Public Types | Public Member Functions | Static Public Attributes
C2DPerfusionAnalysis Class Reference

A class to run an ICA on a heart perfusion series. More...

#include <mia/2d/perfusion.hh>

Public Types

enum  EBoxSegmentation { bs_delta_feature, bs_delta_peak, bs_features, bs_unknown }
 Possible bases for LV-RV heart segmentation. More...
 

Public Member Functions

 C2DPerfusionAnalysis (size_t components, bool normalize, bool meanstrip)
 
P2DFilter get_crop_filter (float scale, C2DBounds &crop_start, EBoxSegmentation approach, const std::string &save_features="") const
 
P2DImage get_feature_image (int index) const
 
int get_LV_idx () const
 
int get_LV_peak_idx () const __attribute__((deprecated))
 
int get_LV_peak_time () const
 
int get_perfusion_idx () const
 
std::vector< C2DFImageget_references () const
 
int get_RV_idx () const
 
int get_RV_peak_idx () const __attribute__((deprecated))
 
int get_RV_peak_time () const
 
bool has_movement () const
 
bool run (const std::vector< C2DFImage > &series)
 
void save_coefs (const std::string &coefs_name) const
 
void save_feature_images (const std::string &base_name) const
 
void set_approach (size_t approach)
 
void set_max_ica_iterations (size_t maxiter)
 
void set_use_guess_model ()
 
 ~C2DPerfusionAnalysis ()
 

Static Public Attributes

static TDictMap< EBoxSegmentationsegmethod_dict
 

Detailed Description

A class to run an ICA on a heart perfusion series.

This class provides the tools for ICA based 2D perfusion image series. This class is specifically designed for the analysis of free breathingly aquired myocardial perfusion images.

Definition at line 41 of file perfusion.hh.

Member Enumeration Documentation

Possible bases for LV-RV heart segmentation.

Enumerator
bs_delta_feature 

Segmentation based on the difference of the LV and RV feature images

bs_delta_peak 

Segmentation based on the difference of the LV and RV peak enhancenemt images

bs_features 

Segmentation based on the LV and RV feature images

bs_unknown 

place holder

Definition at line 44 of file perfusion.hh.

Constructor & Destructor Documentation

C2DPerfusionAnalysis::C2DPerfusionAnalysis ( size_t  components,
bool  normalize,
bool  meanstrip 
)

Constructor

Parameters
componentsnumber of independend components, 0 = auto estimate from [3,4,5,6,7]
normalizenormalize feature images
meanstripstrip mean from mixing time curves
C2DPerfusionAnalysis::~C2DPerfusionAnalysis ( )

Member Function Documentation

P2DFilter C2DPerfusionAnalysis::get_crop_filter ( float  scale,
C2DBounds crop_start,
EBoxSegmentation  approach,
const std::string &  save_features = "" 
) const

Evaluate an image cropping filter. This code is specifically designed to deal with the segmentantion of the left heart ventricle in short axis heart MRI The algorithm evaluates the centers of the LV and the RV and uses the distance between both to estimata a bounding box. Some heuristics are used to check whether the segmentation makes sense

Parameters
scaleenlargement scale of the bounding box to create the cropping region
[out]crop_startreturns the left upper corner of the cropping region that can be used to adjust segmentations
approachon what input data to base thesegmentation on
save_featuresif not empty store feature images in files with this prefix
Returns
the cropping filter or C2DFilterPlugin::ProductPtr() if the segmentation fails.
P2DImage C2DPerfusionAnalysis::get_feature_image ( int  index) const
Parameters
indexof the feature image requested, set negative to request the mean image.
Returns
the requested feature image
int C2DPerfusionAnalysis::get_LV_idx ( ) const
Returns
the LV enhancement IC index of -1 if it could not be identified
int C2DPerfusionAnalysis::get_LV_peak_idx ( ) const
Returns
the LV peak enhancement IC index of -1 if it could not be identified
int C2DPerfusionAnalysis::get_LV_peak_time ( ) const
Returns
the LV maximum peak enhancement time index, or -1 if not identified
int C2DPerfusionAnalysis::get_perfusion_idx ( ) const
Returns
the perfusion enhancement IC index of -1 if it could not be identified
std::vector<C2DFImage> C2DPerfusionAnalysis::get_references ( ) const

Create uncropped reference images that try to omit the movement component in the image series.

int C2DPerfusionAnalysis::get_RV_idx ( ) const
Returns
the RV enhancement IC index of -1 if it could not be identified
int C2DPerfusionAnalysis::get_RV_peak_idx ( ) const
Returns
the RV peak enhancement IC index of -1 if it could not be identified
int C2DPerfusionAnalysis::get_RV_peak_time ( ) const
Returns
the RV maximum peak enhancement time index, or -1 if not identified
bool C2DPerfusionAnalysis::has_movement ( ) const
Returns
true if a periodic component could be identified in the given series
bool C2DPerfusionAnalysis::run ( const std::vector< C2DFImage > &  series)

Run the ICA analysis - keeps a copy of the image series

Parameters
seriesimage series should contain more images thennumber of requested components
void C2DPerfusionAnalysis::save_coefs ( const std::string &  coefs_name) const

Save the mixin matrix to a file.

Parameters
coefs_nameoutput file name
void C2DPerfusionAnalysis::save_feature_images ( const std::string &  base_name) const

Save the feature image to some PNG files.

Parameters
base_nameoutput file name base
void C2DPerfusionAnalysis::set_approach ( size_t  approach)

Set the ICA seperation approach

Parameters
approachFICA_APPROACH_SYMM or FICA_APPROACH_DEFL
Todo:
the parameter should be an enum
void C2DPerfusionAnalysis::set_max_ica_iterations ( size_t  maxiter)

Set the number of ICA iterations

Parameters
maxiter
void C2DPerfusionAnalysis::set_use_guess_model ( )

Use an experimental model to create a initial guess.

Field Documentation

TDictMap<EBoxSegmentation> C2DPerfusionAnalysis::segmethod_dict
static

Dictionary for segmentation method flags

Definition at line 141 of file perfusion.hh.


The documentation for this class was generated from the following file: