2017. Utilizzo non convenzionale dell'Imaging in ... · Utilizzo non convenzionale dell’Imaging...
Transcript of 2017. Utilizzo non convenzionale dell'Imaging in ... · Utilizzo non convenzionale dell’Imaging...
Utilizzononconvenzionaledell’Imaging
inRadioterapia
Chieti,24febbraio2017
Dr.LucaBoldrini
Conventional
Role ofimaging inRadiationOncology
DiagnosisStaging :locoregional,systemic
Characterization :Mpimaging,IB,hybrid approaches
Prognostic evaluationConventional RadiationOncology purposes
segmentation
planning
delivery,motion managementandadaptive approach
acutetoxicity
Follow up:response,relapseandlatetoxicity
Theragnostics
Theuseofdiagnostics
totailor therapeutic approaches
thus facilitating
personalized medicine
Bentzen S.- LancetOncology 2005
Theragnosticimaging paradigm
DiagnosisStaging
Target volumes &Planning
In room imaging (IGRT)Off-line / On-line
Theragnostic Imaging(biologically adapted
prescription)
Response evaluation
Tumor recurrence
Late toxicity
BEFORE AFTERDURING
Modified fromBentzen S.- LancetOncology 2005
Hightechnologyopportunities
Dosesculpting
Courtesy ofVerellen D.- 2013
2DPlanning 3 DPlanning IMRTPlanning
Hightechnologyopportunities
Towards small(anddifferently visible)targetvolumes
UNconventional
Main objective: higher dosedeliverytotargetsandtoxicityreduction withorgans at risk sparing through image
optimization
Metabolic andfunctional imaging:newtargets,dosepainting,newtoxicity paradigms
Adaptive therapy:intra- interfraction,movement
management,autosegmentation
MRIintreatmentroomrequiresafullyintegratedsolution:
1.MRI– LinacdesignedinUMC– Utrecht
8MVaccelerator,FFF
Modified1.5TPhilipsIngenia MRI
Linac mountedinringaroundMRI
Raaymakersetal.PhysMedBiol-2009 CourtesyofUulkevanderHeide
MRIintreatmentroomrequiresafullyintegratedsolution:
2.MRI– 60CoMRIdian®(ViewRay)0.35TMRIsplitmagnet
Realtimeimaging4framespersecond
360Coheads(15.000Cieach)onaringgantry
Boresize:69.3cm
Primarycollimatorsdirectlyunderthesources
MLC:30leaves
GRE:Gradient Echo - Protondensity,T1,T2- 2DGREis 25secperimage
TRUFI:TRUe FastImaging withsteadystatefreeprecession – T1,T2– 25sec3D planning/pilot,0.25sec
treatmentscan
TFL:TurboFlash– T1,mixT1/T2– 3min
EPI:Echo PlanarImaging – T2,mixT1/T2– 0.25secperframe
SE:SpinEcho
CourtesyofVIewRay:00016technicalmanualrevG
Localizationimaging
Imageregistration
Adaptivere-planning(ifneeded)
Treatmentexecution• MRin-linemonitoring
Doseaccumulation
1.5 T 0.35 T
• MRIforin-roomimagingopensanewerainradiationtreatmentworkflow
• Thisnewtechnologybringsmanyexpectationsandmultiplecriticalissues
• Needtomulti-centric cooperation,common lexicon forMRI-RT
• Possibilitytohaveanewtoolforprognosticevaluationduring the treatment execution
• Needtocreatea robustQA fordose accumulation algorithms• Evaluationofimpactofaccumulated doses onoutcome
prediction
Deliverable: MeasurementofOrganMotionCollaboration: Specificmetrics 3D/4D
«Features»toMRIdian
Calipso- VARIAN MRIdian - VIEWRAY
GAMMA.ADAPTIVE:Adaptive
GAMMA.RADIOMICS
Fenotipico
Biologico Genetico
Radiomica
Much morethan vessels andcells...
Hanahan D.andWeinbergRA.- Cell- 2011
Tumorheterogeneity
Vogelstein B.etal- Science- 2013
Tumorheterogeneity
Gerlinger M.etal- NEJM- 2012
Gerlinger M.etal- NEJM- 2012
Tumorheterogeneity
Tumor&treatments heterogeneity
Biomarkers
Noninvasive
Low-cost
Reliable
Easy
RADIOMICS
Tumorheterogeneity management
RadiomicsRadiomics istheprocessofextractionofquantitativefeaturesfromstandardradiologicalimagingforclinicaldecisionmakingtool.
TextureAnalysis,HistogramAnalysisandMorphometricAnalysisrepresentthethreemain
approachesfor featuresextraction.Dedicatedsoftwareneeded.
Lambin P.etal– Eur JCancer - 2012
Lambin P.etal– Eur JCancer - 2012
Radiomics:features extraction
Radiomics evaluation
• Not invasive• Repeatable• Analyzes entire tumor
volume
• Uses diagnostic exams
already available
• Cheap
Histological evaluation
• Invasive• Difficult torepeat• Tumour samplenot
always arerepresentative
ofthewhole volume
(tumorheterogeneity)
• Expensive
Radiomics analysis
Imagecollection Segmentation
Features extraction Analysisandmodeling
Radiomics:features extraction
Gemelli206ptsdatabasetimeframe2008-2014
Pre treatmentT2MRI,HRsequences noise filter
TRGAvailability
173ptsfinaldatabase
24ptsexclusion
9ptsexclusion
1.ROI extraction
2.Pre-processing:- LoG filter application
3.Dataanalysis (Moddicom):- Modelconstruction- Modelvalidation
KBORadiomics:rectal cancer experience
MRI ROIextraction Filterapplication
Kurtosis,Skewness,Entropy
KBORadiomics:rectal cancer experience
s = 0.49
s = 0.69
KBORadiomics:rectal cancer experience
Thefollowing variables were evaluated withmultivariatelogistic analysis for173rectalcancer patientscTcNGTVVolumeGTVSurfaceEquivalent SphereVolume/GTVSurfaceEntropy s=0.49Skewness s=0.69
Final model:
Coefficients:
EstimateStd.Error zvalue Pr(>|z|)
(Intercept) -5.14663.9229-1.3120.18954
cT -1.04420.3584-2.9130.00358**cN 0.53500.34121.5680.11689
Entropy Sigma 0.493.23541.64201.9700.04880*Skewness Sigma 0.69-3.14801.1601-2.7140.00666**---
Signif.codes:0‘***’0.001‘**’0.01‘*’0.05‘.’0.1‘’1
KBORadiomics:rectal cancer experience
FromRadiomics tonomograms
AUC=0.73
Internalvalidation
5000bootstrapresampling
TRIPOD1b
Specificity
Sensitivity
0.0
0.2
0.4
0.6
0.8
1.0
1.0 0.8 0.6 0.4 0.2 0.0
Externalvalidation
25casesMAASTRO
TRIPOD3
AUC=0.77
KBORadiomics:rectal cancer experience
Internalvalidation Externalvalidation
KBO MAASTRO173Patients 25Patients
47/173pCR (28%) 7/25(26%)
T2-w T2-w
Slicethickness3mm Slicethickness3mm
RMGESigna Exite @1.5T RMAchieva @1.5T
KBORadiomics:features extraction
Objectthat presents thesame weave ondifferent scalesscaleinvariance
Kochcurve
Ponteconietal,2016;Crossetal,1998;Waliszewski P,2016
Fractals
FractaldimensionTheparameter that characterizes afractal is thefractal dimension
Measure ofobject’s complexity
Low FD Pronged system
CompactsystemHighFD
Kochcurve
Penrose distribution
S ! = !#$ 1 < '( < 2
MandelbrotB.Thefractalgeometryofnature. 1982
KBORadiomics:features extraction
Personalization byRadiomics
KBORadiomics:features extraction
FromRadiomics tonomograms
?
ClinicalData ImagingData
Datasharing
Genetics
Lambin P. et al - Eur J Cancer - 2012Valentini V. et al - J Clin Oncol - 2011
Datafromdifferentsourcesandcontextscouldhighlyimproveourknowledge
Whatwewouldneed toshare
Whatwearewilling toshare
Whatweareabletoshare
Datasharing
WillemetalBMCPublicHealth,2014
Whichbarriers?Datasharing
- transparencyandcooperation- reproducibilityofresearch- cost-efficiency- preventing redundancies- accelerationofdiscoveryandinnovation- making moreefficientandeffective public health programs
BenefitsDatasharing
ImagingandInterventionalRadiologyforRadiationOncology
Editors:ReginaG.H.Beets-Tan,Wim Oyen,VincenzoValentini
PartI:ImaginginOncology:fromdiagnosistooutcomesPartII:Fromsimulationtodeliveryguidedbyimaging:
technicalaspectsPartIII:ImagingfortumorstagingandvolumedefinitionPartIV:ResponseevaluationandFollowupbyImaging
Lookingto(anear)future
• Newsegmentationandplanningtechniques(e.g.imagingbiomarkers)
• NewparadigmsofIGRTandadaptiverealtimeRT
• Newhybridtechniquesandmachines
• Newprognosticstratificationsystemsandclinicaldecisiontools
• Newradiomics perspectivesandclinicalintegration
• Coordinators
V.Valentini,A.Damiani
• PhDs
A.R.Alitto,S.Chiesa,G.Chiloiro,D.Cusumano
N.Dinapoli,A.Farchione,R.Gatta,V.Lanni
J.Lenkowicz,G.C.Mattiucci,C.Masciocchi,E.Meldolesi
• CollaborationwithMAASTROClinic
A.Dekker,J.VanSoest,P.Lambin
KBOAcknowledgments