mri: a historical perspective
TRANSCRIPT
¡ JohnBap&steFourier§ OfficerinNapoleon'sarmy
§ Mathema&csforthermodynamics
§ InventedtheFourierTransformwhichliesatthecoreofMRItoday
[1]TalGevaJCMR2006
¡ NikolaTesla§ Serbian§ InventorofAC§ Teslacoil§ Sta&onaryterrestrialwaves
§ Unitofmagne&cfieldstrengthnamedaQerhim
[1]TalGevaJCMR2006
¡ Independentlydiscoveredthenuclearmagne&cresonanceeffect
¡ AwardedNobelPrizein1952fortheirdiscoveries[1]TalGevaJCMR2006
§ 1971:RaymondDamadian–relaxa&on&mesfortumorsinratmodelsofcancer
§ 1973:AbeZenuemonandothers,filepatentforthefirsttargetedNMRforevalua&onofinforma&oninsidefromoutside
§ 1974:Magne&cResonanceImagingisbornbytheuseofgradients,thankstoPaulLauterburandPeterMansfield
§ 1975:RichardErnst(AnilKumaret.al)describedtheuseofFTforMRI
[1]TalGevaJCMR2006
§ 1980:Average&metomakeanMRimageis5minutes
§ 1985:Average&metomakeanMRimageis5seconds
§ 1983:T2weightedimagingbeferforhighligh&ngpathology
§ 1984-85:CardiacMRI,bloodflow,CE-MRI,Steadystatefreeprecession
Researchlabstoclinicalprac&ce
[1]TalGevaJCMR2006[3]hfp://www.fonar.com/&melineofmri.htm
¡ 1988:EchoPlanarImagerforpediatrics,LarryCrooksatUCSF
¡ 1987:Mul&pleRFcoilsusedforMRI
¡ 1987–90:ParallelimagingisbornthankstoCarlson
• Gooddataqualitybuttakesalong&me!• Hence,maynotbesuitableforcertainimagingprotocols.• Limitsspa&alandtemporalresolu&ons• Higherspa&alresolu&onaidsinmorphologicalanalysisoftumors–breastDCE-MRI• Temporalresolu&onisimportantforaccuratepharmacokine&canalysis.• Severalapproacheslikekeyhole,parallelimagingandotherfastsequenceshavebeenused.2D FFT
2D IFFT
50
100
150
200
X
1 0 DataprovidedbyBaek
¡ 2008:CompressedsensingisappliedtoMRI
¡ 2012:KeynotespeechesatISMRMoncompressedsensing
¡ CSappliedtoMRSI,DCE-MRI,cardiacimaging,brainimaging,almosteveryknownMRmethod
X2D IFFT
10
20
30
40
50
60
70
10
20
30
40
50
60
70
UniformSampling
X2D IFFT
20
40
60
80
100
120
20
40
60
80
100
120
IncoherentSampling
Completedatareconstruc0on
Wavelet
Transform
DataprovidedbyBaek
[1]DavidL.Donoho,IEEETransac0onsonInforma0ontheory,Vol.52,no.4,April2006[2]Candes,E.J.etal.,IEEETransac0onsonInforma0ontheory,Vol.52,no.2,Feb.2006
• Mostobjectsinnatureareapproximatelysparseinatransformeddomain.• U&lizeaboveconcepttoobtainveryfewmeasurementsandyetreconstructwithhighfidelity [5,6]
Only33%ofcompletedataX
¡ CoilsinIndia
¡ Combina&onofparallelimagingandcompressedsensing
¡ ParalleltransmitMRI
[7] Andrew Thomas et. al.,2007 International Journal of Biomedical Imaging
MRVOLUMECOILSfor
0.5Tesla
BIRDCAGE1:2ra&o,8rung
VOLUMEcoil1:2ra&o
BIRDCAGE1:1ra&o,4rung
BIRDCAGE-1:2ra&o,4rung
§ Tohaveminimumaliasing due to sampling below theNyquist rate, thefollowingproper&esofideasamplinghavetobemet.§ Thenear zero regionaround themain lobeof thePSF shouldbeas largeas
possibleandoutsidethatregion,PSFshouldresemblewhitenoise.
§ The samples should be placed randomly but with a restricted maximum
distancebetweensamples.
§ These two condi&ons are met by Poisson disc sampling but it hasimprac&calgradientrequirements.
§ Constrained random pafern is a normal lasce pafern with samplesshiQed along one dimension randomly by -1, 0 and +1 i.e. constrainedrandomiza&onaddedalongonedirec&on.
§ Ithasmoderategradientrequirements.
§ Performance of compressed sensing (CS) algorithms depends onthesparsity levelof thesignal, thetypeofsamplingpafernusedandthereconstruc&onmethodapplied.
§ The higher the incoherence of the sampling pafern used forundersampling, less aliasing is no&ced resul&ng in befer CSreconstruc&on.
§ ThetheoryofCSrequiresacquisi&onofrandomizedsetofmeasurements(randomsampling),leadingtoincoherantaliasingar&facts.
§ ButrandomsamplingrequiresbiggerchangesinamplitudesandpolarityofMRsystemgradients,whichisnotfeasibleinMRsystem.
Three candidate sampling patterns and their corresponding PSFs: top to bottom: random, Poisson disc and constrained random
Usman et. al., Sampta 2009
21
¡ Joint k-space trajectory allows for the specification of a region of interest(ROI), which improves excitation accuracy at high speedup factors
¡ It allows for magnetic field inhomogenetiy compensation during excitation k-space
¡ Optimized accelerated selective excitation is useful for reducing specific absorption rate(SAR) and shortening multidimensional RF pluses
¡ Normal transmission § In a normal transmission, the data/signals are sent one at a time
over the transmission channel
¡ Parallel transmission § Parallel transmission means simultaneous transmission
of N signals. These signals are sent simultaneously over N different channels
Radartutorial.eu
23
Reconstruc&onofimage
Pulse sequence to generate k-space
1. Target organ 2. K-space trajectory focus on
heart and use spiral
2D-IFT
Pa&ent
Spiral k-space trajectory
24
SAR § Specific absorption rate (SAR) is a measure of the rate at
which energy is absorbed by the body when exposed to a radio frequency (RF) electromagnetic field
§ It is defined as the power absorbed per mass of tissue and has units of watts per kilogram(W/kg)
¡ SAR can be calculated from the electric field within the tissue as:
where ¡ σ is the sample electrical conductivity ¡ E is the RMS electric field ¡ ρ is the sample density
( ) ( )( )
drrrEr
SARsample∫=
2
2ρσ
25
¡ Errors and SAR will be increasing, we need to decrease them ¡ So we have a formula as
C= є + λ (SAR) …2 Here C= cost, є= excitation error, λ = Lagrange multiplier ¡ As λ increases SAR reduces to some extent then remains constant and
error also remain constant L-curve
L-curve Subset virtual observation points for RF
¡ Facultyandmanagement§ Prof.A.N.N.Murthy,Principal,DSCE§ Dr.PremchandraSagar,CEO,DSCE
¡ Students§ PadmaC.R.(coils)§ SnehaPotdar(Paralleltransmitbackground)
¡ MIRCfacultyandstudents,collaboratorsandindustrialpartners
28
¡ The image of the target organ in which we are interested is taken using 2D-FT
¡ The scanned image obtained by parallel transmit from MRI is called as k-space, obtained k-space undergoes sampling at ROI
¡ Then 2D-IFT is done to achieve a desired excitation pattern
Echo-Planar Imaging (EPI) k-space raster. Brazilian journal of physics