Causes  of  Image  Noise  in  PET  

Vibha  Chaswal,  Ph.D.  

Positron  Emission  Tomography  

PET  imaging:  Ideal  case  = Annihilation event

= event of detection of photon in the detector ring

ϒ1  

ϒ2  

e+  

Line of Response (LOR)

Positron emitting nucleus

Two simultaneous (within 6-12 ns time difference) events in the detector make a line of response.

Detector Ring

Data  arrangement  

Count on a Line of Response gets mapped to the corresponding position in a sinogram.

ϒ1  

ϒ2  

θ

t Position (t)

Ang

le (θ

)

(t, θ)

Sinogram

DeviaEons  from  Ideal  case  –  scaGer  events  

= Annihilation event

= event of detection of photon in the detector ring

ϒ1  

ϒ2  

e+  

True Line of Response (LOR)

Detector Ring ϒ2`  

Detected scatter event’s misplaced LOR

True counts

DeviaEons  from  Ideal  case-­‐  Random  counts  

= Annihilation event

= event of detection of photon in the detector ring

ϒ1  

ϒ2  

+  

False (LOR), contributed from two separate annihilation events

Detector Ring

True counts

PET:  Noise  

•  Background  noise:  ScaGer  events  map  misplaced  counts  to  the  sinogram  and  Random  events  map  false  or  spurious  counts  

•  ScaGer  -­‐  Supress  with  collimaEon  and  detectors  with  beGer  energy  resoluEon  

•  Randoms  -­‐  Supress  by  smaller  sampling  window  of  coincidence,  i.e.,  faster  scinEllator  detectors;  and  CollimaEon  

Factors  affecEng  counts  •  Trues  ↑  as  radio-­‐nuclide  concentraEon  ↑  (the  good  stuff)  ↓  as  paEent  size  ↑  (absorpEon  and  scaGer  effects)  •  Randoms  ↑↑  as  count  rate  ↑  (varies  as  square  of  count  rate)    Effects  dominate  image  noise  at  high  injected  acEviEes    Reduce  by  faster  electronics,  faster  crystals  •  Sca/ers  ↓  with  collimaEon  (about  15%  for  2D  PET  and  50%  for  3D  PET)  Reduce  with  collimaEon  

(energy  selecEon  not  efficient  in  PET)  

PET  improvements  leading  to  improved  SNR  

•  Faster  electronics  •  Faster  scinEllators:  shorter  coincidence  Eming  window,  reduced  dead  Eme,  improved  energy  resoluEon  window.    

•  PET/CT  technology,  ACFs  from  CT  scan  (reduced  paEent  movement  related  noise  factors  from  long  transmission  scans),  total  scanning  Eme  reduced  to  10-­‐15  minutes.  

•  Fourier  re-­‐binning  and  staEsEcally-­‐based  algorithms  

Resources  

•  DW  Townsend,  ‘Physical  principles  and  technology  of  clinical  PET  imaging’,  BJR,  March  2004,  Vol  33  (2)  

•  Bushberg  book  •  J  A  Anderson,  ‘IntroducEon  to  PET/CT’,  CRCPD  winter  2004  meeEng  presentaEon  

Thank  You!