cluster_tac.c File Reference

Clustering and segmentation for PET modeling. More...

#include "libtpcmodext.h"

Go to the source code of this file.

Functions
int	clusterTACs (IMG *dimg, IMG *cimg, int nr, DFT *tac, int verbose)

Detailed Description

Clustering and segmentation for PET modeling.

Author

Vesa Oikonen

Definition in file cluster_tac.c.

Function Documentation

clusterTACs()

int clusterTACs	(	IMG *	dimg,
		IMG *	cimg,
		int	nr,
		DFT *	tac,
		int	verbose )

Allocates memory and calculates the values for average TACs for clusters.

Returns

Returns 0 if ok, and >0 in case of an error.

Parameters

dimg	Dynamic image
cimg	Cluster image
nr	Highest cluster ID
tac	Pointer to initiated but empty DFT data
verbose	Verbose level; if zero, then only warnings are printed into stderr

Definition at line 15 of file cluster_tac.c.

  {
  if(verbose>0) printf("clusterTACs(dimg, cimg, %d, tac, %d)\n", nr, verbose);
 
  /* Check the arguments */
  if(dimg==NULL || cimg==NULL || nr<1 || tac==NULL) return(1);
  if(dimg->dimt<1 || cimg->dimt<1) return(1);
  if(cimg->dimx!=dimg->dimx || cimg->dimy!=dimg->dimy || cimg->dimz!=dimg->dimz)
    return(2);
 
  /* Allocate memory for the TACs */
  dftEmpty(tac);
  if(dftSetmem(tac, dimg->dimt, nr+1)) return(3);
  
  /* Set TAC info */
  tac->voiNr=0; tac->frameNr=dimg->dimt; tac->_type=1;
  for(int fi=0; fi<tac->frameNr; fi++) {
    tac->x1[fi]=dimg->start[fi];
    tac->x2[fi]=dimg->end[fi];
    tac->x[fi] =dimg->mid[fi];
  }
  tac->timetype=DFT_TIME_STARTEND;
  tac->timeunit=TUNIT_SEC;
  strcpy(tac->unit, imgUnit(dimg->unit));
 
  /* Calculate one cluster at a time */
  float y[dimg->dimt];
  int clusterID;
  for(clusterID=1; clusterID<=nr; clusterID++) {
    char buf[128]; snprintf(buf, 128, "%06d", clusterID);
    char *p=buf+strlen(buf)-6;
    snprintf(tac->voi[clusterID-1].voiname, MAX_REGIONSUBNAME_LEN+1, "%s", p);
    int n=imgsegmClusterMean(dimg, cimg, clusterID, y, verbose);
    if(verbose>1) printf("  clusterID%d -> %d pixels\n", clusterID, n);
    if(n<0) return(5); else if(n==0) return(6);
    for(int fi=0; fi<tac->frameNr; fi++) tac->voi[clusterID-1].y[fi]=(double)y[fi];
    tac->voi[clusterID-1].size=(double)n*dimg->sizex*dimg->sizey*dimg->sizez;
    tac->voiNr++;
  }
  /* and once more for cluster 0, i.e. the thresholded pixels;  */
  /* note that it is possible that there is no cluster 0 at all */
  clusterID=0;
  sprintf(tac->voi[tac->voiNr].voiname, "%06d", clusterID);
  strcpy(tac->voi[tac->voiNr].name, tac->voi[tac->voiNr].voiname);
  int n=imgsegmClusterMean(dimg, cimg, clusterID, y, verbose);
  if(n<0) return(7);
  if(n>0) {
    for(int fi=0; fi<tac->frameNr; fi++) tac->voi[tac->voiNr].y[fi]=(double)y[fi];
    tac->voi[tac->voiNr].size=(double)n*dimg->sizex*dimg->sizey*dimg->sizez;
    tac->voiNr++;
  }
 
  return(0);
}