How to use image information? Pawe Kukoowicz Verification of - PowerPoint PPT Presentation

IGRT2 How to use image information? Paweł Kukołowicz

Verification of radiotherapy  In space of dose  Comparison of prescribed and delivered dose (dose distribution)  Eg. In-vivo dosimetry  In space of location  Portal control  image based 2/42

Portal control  To minimize the set-up error  There are systematic and random errors in patient positioning  systematic errors deteriorate the dose delivery much more than random errors (3x)  The aim of portal control is to minimize the systematic error! 3/42

Veryfication of geometry  Geometry  Comparison of  reference image and  treatment field image Field edges Center of the beam 4/42

Reference image  Simultor image 5/42

Reference image  Simultor image 6/42

Reference image Digitally Reconstructed Image                 ( d 2 d 2 3 d 3 d I I e 1 1 4 4 0 7/42

DRR digitally reconstructed radiograph 8/42

Quality of DRR  Depends on  slice separation  it is recommended to use 3 mm slice separation  but  3 mm slice separation makes contouring very tedious  interpolation tools  these tools must be checked ! 9/42

Portal images EPID Courtesy of B.Heijmen

Edges zero of the second derivative of intensity 11/42

Matching of reference and portal images  P Y X 12/42

Structures to be matched brain AP lateral 13/42

Structures to be matched H&N lateral AP 14/42

Structures to be matched pelvis AP lateral 15/42

Correction strategies

AP direction 10 Systematic and random errors Single patient, one direction 5 Mean value = systematic error Fraction 0 0 5 10 15 20 25 -5 Standard deviation = random error Cortesy of B.Heijmen -10

Systematic and random errors For a few patients 10 mm patient 3 5 Fraction 0 0 5 10 15 20 25 patient 1 -5 patient 2 Cortesy of B.Heijmen -10

Systematic and random errors – 2D head left Cortesy of B.Heijmen

Systemic and random errors Group of „similar” patients A few patients Mean group error M x : <m i,x >  0 head M y : <m i,y >  0 Distribution of systematic errors  x : SD(m i,x )  y : SD(m i,y ) m 4 m 1 Random group error m 3 left 1    2 x i , x m 2 N i 1    2 y i , y N i Cortesy of B.Heijmen

Distribution of errors 600 prostate patients 750 60  AP 500 40 N  AP N 250 20 0 0 -10 -5 0 5 10 -10 -5 0 5 10 Systematic AP error (mm) Random AP error (mm) Cortesy of B.Heijmen (De Boer and Heijmen, IJROBP, 2001)

Strategies On-line protocols - measure and correct in the same fraction Off-line protocols - measure during first few fractions correct if needed • • SAL Shrinking Action Level (Amsterdam) • NAL No Action Level (Rotterdam) • eNAL extended NAL 22/42

On-line correction AP displacements (mm) Prostate cancer patient 15 most dose delivered with very small errors: correction of both systematic A few MU and random errors image 10 5 0 Fraction 0 5 10 15 20 25 30 35 Remainder MU -5 23/42

Data for on- off-line corrections  2D  EPID  3D  2 orthogonal iamges  CT type control  kV cone beam CT  MV cone beam CT  CT on rails 24/42

NAL (de Boer and Heijmen, IJROBP, 2001)  Fraction 1,2, and 3  set-up a patient according to protocol  portal control, (m ix ,m iy ,m iz ), i =1 ,2,3  before 4th fraction  calculate the systematic error (m x,mean ,m y,mean ,m z,mean)  from 4th fraction on  set-up a patient according to prtocol  shift couch with – (m x,mean ,m y,mean ,m z,mean )  irradiate de Boer and Heijmen, Med. Phys. 2002

No Action Level The random error remains the same! 15 10 No NAL 5 0 Fraction 0 5 10 15 20 25 30 35 Residual error -5 : initial error after set-up Residual error estimate  res   /  N : error after correction

NAL results With no correction With NAL 60 100 80  res  in 40 60 N N 40 20 20 0 0 -10 -5 0 5 10 -10 -5 0 5 10 Systematic AP error (mm) Systematic AP error (mm) 600 prostate patients (De Boer and Heijmen, IJROBP, 2001)

How precise may be radiotherapy? Residual (after NAL) bony anatomy displacements [mm]: LR CC AP  (1) 1.7 1.5 1.6 Prostate  res 1.1 (1) De Boer et al. 2002 1.1 1.1  2.6 2.9 2.7 (2) Kaatee et al. 2002 (2) Cervix  res 1.2 1.7 1.6 (3) De Boer et al. 2003  (3) 2.0 2.4 2.4 Lung (4) De Boer et al. 2004  res 1.3 0.6 1.2 (4)  1.6 1.4 1.6 head & neck  res 1.0 1.1 1.2 28/42

Why on-line verification is not recommended?  Because  It is time consuming.  Systemtic error influence on the margin three times more than random error.  However,  It might be resonble if  random error is large  very high accuracy is needed. 29/42

Margins  Set-up margin  to compensate set-up errors errors measured with  respect to external coordinate system (laser system)  Internal margin  to compensate movement of the target caused by physiology (eg. breathing) errors measured with  respect to internal anatomy coordinate system ( eg. pubis symphisis) Internal Margin Set Up Margin 30/42

How to add margins?  If set-up and internal errors may be treated as not correlated, than we add errors in quadrature          2 2 systematic  tot set up int ernal          2 2 random  int tot set up ernal 31/42

Margins  Two formulas To cover the CTV for 90% of the       patients with the 95% M 2 0 , 7 isodose (analytical solution). tot tot Herk Red, 47: 1121 - 1135, 2000 Margin size which ensures at least       M 2 , 5 0 , 7 95% dose is delivered to (on tot tot average) 99% of the CTV. Stroom, Red, 43: 905-919,1999 32/42

Implementation of geometry control  The most important task in radiotherapy department  „Lens of quality”  This can’t be an incidental action  This must be a program for the systematic monitoring and evaluation of the various aspects of radiotherapy quality 33/42

Data  Must be collected and regurarly analysed  feed back is a must  in our department ones a year all results are presented to doctors, radiation technologiests and physicists  big errors must be analysed as quickly as possible  conclusions must be drawn  group systematic errors (mean of means) play an important role in general evaluation of the quality of work and quality of equipment  group systematic error should not be different from zero 34/42

Breathing and related problems 35/42

CT for planning  Artefacts with breathing control without breathing control 36/42

Changes of GTV position In relations to bronchial tree 37/42

CT for planning  With breathing control RPM system 38/42

Pattern of breathing Patient A Patient B 39/42

Deep-inspiration breath hold technique  DIBH  for patients with left breast cancer  for some of tchem (they have to inhale and keep inhale for some time 10 – 15 sec) 40/42

DIBH - advantages 41/42

Thank you for your attention! p.kukolowicz@zfm.coi.pl