices_number def _find_slice(self, triangle): split_slices = self.split_slices # because of performance: 1.5 times slower without this mn = min(triangle) mx = max(triangle) for i in range(self.number_of_split_slices): v = split_slices[i][KEY_VERTICES] # extracted for performance slice_start = v[KEY_START] if slice_start <= mn and mx < v[KEY_END]: return i, [triangle[j] - slice_start for j in range(3)] return None, triangle ####################################### Split for Picking ####################################### def get_pick_vertices_slice(self, slice_number=0): """ Read vertices slice, to be used by WebGL visualizer with pick. """ slice_number = int(slice_number) slice_triangles = self.get_data('triangles', slice(slice_number * SPLIT_PICK_MAX_TRIANGLE, min(self.number_of_triangles, (slice_number + 1) * SPLIT_PICK_MAX_TRIANGLE))) result_vertices = [] for triang in slice_triangles: result_vertices.append(self.vertices[triang[0]]) result_vertices.append(self.vertices[triang[1]]) result_vertices.append(self.vertices[triang[2]]) return numpy.array(result_vertices) def get_pick_vertex_normals_slice(self, slice_number=0): """ Read vertex-normals slice, to be used by WebGL visualizer with pick. """ slice_number = int(slice_number) slice_triangles = self.get_data('triangles', slice(slice_number * SPLIT_PICK_MAX_TRIANGLE, min(self.number_of_triangles, (slice_number + 1) * SPLIT_PICK_MAX_TRIANGLE))) result_normals = [] for triang in slice_triangles: result_normals.append(self.vertex_normals[triang[0]]) result_normals.append(self.vertex_normals[triang[1]]) result_normals.append(self.vertex_normals[triang[2]]) return numpy.array(result_normals) def get_pick_triangles_slice(self, slice_number=0): """ Read triangles slice, to be used by WebGL visualizer with pick. """ slice_number = int(slice_number) no_of_triangles = (min(self.number_of_triangles, (slice_number + 1) * SPLIT_PICK_MAX_TRIANGLE) - slice_number * SPLIT_PICK_MAX_TRIANGLE) triangles_array = numpy.arange(no_of_triangles * 3).reshape((no_of_triangles, 3)) return triangles_array def get_urls_for_pick_rendering(self): """ Compose URLS for the JS code to retrieve a surface for picking. """ vertices = [] triangles = [] normals = [] number_of_split = self.number_of_triangles // SPLIT_PICK_MAX_TRIANGLE if self.number_of_triangles % SPLIT_PICK_MAX_TRIANGLE > 0: number_of_split += 1 for i in range(number_of_split): param = "slice_number=" + str(i) vertices.append(paths2url(self, 'get_pick_vertices_slice', parameter=param, flatten=True)) triangles.append(paths2url(self, 'get_pick_triangles_slice', parameter=param, flatten=True)) normals.append(paths2url(self, 'get_pick_vertex_normals_slice', parameter=param, flatten=True)) return vertices, normals, triangles def get_url_for_region_boundaries(self, region_mapping): return paths2url(self, 'generate_region_boundaries', datatype_kwargs={'region_mapping': region_mapping.gid}) def center(self): """ Compute the center of the surface as the mean spot on all the three axes. """ # is this different from return numpy.mean(self.vertices, axis=0) ? return [float(numpy.mean(self.vertices[:, 0])), float(numpy.mean(self.vertices[:, 1])), float(numpy.mean(self.vertices[:, 2]))] def generate_region_boundaries(self, region_mapping): """ Return the full region boundaries, including: vertices, normals and lines indices. """ boundary_vertices = [] boundary_lines = [] boundary_normals = [] array_data = region_mapping.array_data for slice_idx in range(self.number_of_split_slices): # Generate the boundaries sliced for the off case where we might overflow the buffer capacity slice_triangles = self.get_triangles_slice(slice_idx) slice_vertices = self.get_vertices_slice(slice_idx) slice_normals = self.get_vertex_normals_slice(slice_idx) first_index_in_slice = self.split_slices[str(slice_idx)][KEY_VERTICES][KEY_START] # These will keep track of the vertices / triangles / normals for this slice that have # been processed and were found as a part of the boundary processed_vertices = [] processed_triangles = [] processed_normals = [] for triangle in slice_triangles: triangle += first_index_in_slice # Check if there are two points from a triangles that are in separate regions # then send this to further processing that will generate the corresponding # region separation lines depending on the 3rd point from the triangle rt0, rt1, rt2 = array_data[triangle] if rt0 - rt1: reg_idx1, reg_idx2, dangling_idx = 0, 1, 2 elif rt1 - rt2: reg_idx1, reg_idx2, dangling_idx = 1, 2, 0 elif rt2 - rt0: reg_idx1, reg_idx2, dangling_idx = 2, 0, 1 else: continue lines_vert, lines_ind, lines_norm = self._process_triangle(triangle, reg_idx1, reg_idx2, dangling_idx, first_index_in_slice, array_data, slice_vertices, slice_normals) ind_offset = len(processed_vertices) / 3 processed_vertices.extend(lines_vert) processed_normals.extend(lines_norm) processed_triangles.extend([ind + ind_offset for ind in lines_ind]) boundary_vertices.append(processed_vertices) boundary_lines.append(processed_triangles) boundary_normals.append(processed_normals) return numpy.array([boundary_vertices, boundary_lines, boundary_normals]) @staticmethod def _process_triangle(triangle, reg_idx1, reg_idx2, dangling_idx, indices_offset, region_mapping_array, vertices, normals): """ Process a triangle and generate the required data for a region separation. :param triangle: the actual triangle as a 3 element vector :param reg_idx1: the first vertex that is in a 'conflicting' region :param reg_idx2: the second vertex that is in a 'conflicting' region :param dangling_idx: the third vector for which we know nothing yet. Depending on this we might generate a line, or a 3 star centered in the triangle :param indices_offset: to take into account the slicing :param region_mapping_array: the region mapping raw array for which the regions are computed :param vertices: the current vertex slice :param normals: the current normals slice """ def _star_triangle(point0, point1, point2, result_array): """ Helper function that for a given triangle generates a 3-way star centered in the triangle center """ center_vertex = [(point0[i] + point1[i] + point2[i]) / 3 for i in range(3)] mid_line1 = [(point0[i] + point1[i]) / 2 for i in range(3)] mid_line2 = [(point1[i] + point2[i]) / 2 for i in range(3)] mid_line3 = [(point2[i] + point0[i]) / 2 for i in range(3)] result_array.extend(center_vertex) result_array.extend(mid_line1) result_array.extend(mid_line2) result_array.extend(mid_line3) def _slice_triangle(point0, point1, point2, result_array): """ Helper function that for a given triangle generates a line cutting thtough the middle of two edges. """ mid_line1 = [(point0[i] + point1[i]) / 2 for i in range(3)] mid_line2 = [(point0[i] + point2[i]) / 2 for i in range(3)] result_array.extend(mid_line1) result_array.extend(mid_line2) # performance opportunity: we are computing some values available in caller p0 = vertices[triangle[reg_idx1] - indices_offset] p1 = vertices[triangle[reg_idx2] - indices_offset] p2 = vertices[triangle[dangling_idx] - indices_offset] n0 = normals[triangle[reg_idx1] - indices_offset] n1 = normals[triangle[reg_idx2] - indices_offset] n2 = normals[triangle[dangling_idx] - indices_offset] result_vertices = [] result_normals = [] dangling_reg = region_mapping_array[triangle[dangling_idx]] reg_1 = region_mapping_array[triangle[reg_idx1]] reg_2 = region_mapping_array[triangle[reg_idx2]] if dangling_reg != reg_1 and dangling_reg != reg_2: # Triangle is actually spanning 3 regions. Create a vertex in the center of the triangle, which connects to # the middle of each edge _star_triangle(p0, p1, p2, result_vertices) _star_triangle(n0, n1, n2, result_normals) result_lines = [0, 1, 0, 2, 0, 3] elif dangling_reg == reg_1: # Triangle spanning only 2 regions, draw a line through the middle of the triangle _slice_triangle(p1, p0, p2, result_vertices) _slice_triangle(n1, n0, n2, result_normals) result_lines = [0, 1] else: # Triangle spanning only 2 regions, draw a line through the middle of the triangle _slice_triangle(p0, p1, p2, result_vertices) _slice_triangle(n0, n1, n2, result_normals) result_lines = [0, 1] return result_vertices, result_lines, result_normals # TODO consider using just an enum on surface to indicate type, avoid excess classes.
[docs]class WhiteMatterSurface(Surface): """White matter - gray matter interface surface.""" __tablename__ = None __mapper_args__ = {'polymorphic_identity': WHITE_MATTER} _ui_name = "A white matter - gray surface" surface_type = basic.String(default=WHITE_MATTER)
[docs]class CorticalSurface(Surface): """Cortical or pial surface.""" _ui_name = "A cortical surface" surface_type = basic.String(default=CORTICAL, order=-1) __tablename__ = None __mapper_args__ = {'polymorphic_identity': CORTICAL}
[docs]class SkinAir(Surface): """Skin - air interface surface.""" __tablename__ = None __mapper_args__ = {'polymorphic_identity': OUTER_SKIN} _ui_name = "Skin" surface_type = basic.String(default=OUTER_SKIN) @classmethod def from_file(cls, source_file="outer_skin_4096.zip", instance=None): return super(SkinAir, cls).from_file(source_file, instance)
[docs]class BrainSkull(Surface): """Brain - inner skull interface surface.""" __tablename__ = None __mapper_args__ = {'polymorphic_identity': INNER_SKULL} _ui_name = "Brain - inner skull interface surface." surface_type = basic.String(default=INNER_SKULL) @classmethod def from_file(cls, source_file="inner_skull_4096.zip", instance=None): return super(BrainSkull, cls).from_file(source_file, instance)
[docs]class SkullSkin(Surface): """Outer-skull - scalp interface surface.""" __tablename__ = None __mapper_args__ = {'polymorphic_identity': OUTER_SKULL} _ui_name = "Outer-skull - scalp interface surface" surface_type = basic.String(default=OUTER_SKULL) @classmethod def from_file(cls, source_file="outer_skull_4096.zip", instance=None): return super(SkullSkin, cls).from_file(source_file, instance)
[docs]class OpenSurface(Surface): """Base class for open surfaces.""" __tablename__ = None
[docs]class EEGCap(OpenSurface): """EEG cap surface.""" __mapper_args__ = {'polymorphic_identity': EEG_CAP} _ui_name = "EEG Cap" surface_type = basic.String(default=EEG_CAP) @classmethod def from_file(cls, source_file="scalp_1082.zip", instance=None): return super(EEGCap, cls).from_file(source_file, instance)
[docs]class FaceSurface(OpenSurface): """Face surface.""" __mapper_args__ = {'polymorphic_identity': FACE} _ui_name = "Face surface" surface_type = basic.String(default=FACE) @classmethod def from_file(cls, source_file="face_8614.zip", instance=None): return super(FaceSurface, cls).from_file(source_file, instance)
[docs]def make_surface(surface_type): """ Build a Surface instance, based on an input type :param surface_type: one of the supported surface types :return: Instance of the corresponding surface lass, or None """ if surface_type in [CORTICAL, "Pial"] or surface_type.startswith("Cortex"): return CorticalSurface() elif surface_type == INNER_SKULL: return BrainSkull() elif surface_type == OUTER_SKULL: return SkullSkin() elif surface_type in [OUTER_SKIN, "SkinAir"]: return SkinAir() elif surface_type == EEG_CAP: return EEGCap() elif surface_type == FACE: return FaceSurface() elif surface_type == WHITE_MATTER: return WhiteMatterSurface() return None
[docs]def center_vertices(vertices): """ Centres the vertices using means along axes. :param vertices: a numpy array of shape (n, 3) :returns: the centered array """ return vertices - numpy.mean(vertices, axis=0).reshape((1, 3))
ALL_SURFACES_SELECTION = [{'name': 'Cortical Surface', 'value': CORTICAL}, {'name': 'Brain Skull', 'value': INNER_SKULL}, {'name': 'Skull Skin', 'value': OUTER_SKULL}, {'name': 'Skin Air', 'value': OUTER_SKIN}, {'name': 'EEG Cap', 'value': EEG_CAP}, {'name': 'Face Surface', 'value': FACE}, {'name': 'White Matter Surface', 'value': WHITE_MATTER}]