This chapter provides a list of tips that address various SPHARM-MAT issues, including strengths, limitations, parameter settings, options, etc.
Matlab starts navigation for file selection from the current directory. Thus, to minimize the effort of finding your data, you can change matlab current directory to your data directory if you want to work on the same data for a while.
We notice that the ADC values of SPHARM-PDM results are often extremely high; see Surface Visualization (ADC_ParaMap, Solid with Mesh) for an example. Some relevant discussion is provided below.
Input
Parameterization using CALD as Method can take a binary object, a triangular mesh, or a quadrilateral mesh as an input. For a binary object, SPHARM-MAT extracts its voxel surface and describes it as a triangular mesh. For a quadrilateral mesh, SPHARM-MAT converts it to a triangular mesh. In any case, the resulting triangular mesh needs to be a genus zero surface (i.e., with a spherical topology). If you work with binary images, be sure to run topology fix first (see Data Preparation).
Output
Results of both initial parameterization and optimized parameterization are saved. See Surface Visualization (ADC_ParaMap, Solid with Mesh) for an example.
Spherical parameterization using CALD has a few adjustable parameters shown below. In most cases, you can just use the default setting; see Exercise 3.1 Surface Meshes (CALD) and Exercise 3.2 Voxel Surfaces (CALD).
Parameterization using PDM as Method takes a binary object as input. The surface of the binary object needs to be genus zero (i.e., with a spherical topology). To meet this requirement, be sure to run topology fix first (see Data Preparation).
Parameterization using PDM as Method has a few adjustable parameters shown below. In most cases, you can just use the default setting; see Exercise 3.3 Voxel Surfaces (PDM).
See also
SPHARM-PDM manual for further details about these parameters.
iter: Max number of iterations (default 500)
label: Binary volume is formed by all the voxels with value equal to the specified label (default 1)
The SPHARM coefficients up to a user-desired degree are computed in SPHARM expansion step. The object surface can be reconstructed using these coefficients, and using more coefficients leads to a more detailed reconstruction; see Surface Visualization (icosa4, Solid with Mesh).
SPHARM Expansion has the following parameter for users to specify the maximum degree; see Exercise 4.1 SPHARM-MAT Expansion.
The SPHARM coefficients computed by SPHARM-MAT (see Exercise 4.1 SPHARM-MAT Expansion) are complex numbers. The SPHARM coefficients computed by SPHARM-PDM (see Exercise 4.2 SPHARM-PDM Expansion) are real numbers, where the imaginary part of each coefficient is ignored.
Alignment using FOE has a few adjustable parameters shown below. In most cases, you can just use the default setting; see Exercise 5.1 FOE Alignment.
Alignment using SHREC has a few adjustable parameters shown below. In most cases, you can just use the default setting; see Exercise 5.2 SHREC Alignment.
Initial SPHARM models for SHREC alignment: If the FOE of your data is a real ellipsoid having three distinct axes (e.g., like hippocampus), you may want to run FOE alignment first before doing SHREC. Note that the final results of FOE alignment do not preserve the original geometric information in the object space. Therefore, be sure to apply SHREC only to the intermediate results (i.e., *_prm.mat files in Exercise 5.2 SHREC Alignment) of FOE alignment. For these intermediate results, the orientation and the location of the objects have not been transformed in the object space; however, their parameter nets have been aligned to the canonical position so that an initial surface correspondence has been established among then.
In SHREC alignment, a template needs to be specified so that all the individual SPHARM models can be registered to it. This template should be carefully picked or prepared. For a case-control group study, it could be an individual control object or the mean shape of a set of typical control objects (see Exercise 7.2 Average Objects). In addition, its underlying parameterization should be oriented to a canonical position. If the FOE of the template is a real ellipsoid, FOE alignment can be applied to achieve this goal, as described in Alignment SHREC Initial Models; otherwise, manual adjustment could be used to rotate the parameter net to a user-desired position.
The Expand & Align button is provided for performing SPHARM expansion and FOE alignment using SPHARM-PDM. It has a few adjustable parameters shown below. In most cases, you can just use the default setting; see Exercise 4.2 SPHARM-PDM Expansion and Exercise 5.3 SPHARM-PDM FOE Alignment.
See also
SPHARM-PDM manual for further details about these parameters.
flipTemplate: Set a fliptemplate for normalization of axis-flips (could be empty)
subdivLevel: Set subdivision level for linear icosahedron subdivision
spharmDegree: Set the degree of spherical harmonic expansion
regTemplate: Specify a registration template (could be empty)
Statistical analysis using t-map as method has a few adjustable parameters shown below. In most cases, you can just use the default setting; see Exercise 6.1 T Test.
Statistical analysis using PCA as method has a few adjustable parameters shown below. In most cases, you can just use the default setting; see Exercise 6.2 PCA.
DisplayRes using res-t-map as method has a few adjustable parameters shown below. In most cases, you can just use the default setting; see Exercise 6.1 T Test.
DisplayRes using res-PCA as method has a few adjustable parameters shown below. In most cases, you can just use the default setting; see Exercise 6.2 PCA.
Import utility does the following conversion (see Exercise 7.1 Import)
Import has the following parameter to adjust
SPHARM-MAT uses matlab format to store 3D binary objects (before topology fix: *_bim.mat; after topology fix: *_fix.mat). SPHARM-PDM uses gipl format to store 3D binary objects (*.gipl). FormatConvert under the Utils pop-up menu can be used for performing format conversion with the following three options:
SPHARM-MAT uses matlab format to store surfaces (original: *_obj.mat; after initial parameterization: *_fix.mat; after parameterization optimization: *_smo.mat). SPHARM-PDM uses META format to store object meshes (*_surf.meta) and parameter meshes (*_para.meta). FormatConvert under the Utils pop-up menu can be used for performing conversion between these formats:
SPHARM-MAT uses matlab format to store SPHARM models:
SPHARM-PDM uses META format (i.e., surface meshes) and COEF format (i.e., SPHARM coefficients) to store SPHARM models:
FormatConvert under the Utils pop-up menu can be used for performing conversion between these formats:
TopologyFix under the Utils pop-up menu has two options for customizing 3D hole filing process if Inhouse_Fix is selected as Method:
TopologyFix under the Utils pop-up menu has two options for customizing 3D hole filing process if PDM_Fix is selected as Method:
DisplayObjs under the Utils pop-up menu has three options for selection of visualizing the surface in object and/or parameter spaces:
DisplayObjs under the Utils pop-up menu has a few options for selecting mesh structures:
DisplayObjs under the Utils pop-up menu has three options for shading the surface:
DisplayObjs under the Utils pop-up menu has an option for showing additional information
none: No additional information is shown.
DisplayObjs under the Utils pop-up menu has three options for exporting the figure:
DisplayObjs under the Utils pop-up menu has an option called Degree for specifying the user-desired degree for SPHARM reconstruction. If empty, all the available coefficients are used for SPHARM reconstruction; otherwise, the reconstructed surface is generated only using those coefficients up to the user-specified degree. See Surface Visualization (icosa4, Solid with Mesh).
DisplayObjs under the Utils pop-up menu has an option called Template for specifying a template. This is often used when you want to display registered SPHARM models. If the template file exists, the root mean square distance between each input SPHARM model and the specified template model will be calculated and displayed on the z-axis. See Surface Visualization (quad64, template).
AverageObjs under the Utils pop-up menu aims to create an average surface, which can be used as the atlas in group analysis. In a typical study, you can select a set of control objects and run this utility to create an atlas that represents an average normal surface. The method is the following: (1) let the atlas be the first model; (2) align each model to the atlas using SHREC; (3) let the atlas be the mean of all the data; and (4) repeat (2) and (3) until the atlas converges. See Surface Visualization (icosa4): for an example atlas. Once the atlas is created, you can use SHREC to register all the individual objects to the atlas (instead of using an individual object as the template, as shown in Exercise 5.2 SHREC Alignment). AverageObjs has the following parameter to adjust
ScaleObjs under the Utils pop-up menu can be used to scale a set of SPHARM objects. The scaling factors need to be pre-computed by users and provided as a csv file as an input file. See Exercise 7.2 Average Objects: for an example. ScaleObjs has the following parameter to adjust