— Class GraphAnalyser More...

#include <mlGraphAnalyser.h>

Public Member Functions
void	sumupST_DrainVolume (bool bIgnoreOrientation=false)
	cumulate DrainVolumes given at the leafs

bool	calcDistanceMatrix (MatrixTemplate< double > &dm, std::vector< int > &vId, int nMaxNodes)
	Calculate Distance Matrix for rooted Graph, taking hierarchy into account: dm(n1, n2) := minimal distance between n1 and a conjoint parent node of n1 and n2 nMaxNodes: maximal Number of notes taken into account Nodes are cumulated in order of descending hierarchy to get a equably resolution dm: Matrix to store result, size is matched automatically.

bool	calcNodeClassifikator (CDoubleArray pArAssoc, std::vector< int > pvId, int nMaxNodes)
	stack Node classificators in (*pArAssoc) (n1,n1): classificators based on single Node n1 respect to global tree structure [0]: relative X position [1]: relative Y position [2]: relative Z position [3]: 0/1 left resp.

bool	calcNodeBaseFromPosition (const Vector3 &vPosition, const std::vector< int > &vId, std::vector< CIdxFloat > *pvResult)
	Calculate representation of spatial position in node coordinates.

bool	calcPositionFromNodeBase (std::vector< CIdxFloat > vBase, Vector3 *pvResult)
	Calculate spatial vector position form relative node coordinates.

double	calcSpannedVolume (const std::vector< int > &vId)
	Calculate Volume of minimal cube including all supplied nodes.

short	sampleConnectedCluster ()
	identify connected subgraph return value is number of identified cluster

void	tansferLabels (Skeleton::LABELTYPE NType, float fThresh=0.1f)
	transfer Vessel-Edge labels to Skeleton

void	tansferLabelsInverse (void)
	transfer skeleton labels to vessel edge label (take label of most frequent skeleton)

void	scanSkeletonLabelMinMax (float min, float max)
	evaluate min/max value of skeleton labels

void	collectLeafs (VesselNode pRoot, std::vector< VesselNode > &pLeafs)
	Collect all leafs which are children of node `pRoot` and append them to `pLeafs`.

Initialization
	GraphAnalyser (Graph *pGraph)
	constructor: provide pointer `pGraph` for easy access to Graph object

evaluation of Graph geometry
void	measureLength ()
	Get the euclidean length of all graph edges and set edge 'Length' property in the `Graph`.

void	measureVolume ()
	Get the euclidean volume of all graph edges and set edge 'Volume' property in the `Graph`.

void	measureST_Volume (bool bIgnoreOrientation=false)
	Get the euclidean volume of all graph edges connected with given root, nodes cumulate them according to orientation and set edge 'ST_Volume' property in the `Graph`.

float	measureAvMinDistance ()
	Get the average minimal distance over all skeletons of an edge and set edge 'averageMinDistance' property in the `Graph`.

void	measureSkeletonArea (int nAvLength=5)
	measure floating average of SkeletonArea and set area property

Vector3	measureBarycenter ()
	Get barycenter position of all the `Graph`.

void	measureST_Barycenter (bool bIgnoreOrientation=false)
	Get the euclidean volume of all graph edges connected with given root, nodes cumulate them according to orientation and set edge 'ST_Barycenter' property in the `Graph`.

void	sumupST_Value (VesselNode::NodeGetFkt getFkt, VesselNode::NodeSetFkt setFkt, bool bIgnoreOrientation=false)

void	measureMaxPathValue (VesselNode *pRootNode, VesselEdge::EdgeGetFkt edgeGetFkt, VesselNode::NodeGetFkt nodeGetFkt, VesselNode::NodeSetFkt nodeSetFkt)
	For each node calculate the maximum sum of edge values in a single path below this node. CAUTION: Works with directed, acyclic graphs only!

void	searchMaxSubTrees (std::vector< VesselEdge * >::iterator iBegin, std::vector< VesselEdge * >::iterator iEnd, int nNum, std::vector< VesselEdge * > pvMaxEdges, float pfActValue, float *pfActSum)
	calculate set of edges representing nNum subtrees maximizing selected feature stored in label property of the edges

void	searchMaxSubTrees (std::vector< VesselEdge * > vStartEdges, unsigned int nNum, std::vector< VesselEdge * > pvMaxEdges, float pfActValue, float *pfActSum)
	calculate set of edges representing nNum subtrees maximizing selected feature stored in label property of the edges

void	searchMaxSubTrees (std::vector< VesselEdge * > vStartEdges, int nPos, unsigned int nNum, std::vector< VesselEdge * > pvMaxEdges, float pfActValue, float *pfActSum)
	calculate set of edges representing nNum subtrees maximizing selected feature stored in label property of the edges

void	searchSubTreeSeparation (VesselNode pRoot, const Vector3 &separationDirection, std::vector< VesselEdge > &vClassifiedEdges)

void	searchST_MajorBifurcation (VesselEdge pStartEdge, float fMinorLimit, std::vector< VesselEdge > *pvEdges, bool bClear=true)
	search for first major bifurcation following main edge starting with pStartEdge, use label property to differentiate major and minor branches.

void	searchST_MajorBifurcation (VesselEdge pStartEdge, float fMinorLimitStart, float fMinorLimitStep, int nMinorLimitNumber, std::vector< VesselEdge > *pvEdges, bool bClear=true)
	search for first major bifurcation following main edge starting with pStartEdge, use label property to differentiate major and minor branches.

evaluation of Graph structure
VesselNode *	suggestRoot (float fSignificanceLevel=0.7f, int nPreferedDirection=0)
	Search for root-node: look for leaf with thick edge seated remote from the barycenter of the Graph.

const VesselNode *	getParent (const VesselNode pN1, const VesselNode pN2) const
	get pointer to common parent if exists or NULL

const VesselNode *	getParent (const VesselEdge pE1, const VesselEdge pE2) const

VesselNode *	getParent (VesselNode pN1, VesselNode pN2)

VesselNode *	getParent (VesselEdge pE1, VesselEdge pE2)

Detailed Description

— Class GraphAnalyser

class for user-friendly hosting of implemented methods to analyze graph structure:

measure euclidean length of all edges and set edge 'Length' property

Definition at line 148 of file mlGraphAnalyser.h.

Constructor & Destructor Documentation

◆ GraphAnalyser()

GraphAnalyser::GraphAnalyser ( Graph * pGraph )

inlineexplicit

constructor: provide pointer pGraph for easy access to Graph object

Definition at line 157 of file mlGraphAnalyser.h.

Member Function Documentation

◆ calcDistanceMatrix()

bool GraphAnalyser::calcDistanceMatrix	(	MatrixTemplate< double > &	dm,
		std::vector< int > &	vId,
		int	nMaxNodes )

Calculate Distance Matrix for rooted Graph, taking hierarchy into account: dm(n1, n2) := minimal distance between n1 and a conjoint parent node of n1 and n2 nMaxNodes: maximal Number of notes taken into account Nodes are cumulated in order of descending hierarchy to get a equably resolution dm: Matrix to store result, size is matched automatically.

◆ calcNodeBaseFromPosition()

bool GraphAnalyser::calcNodeBaseFromPosition	(	const Vector3 &	vPosition,
		const std::vector< int > &	vId,
		std::vector< CIdxFloat > *	pvResult )

Calculate representation of spatial position in node coordinates.

Coordinates are given by nadir and distance of the given point respective to the plane, defined by the tree nearest nodes.

◆ calcNodeClassifikator()

bool GraphAnalyser::calcNodeClassifikator	(	CDoubleArray *	pArAssoc,
		std::vector< int > *	pvId,
		int	nMaxNodes )

stack Node classificators in (*pArAssoc) (n1,n1): classificators based on single Node n1 respect to global tree structure [0]: relative X position [1]: relative Y position [2]: relative Z position [3]: 0/1 left resp.

right lung

(n1,n2): second order classificators based on association of two nodes n1 and n2 (n2 < n1) [0]: grade of node n1 being parent node [1]: grade of node n1 being parent node [2]: no conjoint path of n1 and n2 -> attribute 'parent node meaningless' [3]: euclidean distance of n1 and n2

At the moment the values used are not pure fuzzy classificator: they did not describe the peculiarity of a specific attribute like 'belong to upper left region' in the range 0..1. The aim is to reach second order classificator for the association matrix of two graphs based on relative similarity of the two graphs not biased by specifications of the single graph only.

◆ calcPositionFromNodeBase()

bool GraphAnalyser::calcPositionFromNodeBase	(	std::vector< CIdxFloat >	vBase,
		Vector3 *	pvResult )

Calculate spatial vector position form relative node coordinates.

The coordinates are interpreted in terms of plane coordinates and nadir distance with respect to three nodes supplied by index-nr.

◆ calcSpannedVolume()

double GraphAnalyser::calcSpannedVolume ( const std::vector< int > & vId )

Calculate Volume of minimal cube including all supplied nodes.

◆ collectLeafs()

void GraphAnalyser::collectLeafs	(	VesselNode *	pRoot,
		std::vector< VesselNode * > &	pLeafs )

Collect all leafs which are children of node pRoot and append them to pLeafs.

◆ getParent() [1/4]

const VesselNode * GraphAnalyser::getParent	(	const VesselEdge *	pE1,
		const VesselEdge *	pE2 ) const

◆ getParent() [2/4]

const VesselNode * GraphAnalyser::getParent	(	const VesselNode *	pN1,
		const VesselNode *	pN2 ) const

get pointer to common parent if exists or NULL

◆ getParent() [3/4]

VesselNode * GraphAnalyser::getParent	(	VesselEdge *	pE1,
		VesselEdge *	pE2 )

◆ getParent() [4/4]

VesselNode * GraphAnalyser::getParent	(	VesselNode *	pN1,
		VesselNode *	pN2 )

◆ measureAvMinDistance()

float GraphAnalyser::measureAvMinDistance ( )

Get the average minimal distance over all skeletons of an edge and set edge 'averageMinDistance' property in the Graph.

Return global maximum of distances.

◆ measureBarycenter()

Vector3 GraphAnalyser::measureBarycenter ( )

Get barycenter position of all the Graph.

◆ measureLength()

void GraphAnalyser::measureLength ( )

Get the euclidean length of all graph edges and set edge 'Length' property in the Graph.

◆ measureMaxPathValue()

void GraphAnalyser::measureMaxPathValue	(	VesselNode *	pRootNode,
		VesselEdge::EdgeGetFkt	edgeGetFkt,
		VesselNode::NodeGetFkt	nodeGetFkt,
		VesselNode::NodeSetFkt	nodeSetFkt )

For each node calculate the maximum sum of edge values in a single path below this node. CAUTION: Works with directed, acyclic graphs only!

◆ measureSkeletonArea()

void GraphAnalyser::measureSkeletonArea ( int nAvLength = 5 )

measure floating average of SkeletonArea and set area property

◆ measureST_Barycenter()

void GraphAnalyser::measureST_Barycenter ( bool bIgnoreOrientation = false )

Get the euclidean volume of all graph edges connected with given root, nodes cumulate them according to orientation and set edge 'ST_Barycenter' property in the Graph.

Parameters

bIgnoreOrientation state weather to use defined orientation of directed Graph or scan tree independent

◆ measureST_Volume()

void GraphAnalyser::measureST_Volume ( bool bIgnoreOrientation = false )

Get the euclidean volume of all graph edges connected with given root, nodes cumulate them according to orientation and set edge 'ST_Volume' property in the Graph.

Parameters

bIgnoreOrientation state weather to use defined orientation of directed Graph or scan tree independent

◆ measureVolume()

void GraphAnalyser::measureVolume ( )

Get the euclidean volume of all graph edges and set edge 'Volume' property in the Graph.

◆ sampleConnectedCluster()

short GraphAnalyser::sampleConnectedCluster ( )

identify connected subgraph return value is number of identified cluster

◆ scanSkeletonLabelMinMax()

void GraphAnalyser::scanSkeletonLabelMinMax	(	float *	min,
		float *	max )

evaluate min/max value of skeleton labels

◆ searchMaxSubTrees() [1/3]

void GraphAnalyser::searchMaxSubTrees	(	std::vector< VesselEdge * >	vStartEdges,
		int	nPos,
		unsigned int	nNum,
		std::vector< VesselEdge * > *	pvMaxEdges,
		float *	pfActValue,
		float *	pfActSum )

inline

calculate set of edges representing nNum subtrees maximizing selected feature stored in label property of the edges

Parameters

vStartEdges	base Edges to start search with
nPos	position of start edge
nNum	number of subtress to be selected
pvMaxEdges	set of Edges representing max subtrees
pfActValue	minimal feature value of currently selected subtrees
pfActSum	total feature sum of currently selected subtrees

Definition at line 229 of file mlGraphAnalyser.h.

References mlrange_cast().

◆ searchMaxSubTrees() [2/3]

void GraphAnalyser::searchMaxSubTrees	(	std::vector< VesselEdge * >	vStartEdges,
		unsigned int	nNum,
		std::vector< VesselEdge * > *	pvMaxEdges,
		float *	pfActValue,
		float *	pfActSum )

inline

calculate set of edges representing nNum subtrees maximizing selected feature stored in label property of the edges

Parameters

vStartEdges	base Edges to start search with
nNum	number of subtress to be selected
pvMaxEdges	set of Edges representing max subtrees
pfActValue	minimal feature value of currently selected subtrees
pfActSum	total feature sum of currently selected subtrees

Definition at line 217 of file mlGraphAnalyser.h.

References mlrange_cast().

◆ searchMaxSubTrees() [3/3]

void GraphAnalyser::searchMaxSubTrees	(	std::vector< VesselEdge * >::iterator	iBegin,
		std::vector< VesselEdge * >::iterator	iEnd,
		int	nNum,
		std::vector< VesselEdge * > *	pvMaxEdges,
		float *	pfActValue,
		float *	pfActSum )

calculate set of edges representing nNum subtrees maximizing selected feature stored in label property of the edges

Parameters

nNum	number of subtress to be selected
pvMaxEdges	set of Edges representing max subtrees
pfActValue	minimal feature value of currently selected subtrees
pfActSum	total feature sum of currently selected subtrees

◆ searchST_MajorBifurcation() [1/2]

void GraphAnalyser::searchST_MajorBifurcation	(	VesselEdge *	pStartEdge,
		float	fMinorLimit,
		std::vector< VesselEdge * > *	pvEdges,
		bool	bClear = true )

search for first major bifurcation following main edge starting with pStartEdge, use label property to differentiate major and minor branches.

The graph must be directed.

Parameters

pStartEdge	edge to start with
fMinorLimit	relative feature limit characterizing minor branches
pvEdges	resulting branches of first major bifurcation
bClear	clear old entries in result edge vector

◆ searchST_MajorBifurcation() [2/2]

void GraphAnalyser::searchST_MajorBifurcation	(	VesselEdge *	pStartEdge,
		float	fMinorLimitStart,
		float	fMinorLimitStep,
		int	nMinorLimitNumber,
		std::vector< VesselEdge * > *	pvEdges,
		bool	bClear = true )

search for first major bifurcation following main edge starting with pStartEdge, use label property to differentiate major and minor branches.

Starting from fMinorLimitStart the major/minor limit is stepwise reduced until a bifurcation is found. The graph must be directed.

Parameters

pStartEdge	edge to start with
fMinorLimitStart	start value of relative feature limit characterizing minor branches
fMinorLimitStep	value for stepwise reduction of the relative featur limit
nMinorLimitNumber	number of trails
pvEdges	resulting branches of first major bifurcation
bClear	clear old entries in result edge vector

◆ searchSubTreeSeparation()

void GraphAnalyser::searchSubTreeSeparation	(	VesselNode *	pRoot,
		const Vector3 &	separationDirection,
		std::vector< VesselEdge * > &	vClassifiedEdges )

◆ suggestRoot()

VesselNode * GraphAnalyser::suggestRoot	(	float	fSignificanceLevel = 0.7f,
		int	nPreferedDirection = 0 )

Search for root-node: look for leaf with thick edge seated remote from the barycenter of the Graph.

◆ sumupST_DrainVolume()

void GraphAnalyser::sumupST_DrainVolume ( bool bIgnoreOrientation = false )

inline

cumulate DrainVolumes given at the leafs

Parameters

bIgnoreOrientation state weather to use defined orientation of directed Graph or scan tree independent

Definition at line 300 of file mlGraphAnalyser.h.

References mlrange_cast().

◆ sumupST_Value()

void GraphAnalyser::sumupST_Value	(	VesselNode::NodeGetFkt	getFkt,
		VesselNode::NodeSetFkt	setFkt,
		bool	bIgnoreOrientation = false )

Parameters

getFkt	object bound function pointer to get Node parameter
setFkt	object bound function pointer to set Node parameter
bIgnoreOrientation	state weather to use defined orientation of directed Graph or scan tree independent

◆ tansferLabels()

void GraphAnalyser::tansferLabels	(	Skeleton::LABELTYPE	NType,
		float	fThresh = 0.1f )

transfer Vessel-Edge labels to Skeleton

◆ tansferLabelsInverse()

void GraphAnalyser::tansferLabelsInverse ( void )

transfer skeleton labels to vessel edge label (take label of most frequent skeleton)

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

MeVisLab/Standard/Sources/Shared/MLVesselGraph/mlGraphAnalyser.h

Public Member Functions

evaluation of Graph structure

Detailed Description

Constructor & Destructor Documentation

◆ GraphAnalyser()

Member Function Documentation

◆ calcDistanceMatrix()

◆ calcNodeBaseFromPosition()

◆ calcNodeClassifikator()

◆ calcPositionFromNodeBase()

◆ calcSpannedVolume()

◆ collectLeafs()

◆ getParent() [1/4]

◆ getParent() [2/4]

◆ getParent() [3/4]

◆ getParent() [4/4]

◆ measureAvMinDistance()

◆ measureBarycenter()

◆ measureLength()

◆ measureMaxPathValue()

◆ measureSkeletonArea()

◆ measureST_Barycenter()

◆ measureST_Volume()

◆ measureVolume()

◆ sampleConnectedCluster()

◆ scanSkeletonLabelMinMax()

◆ searchMaxSubTrees() [1/3]

◆ searchMaxSubTrees() [2/3]

◆ searchMaxSubTrees() [3/3]

◆ searchST_MajorBifurcation() [1/2]

◆ searchST_MajorBifurcation() [2/2]

◆ searchSubTreeSeparation()

◆ suggestRoot()

◆ sumupST_DrainVolume()

◆ sumupST_Value()

◆ tansferLabels()

◆ tansferLabelsInverse()