Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision Both sides next revision
development:fwdarch [2018/10/21 15:02]
42.49.180.224 [Refurbishing the FORWARD module]
development:fwdarch [2017/08/17 11:21] (current)
127.0.0.1 external edit
Line 47: Line 47:
  
  
-========Refurbishing ​the FORWARD module ========+======= ​Architecture:​ defining a table for the possible implementations ​=======
  
-The forward module routines generate leadfields ​which are used used in different contexts (e.g. beamformer). To build the forward model we make use of various methods and we start from very different geometrical descriptions,​ generally representing ​the head shape.\\+The table contains the cells that indicate **FROM** ​which kind of input **TO** which kind of method I want to goThe content is a set of programmatic steps which leads from a specific anatomical information (see the first columnto the costruction of the '​vol'​ structure for a forward model method (see the first row of the table).
  
-Let's list some of the the methods currently supported by FieldTrip (in parentheses which modality they are used for):\\ +The tables ​are different ​for EEG and MEG:
-  * ASA bem (EEG+
-  * Bemcp (EEG) +
-  * Dipoli bem (EEG) +
-  * OpenMEEG bem (EEG/MEG+
-  * MNE bem (do we want to adopt this?) +
-  * halfspace medium (EEG) +
-  * Infinite medium (EEG/MEG) +
-  * Infinite slab (EEG) +
-  * Local spheres (EEG/MEG) +
-  * Single shell (MEG) +
-  * Concentric spheres (EEG)+
  
-According to the different anatomy acquisition techniques it is possible to distinguish among the following types of input:\\ +* For EEG:
-  * CT scan +
-  * MRI scan +
-  * Headshape, points +
-  * Headshape, closed surface +
-  * Nothing+
  
-All the routines which create ​forward model need description of the volume conductor model as inputThis model is represented in vol structure which contains ​set of fields, among which vol.method ​(e.gvol.method = '​dipoli'​)All other fields are specific for the various methods and are generated in pipeline which can be summarized in the following picture:+|  ^ Infinite ​ ^ Halfspace ​ ^ Slab ^ Single sphere ​ ^ Concentric spheres ^ FEM ^ FDM ^ ASA BEM ^ Bemcp ^ Dipoli BEM ^ OpenMEEG BEM ^  
 +^ 1.CT      |[[#INF1 method|INF]]|[[#​HS1 method|HS1]]|[[#​SL1 method|SL1]]|[[#​SPH1 method|SPH1]]|[[#​CS1 method|CS1]]|[[#​FEM1 method|FEM1]]|[[#​FDM1 method|FDM1]]|[[#​ASA1 method|ASA]] |[[#BEMCP1 method|BEMCP1]]|[[#​Dipoli1 method|DIP1]] |[[#​OpenMEEG1 method|OM1]]| 
 +^ 2.MRI      | [[#INF2 method|INF]]|[[#​HS2 method|HS2]]|[[#​SL2 method|SL2]]|[[#​SPH2 method|SPH2]]|[[#​CS2 method|CS2]]|[[#​FEM2 method|FEM2]]|[[#​FDM2 method|FDM2]]|[[#​ASA2 method|ASA]]|[[#​BEMCP2 method|BEMCP2]]|[[#​Dipoli2 method|DIP2]]|[[#​OpenMEEG2 method|OM2]]| 
 +^ 3.Headshape points ​   |[[#INF3 method|INF]]|[[#​HS3 method|HS3]]|[[#​SL3 method|SL3]] |[[#SPH3 method|SPH3]]|[[#​CS3 method|CS3]]|n.a.|n.a.|[[#ASA3 method|ASA]]|[[#​BEMCP3 method|BEMCP3]]|[[#​Dipoli3 method|DIP3]]|[[#​OpenMEEG3 method|OM3]]| 
 +^ 4.Headshape triangulation|[[#​INF4 method|INF]]|[[#​HS4 method|HS4]]|[[#​SL4 method|SL4]] |[[#SPH4 method|SPH4]]|[[#​CS4 method|CS4]]|n.a.|n.a.|[[#​ASA4 ​method|ASA]]|[[#​BEMCP4 method|BEMCP4]]|[[#​Dipoli4 method|DIP4]]|[[#​OpenMEEG4 method|OM4]]| 
 +^ 5.Nothing ​     |[[#INF5 method|INF]] |[[#HS5 method|HS5]]|[[#​SL5 method|SL5]]|[[#​SPH5 method|SPH5]]|n.a.|n.a.|n.a.|n.a. |n.a. |n.a. |n.a. |
  
-{{ :development:​wikie_scheme2.jpg?​nolink&​amp;​600 |}}+* For MEG:
  
-The leftmost box indicates ​set of functions (Matlab/​external toolboxes, FieldTrip) that deal with anatomical volumes (3D matrix filled with intensity values - a set of voxels).  +|  ^ Infinite ​ ^ Single sphere ^ Local spheres ^ Single shell ^ FEM ^ OpenMEEG BEM ^ 
-This step manages the volumetric information (MRI/CT scans) and transforms the anatomy into processed anatomy ('​ana2ana'​ functions), transforms the anatomy into binary segmentation ('​ana2seg'​ functions) or manipulates segmented images ('​seg2seg'​ functions, i.emorphology operators).+^ 1.CT           ​|[[#​INF1 method |INF]]|[[#​SPH1 method |SPH1]]|[[#​LS1 method|LS1]]|[[#​SH1 method|SH1]]| [[#FEM1 method|FEM1]]|[[#​OpenMEEG1 method|OM1]]| 
 +^ 2.MRI          |[[#INF2 method|INF]]|[[#​SPH2 method |SPH2]]|[[#​LS2 method|LS2]]|[[#​SH2 method|SH2]]| [[#FEM2 method|FEM2]]|[[#​OpenMEEG2 method|OM2]]| 
 +^ 3.Headshape points ​   |[[#INF3 method |INF]]|[[#​SPH3 method|SPH3]]|[[#​LS3 method|LS3]]|[[#​SH3 method|SH3]]|n.a.|[[#​OpenMEEG3 method|OM3]]|  
 +^ 4.Headshape triangulation|[[#​INF4 method |INF]]|[[#​SPH4 method|SPH4]]|[[#​LS4 method|LS4]]|[[#​SH4 method|SH4]]|n.a.|[[#​OpenMEEG4 method|OM4]]| 
 +^ 5.Nothing ​     |[[#INF5 method |INF]]|[[#​SPH5 method|SPH5]]|n.a.|n.a.| n.a.|n.a.|
  
-The content of the central box is a set of functions which refines the geometrical description derived from the volumetric analysis. Any visual quality control is processed in this section, including interactions with the operator. The input can be a segmented volume, a triangulated set of points or a cloud of points. The output is a geometry (geom), defined as a geometrical description other than volumetric, like for example a triangulated mesh, a cloud of points representing the head, a 3D lattice of tetrahedrons/​hexahedrons,​ etc.\\ 
  
-The third block (Post-processing) performs all automatic steps to build the description of the volume conductor ('​vol'​ structure with various fields). This last step provides the piece of information ​ which is fed to the forward solution algorithms ([[#The functions in Forward module|the various lead field computation routines]]). This post-processing functions are **not supposed to change the meshes anymore** and call method-specific head model functions (**[[#A list of the headmodel functions|ft_headmodel_XXX]]** routines) 
  
-As a general rule all volumetric geometry descriptions should not contain holes or be hollow. All mesh geometries (excluded 3D meshes for FEM) should not intersect with each other or be nested (apart for concentric spheres).\\ 
 ====== Overview on the methods ====== ====== Overview on the methods ======
 The following paragraphs describe in synthesis the pipelines for the various methods sketched in the tables. The methods are ordered according to the input type (e.g. CT scans) they use to generate the '​vol'​ structure. The following paragraphs describe in synthesis the pipelines for the various methods sketched in the tables. The methods are ordered according to the input type (e.g. CT scans) they use to generate the '​vol'​ structure.
Line 787: Line 777:
 All routines that create head models must allow the transition from all objects to all others (if possible), and from voxel based images to geometrical objects. All routines that create head models must allow the transition from all objects to all others (if possible), and from voxel based images to geometrical objects.
 For example the routine prepare_mesh_manual segments boundaries directly from the MRI scans and generates connected lines (if goes from ana to 1d objects).\\ For example the routine prepare_mesh_manual segments boundaries directly from the MRI scans and generates connected lines (if goes from ana to 1d objects).\\
 +
 +
 +
 +
 +
 +
 +