AAPM 2006

3D Surface Imaging 3D Surface Imaging for PBI Patient Setupfor PBI Patient SetupG.T.Y. ChenG.T.Y. Chen11, Ph.D., M. Riboldi, Ph.D., M. Riboldi22, Ph.D., Ph.D.

ChristophChristoph BertBert33, Ph.D., D.P. Gierga, Ph.D., D.P. Gierga11, Ph.D., Ph.D.

1Massachusetts General HospitalHarvard Medical School

2TBMLab - Department of BioengineeringPolitecnico di Milano University

3GSI Darmstadt, Germany

AAPM 2006

WG Conner 1975 WG Conner 1975 –– Motion detection Motion detection /cancellation imaging/cancellation imaging

JA Purdy 1978

AAPM 2006

Video-based patient set-up

3-Dimensional:

• laser interferometry

- surface height maps

• stereo-photogrammetry

- point-based registration

- surface registration

2-Dimensional:

• video images with single / orthogonal cameras

accuracy

simplicityMarco Riboldi

AAPM 2006

3-D point-based stereo-photogrammetry

BunkerControl roomSync-power and composite video

Digital data

3-D markers coordinates

gantry

radiationbeam

control points

TVC 1TVC 2IR flash IR flash

Isocenter localization

MotionAnalyzer

CPU

AAPM 2006

Applications of 3-D point-based stereo-photogrammetry

G Baroni, CAS 2000Breast irradiation

AAPM 2006

3-D surface scanning

G Baroni, MBEC 2003Breast irradiation

Opto-electronic surface sensingHybrid registration (surface fiducials / laser spots

JW Sohn, AAPM 2004Breast irradiation

Handheld laser scanning

AAPM 2006

Recent References / Others

2-D patient set-up:1. Milliken BD, Rubin SJ, Hamilton RJ, Johnson LS, Chen GT. Performance of a video-image-subtraction-based patient

positioning system. Int J Radiat Oncol Biol Phys. 1997 Jul 1;38(4):855-66.2. Johnson LS, Milliken BD, Hadley SW, Pelizzari CA, Haraf DJ, Chen GT. Initial clinical experience with a video-based

patient positioning system. Int J Radiat Oncol Biol Phys. 1999 Aug 1;45(1):205-13.

3-D point-based stereo-photogrammetry:1. Rogus RD, Stern RL, Kubo HD. Accuracy of a photogrammetry-based patient positioning and monitoring system for

radiation therapy. Med Phys. 1999 May;26(5):721-8.2. Baroni G, Ferrigno G, Orecchia R, Pedotti A. Real-time opto-electronic verification of patient position in breast cancer

radiotherapy. Comput Aided Surg. 2000;5(4):296-306.3. Soete G, Van de Steene J, Verellen D, Vinh-Hung V, Van den Berge D, Michielsen D, Keuppens F, De Roover P,

Storme G. Initial clinical experience with infrared-reflecting skin markers in the positioning of patients treated by conformal radiotherapy for prostate cancer. Int J Radiat Oncol Biol Phys. 2002 Mar 1;52(3):694-8.

4. Weiss E, Vorwerk H, Richter S, Hess CF. Interfractional and intrafractional accuracy during radiotherapy of gynecologic carcinomas: a comprehensive evaluation using the ExacTrac system. Int J Radiat Oncol Biol Phys. 2003 May 1;56(1):69-79.

Surface registration:1. Moore C, Lilley F, Sauret V, Lalor M, Burton D. Opto-electronic sensing of body surface topology changes during

radiotherapy for rectal cancer. Int J Radiat Oncol Biol Phys. 2003 May 1;56(1):248-58.2. MacKay RI, Graham PA, Logue JP, Moore CJ. Patient positioning using detailed three-dimensional surface data for

patients undergoing conformal radiation therapy for carcinoma of the prostate: a feasibility study. Int J Radiat OncolBiol Phys. 2001 Jan 1;49(1):225-30.

3. Baroni G, Troia A, Riboldi M, Orecchia R, Ferrigno G, Pedotti A. Evaluation of methods for opto-electronic body surface sensing applied to patient position control in breast radiation therapy. Med Biol Eng Comput. 2003 Nov;41(6):679-88.

4. Sohn J, Kim S, Chvetsov A, Suh T, Jin H, Farr J. Three-Dimensional Surface Image Registration For Image-Guided IMRT To Breast. AAPM 2004 Proceedings.

AAPM 2006

Clinical Implementation ofClinical Implementation ofIGRT for PBI : MicrocosmIGRT for PBI : Microcosm

Multiple Approaches to IGRT Multiple Approaches to IGRT Organ deformationOrgan deformationRespiration Respiration Quantify accuracy of methodsQuantify accuracy of methodsApply statistical rigor to IGRTApply statistical rigor to IGRTEtcEtc……

AAPM 2006

Challenge: Setup of BreastChallenge: Setup of Breast•Irradiate involved quadrant of breast (vs. whole)•4.0 Gy X 8 fractions BID / 4 days (vs. 6 wks)•Escalate dose; minimize NT irradiation

Mini tangents and electron boost

AAPM 2006

Imaging OptionsImaging Options

Image target (Image target (seromaseroma) ) directlydirectly (in room CT)(in room CT)

Surrogates (more commonly used)Surrogates (more commonly used)Skin tattoos Skin tattoos –– aligned to lasersaligned to lasersChest wall Chest wall –– imaged in radiographsimaged in radiographsBreast surface Breast surface –– 3D video imaging 3D video imaging ((newnew))Clips near seroma Clips near seroma –– imaged by radiographsimaged by radiographs

AAPM 2006

Sources of UncertaintySources of Uncertainty

Elasticity of skin, arm position / lasersElasticity of skin, arm position / lasersChest wall is weakly coupled to tumor / breast Chest wall is weakly coupled to tumor / breast tissue tissue Skin /seroma correlation Skin /seroma correlation -- deformation affects deformation affects accuracyaccuracyClip migration / seroma shrinkage affects Clip migration / seroma shrinkage affects radiographic accuracyradiographic accuracy

Conebeam CT Conebeam CT ––beforebefore not during Rxnot during Rx

AAPM 2006

Outline:Outline:

1)1) System DescriptionSystem Description2)2) System PerformanceSystem Performance3)3) Patient StudiesPatient Studies4)4) Target Registration AnalysisTarget Registration Analysis5)5) SummarySummary

AAPM 2006

1. Surface Imaging Hardware1. Surface Imaging Hardware

POD 1

POD 2

GantryAlignRT

Linac1Linac2(Protons)

AAPM 2006

Each pod containsEach pod contains……..

SpeckleCamera

SpeckleCamera

ProjectionTexture

CamFlash

AAPM 2006

Speckle imageSpeckle imageTwo camera podsTwo camera podsFlash modeFlash mode

Speckle patternSpeckle pattern6 images captured6 images captured

Surface Surface coordscoords in 3Din 3DSurface matching to Surface matching to maximize maximize congruence between congruence between reference and Rx reference and Rx surfacessurfaces

Torso phantom

AAPM 2006

Reconstructed surface imageReconstructed surface image

AAPM 2006

Wire Frame Wire Frame ObjObj imagesimages

Zoomed Surface Patch

AAPM 2006

Interactive DemoInteractive Demo

Use file Marco 2006 Use file Marco 2006 phantom.objphantom.objD:D:\\Research Projects 06Research Projects 06\\spine 3d spine 3d videovideo\\viewerviewer\\SurfaceView.exeSurfaceView.exe

AAPM 2006

3D Surface Alignment Process3D Surface Alignment Process(analogous to conventional IGRT)(analogous to conventional IGRT)

Define reference surface (CT or 1Define reference surface (CT or 1stst Rx)Rx)Acquire daily surface image (after laser Acquire daily surface image (after laser setup)setup)Match daily 3D image with ref image Match daily 3D image with ref image through surface matchingthrough surface matchingAdjust patient positionAdjust patient position(Verify post move)(Verify post move)

AAPM 2006

Alignment ScreenAlignment Screen

ROI

4 DOF 6 DOFStandard Couch

AAPM 2006

2.Characterize System Performance2.Characterize System Performance

What is the smallest misalignment detectable by What is the smallest misalignment detectable by this 3D video system?this 3D video system?

Performed phantom and Performed phantom and calculationalcalculational experiments experiments to measure system performance.to measure system performance.

C. Bert et al Medical Physics 2005, M Riboldi 2006

AAPM 2006

Ground TruthGround Truth

Breast phantom

High precision mechanical stage

(digital micrometer, 1/100 mm)

AAPM 2006

Phantom Surface ModelPhantom Surface Model

ROI

AAPM 2006

Determine System AccuracyDetermine System Accuracy

Acquire reference surfaceAcquire reference surfaceMove mechanical stage known amount Move mechanical stage known amount Acquire Acquire ““dailydaily”” surface image / query system surface image / query system ––how much did it move?how much did it move?Compare Compare ground truthground truth (known move) with (known move) with AlignRT calculated shift.AlignRT calculated shift.

AAPM 2006

Phantom Study ResultsPhantom Study Results•18 readings / data points

•range of shifts: [-2, +2] mm in each direction (VRT, LNG, LAT)

•compared AlignRT-suggested shifts vs. digital micrometers (ground truth)

VRT LNG LAT 3-D

MEAN 0.011 0.011 0.011 0.1

1SD 0.144 0.144 0.144 0.227

Translation differences [mm]

Minimum detectable translational shifts is subMinimum detectable translational shifts is sub--mm.mm.

AAPM 2006

High precision mechanical stage (digital micrometer, 1/100 mm)

Can system detect small angular misalignments?

VRT

LNG

LAT

Phantom virtual experimentPhantom virtual experiment

AAPM 2006

Results – 6 DOF transformation parameters

Virtual phantom experiment

-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

VRT LNG LAT ROT PR1 PR2

diffe

renc

es [m

m] [

deg]

mean±SD

AAPM 2006

3.Patient Studies3.Patient Studies

Analyze multiple methods of PBI setup (Laser, Analyze multiple methods of PBI setup (Laser, Chest Wall, Iris, AlignRT) Chest Wall, Iris, AlignRT) Comparative / Comparative / quantitative analysis of method accuracy.quantitative analysis of method accuracy.Metric: residual displacements on target localization Metric: residual displacements on target localization after alignment (after alignment (Target Registration ErrorTarget Registration Error))Data analysis performed on a Data analysis performed on a statistical basisstatistical basis (non (non parametric Friedman ANOVA and Wilcoxon tests)parametric Friedman ANOVA and Wilcoxon tests)

AAPM 2006

Patient Imaging ProtocolPatient Imaging Protocol

Patient alignedby lasers

OrthogonalIRIS images:

Clip based moveTreat

ImageSMT

ImageSME

ImageSML

SMRfrom CT

SMT1 as reference for VisionRT TRE evaluation

DRRsfromCT

AAPM 2006

Breast Alignment AnalysisBreast Alignment Analysis

AAPM 2006

4.Target Registration Error4.Target Registration Error

The TRE is the vector difference between the The TRE is the vector difference between the target as aligned by method target as aligned by method a,b,ca,b,c…… and ground and ground truth.truth.Ground truth: defined by clips (DRRs, Planning Ground truth: defined by clips (DRRs, Planning CT)CT)

AAPM 2006

Laser TRELaser TRE

Align breast by laser; take radiographs of clips (ground Align breast by laser; take radiographs of clips (ground truth). Calculate shifts needed to bring clip into congruence truth). Calculate shifts needed to bring clip into congruence with ref image clip position (TRE of Lasers)with ref image clip position (TRE of Lasers)

GT

Lasers

AAPM 2006

Chest Wall (CW) TREChest Wall (CW) TREBegin with perfect IRIS clip alignment; align CW to ref DRR; Begin with perfect IRIS clip alignment; align CW to ref DRR;

vector difference is CW TREvector difference is CW TRE

SupInf

Ant

Post

TClips

Chest Wall

TSupInf

Ant

Post

TClips

Chest Wall

T

Chest Wall alignment

AAPM 2006

AlignRT TREAlignRT TRE

Begin with perfect IRIS clip alignment; match surfaces; apply Begin with perfect IRIS clip alignment; match surfaces; apply transformation to isocenter. Difference vector is AlignRT transformation to isocenter. Difference vector is AlignRT TRE.TRE.

Surface mismatch

Matched surfaces

TRE

AAPM 2006

IRIS (radiographic) SystemIRIS (radiographic) System

IRISX-ray Tube retract

Detector Arms retract

Steve JiangGreg SharpRoss Berbeco

Berbeco et al Phys Med Biol 2004:49:243-257

AAPM 2006

Radiographic (IRIS) TRERadiographic (IRIS) TREAcquire IRIS radiographs; calculate shifts; make shifts; reAcquire IRIS radiographs; calculate shifts; make shifts; re--image, image, DIPS. IRIS TRE is inexactness in repositioning patient EXACTLY tDIPS. IRIS TRE is inexactness in repositioning patient EXACTLY to o calculated shift; calculated shift; ieie residual error in setup.residual error in setup.

IRIS alignment

AAPM 2006

TRE Analysis ResultsTRE Analysis ResultsTRE analysis

Median 25%-75% Min-Max Laser

IrisVisionRT

CW

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.43D

TR

E [c

m]

MediansLaser: .79 cmIris: .22VRT .32CW: .57

AAPM 2006

Statistical AnalysisStatistical Analysis

• Is there a meaningful difference between Laser, Iris, Chest Wall and VisionRT TREs?

YES -> Friedman ANOVA test (p<0.00004)

• Where is the difference?

Wilcoxon rank test:

LASER vs IRIS p<0.0037

LASER vs AlignRT p<0.0044

IRIS vs AlignRT p<0.21

CW vs LASER p<0.11

AAPM 2006

Statistical Analysis ResultsStatistical Analysis Results

• Laser, Iris, CW and VisionRT can be divided in 2 groups, in terms of TRE results:

CW

Laser

Iris

AlignRT

Median TRE

≈ 6-8 mm

Median TRE

≈ 2-3 mm

AAPM 2006

QuestionQuestion

If the intrinsic accuracy of surface based If the intrinsic accuracy of surface based alignment is <0.5mm (as shown in precision alignment is <0.5mm (as shown in precision phantom experiments), then why are patient phantom experiments), then why are patient TRETRE’’ss on the order of 3mm?on the order of 3mm?(TRE of IRIS radiographic clip alignment is (TRE of IRIS radiographic clip alignment is about 2mm)about 2mm)Deformation? Respiration? Other effects?Deformation? Respiration? Other effects?

AAPM 2006

SMT2 SMT3 SMT4

SMT5 SMT6 SMT7

SMT8

Is there breast deformation?Patient 4: Generally, reference surface andtreatment surfaces are within 2mm after6 DOF fit. (green areas)

AAPM 2006

Patient 5

AAPM 2006

Can we use CT Breast SurfacesCan we use CT Breast Surfacesas reference image?as reference image?

Rx: Rt BreastRx: Lt Breast

Estimated magnitude Estimated magnitude ––using GE Workstation to measure using GE Workstation to measure ––Pt 4 ~ 5 mm; Pt 5 ~ 3Pt 4 ~ 5 mm; Pt 5 ~ 3--4 mm4 mm

AAPM 2006

Texture Images reveal setupTexture Images reveal setup

AAPM 2006

Ongoing Studies:Ongoing Studies:

TRE as a function of breast size and height TRE as a function of breast size and height above chest wall. above chest wall. Protocol extended to 300 PBI patients.Protocol extended to 300 PBI patients.IntraIntra--fractional dosimetric variations due to fractional dosimetric variations due to breathingbreathing

AAPM 2006

Why surface imaging if we have Why surface imaging if we have ground truth by clips?ground truth by clips?

Faster Faster ––guide to optimal position, verify as guide to optimal position, verify as neededneededReduce radiation in comparison to radiographsReduce radiation in comparison to radiographsSurveillanceSurveillance during Rxduring RxNot every machine has conebeam CT or OBINot every machine has conebeam CT or OBIApplications in charged particle beam Applications in charged particle beam radiotherapy.radiotherapy.

AAPM 2006

5.Summary5.Summary

Determined that TRE of 3D surface imaging Determined that TRE of 3D surface imaging system superior to conventional methods.system superior to conventional methods.

Applied statistics to provide significanceApplied statistics to provide significance

Respiration remains issue if accuracy < 2 Respiration remains issue if accuracy < 2 mm is desired.mm is desired.

Deformation is minimal in patients studiedDeformation is minimal in patients studied

3D technology promising for PBI setup3D technology promising for PBI setup

AAPM 2006

AcknowledgementsAcknowledgements

Steve Jiang ,Greg Sharp, Julie Steve Jiang ,Greg Sharp, Julie TurcotteTurcotteSimon Powell, Alphonse Taghian, Angela Katz, Simon Powell, Alphonse Taghian, Angela Katz, Ellen Kornmehl Ellen Kornmehl MGH Radiation TherapistsMGH Radiation TherapistsTechnical advice: Norman Smith / Ivan Meir Technical advice: Norman Smith / Ivan Meir VisionRTVisionRTStudy conducted under IRB protocol.Study conducted under IRB protocol.

AAPM 2006

ReferencesReferences

Bert, C et al: Bert, C et al: A phantom evaluation of a stereo-vision surface imaging system for radiotherapy patient setup Med Phys 32:9 2005Bert, C. et al: Initial Clinical Experience with Bert, C. et al: Initial Clinical Experience with Surface Imaging for PBI: Surface Imaging for PBI: IntInt J J RadRad Onc Bio Onc Bio Phys March 15, 2006Phys March 15, 2006Riboldi, M: submitted for publication 2006Riboldi, M: submitted for publication 2006GiergaGierga, DP: Oral presentation at ASTRO 2006, DP: Oral presentation at ASTRO 2006