Basics of Clinical X-Ray Computed Tomography

Transcription

1 Siemens 3=18-slice dual source cone-beam spiral CT (005) Basics of Clinical X-Ra Computed Tomograph EMI parallel beam scanner (197) Marc Kachelrieß Institute of Medical Phsics Universit of Erlangen-Nürnberg German 1536 views per rotation in 0.33 s 3 (67+35) -bte channels per view 600 MB/s data transfer rate 180 views per rotation in 300 s GE LightSpeed 5 GB data sie tpical 160 positions per view Toshiba Aquilion What does CT Measure? Polchromatic Radon transform p( L ) = ln de w( E ) e Siemens Somatom Definition µ (r, E ) dl µ ( r, E ) with normalied detected spectrum: 1 = de w( E ) Philips Brilliance Widel used monochromatic approimation: p ( L ) dl µ ( r, Eeff ) with the effective energ being around 70 kev Dual Source CT-Performance (Best-of Values) CT Basics From Single-Slice to Cone-Beam Spiral CT Trot Technolog s 4 Basic parameters Detector concepts, tube technolog Scan trajectories, scan modes Spiral CT Algorithms D filtered backprojection Spiral -interpolation ASSR and EPBP (cone-beam recon.) Phase-correlated CT (e.g. cardiac CT) Image qualit and dose Spatial resolution (PSF, SSP, MTF) Relation of noise, dose and resolution Dose values (CTDI, patient dose) Dose reduction techniques collimation tp. 30 cm scan1 slices/s 13 mm / s mm 0 mm, 30 s s 1 mm 10 mm, 30 s s 1 mm 8 mm, 30 s s 4 1 mm 4 1 mm, 30 s s mm mm, 1 s s mm s mm mm, 0. s mm, 3 s 500 assuming a breath-hold limit of 30 s factor 4 converts from head FOM to full bod FOM assuming p = 1, otherwise Seff is increased assuming p = 1.5 since IQ is independent of pitch for MSCT Seite 1 1

2 Fan-Beam Geometr (transaial / in-plane / --plane) -ra tube p.a. field of measurement (FOM) and object Acquisition Reconstruction lateral 180 detector (tp channels) (illustration without quarter detector offset) Object, Image a.p. Sinogram, Rawdata Data Completeness In the order of 1000 projections with 1000 channels are acquired per detector slice and rotation. (illustration without quarter detector offset) Each object point must be viewed b an angular interval of 180 or more. Otherwise image reconstruction is not possible. Basic Parameters (best-of values tpical for modern scanners) In-plane resolution: mm Nominal slice thickness: S = mm Effective slice thickness: S eff = mm Tube (ma. values): 100 kw, 140 kv, 800 ma Effective tube current: mas eff = 10 mas 1000 mas Rotation time: T rot = s Simultaneousl acquired slices: M = 4 64 Table increment per rotation: d = 50 mm Pitch value: p = Scan speed: up to 16 cm/s Temporal resolution: ms Demands on the Mechanical Design Continuous data acquisition in spiral scanning mode Able to withstand ver fast rotation Centrifugal force at 550 mm with 0.5 s: F = 9 g with 0.4 s: F = 14 g with 0.3 s: F = 5 g Mechanical accurac better than 0.1 mm Compact and robust design Short installation times Long service intervals Low cost Seite

3 Tube Technolog Demands on X-Ra Sources conventional tube high performance tube (rotating anode, helical wire emitter) (rotating cathode, anode + envelope, flat emitter) cooling oil cooling oil cathode cathode anode Photo courtes of GE RF Available as multi-row arras Ver fast sampling (tp. 300 µs) Favourable temporal characteristics (deca time < 10 µs) High absorption efficienc High geometrical efficienc High count rate (up to 108 cps*) Adequate dnamic range (at least 0 bit) cooling oil RF B RF anode C Photo courtes of Siemens Demands on CT Detector Technolog C cathode B C C anode tan φ = cathode anode High instantaneous power levels (tp kw) Increasing with rotation speed High continuous power levels (tp. >5 kw) High cooling rates (tp. >1 MHU/minute) High tube current variation (low inertia) Compact and robust design anode * in the order of 105 counts per reading and 103 readings per second Straton Tube Multirow Detectors for Multi-Slice CT 006 Adaptive Arra Technolog 40 mm GE 64 / 0.37 s / mm 40 mm β Philips 64 / 0.4 s / mm 19 mm 16 channels (of 103) shown Siemens 64 / 0.33 s / mm 4 1. mm 3 mm mm Toshiba M = 64 / 0.4 s / 3. Data courtes of Siemens Medical Solutions, Forchheim, German Number of simultaneousl acquired slices M / Rotation time trot / Cone-angle Γ Seite 3 3

4 Rows vs. Slices Scan Trajectories FOM 4 Rows 4 Slices 6 Rows 4 Slices 1 Rows 4 Slices 4 Rows 8 Slices M = 1 4 M = 1 4 M = 4 1 M = 4 d d p = 1.5 p = 0. 9 M S M S 1 p = N Spiral Sequence Circle rot CT Basics From Single-Slice to Cone-Beam Spiral CT Animation b Udo Buhl, Aachen Technolog Basic parameters Detector concepts, tube technolog Scan trajectories, scan modes Algorithms D filtered backprojection Spiral -interpolation ASSR and EPBP (cone-beam recon) Phase-correlated CT (e.g. cardiac CT) Image qualit and dose Spatial resolution (PSF, SSP, MTF) Relation of noise, dose and resolution Dose values (CTDI, patient dose) Dose reduction techniques Emission vs. Transmission D: In-Plane Geometr Emission tomograph Infinitel man sources No source trajector Detector trajector ma be an issue 3D reconstruction relativel simple Transmission tomograph A single source Source trajector is the major issue Detector trajector is an important issue 3D reconstruction etremel difficult Decouples from longitudinal geometr Useful for man imaging tasks Eas to understand D reconstruction Rebinning = resampling, resorting Filtered backprojection Seite 4 4

5 Fan-beam geometr Parallel-beam geometr transaial rebinning ( β, α ) ( ξ, ϑ) Fan-beam geometr Parallel-beam geometr In-Plane Parallel Beam Geometr β R F α ϑ ξ ϑ ξ ( β, α ) ( ξ, ϑ) Measurement: p( ϑ, ξ ) = Rf ( ϑ, ξ ) = d d f (, ) δ cosϑ + sinϑ ξ ( ) Measurement: FBP (Filtered Backprojection) ( cosϑ + sinϑ ) p( ϑ, ξ ) = d d f (, ) δ ξ D Backprojection (Discrete Version of the Transpose Radon Transform) 1D FT: π iξu π iu( cosϑ + sinϑ) dξ p( ϑ, ξ ) e = d d f (, ) e Central slice theorem: P ( ϑ, u) = F( u cosϑ, u sin ϑ ) Inversion: f (, ) = π 0 π π iu( cosϑ + sinϑ) dϑ du u P ( ϑ, u) e = dϑ p( ϑ, ξ ) k( ξ ) 0 ξ = cosϑ + sinϑ p( ϑ, ξ ) Add ra value to each piel in the vicinit of the ra. Seite 5 5

6 Spiral CT Scanning Principle Filtered Backprojection (FBP) Scan trajector Start of spiral scan 1. Filter projection data with the reconstruction kernel.. Backproject the filtered data into the image: Direction of continuous patient transport Reconstruction kernels balance between spatial resolution and image noise. 1996: 1998: 00: 1 5 mm, 0.75 s 4 1 mm, 0.5 s mm, 0.4 s 0 0 t 004: mm, 0.33 s Kalender et al., Radiolog 173(P):414 (1989) and 176: (1990) Spiral -Interpolation for Single-Slice CT M=1 without -interpolation with -interpolation d p= M S = R Spiral -interpolation is tpicall a linear interpolation between points adjacent to the reconstruction position to obtain circular scan data. Spiral -Filtering for Multi-Slice CT M=,, 6 d p= M S CT Angiograph: Aillo-femoral bpass 1.5 M=4 10 cm in 40 s 0.5 s per rotation 4.5 mm collimation pitch 1.5 = R Spiral -filtering is collecting data points weighted with a triangular or trapeoidal distance weight to obtain circular scan data. Seite 6 6

7 The Cone-Beam Problem Animation b Siemens 1 5 mm 0.75 s 4 1 mm 0.5 s mm s mm s mm << 1 s? Cone-Beam Artifacts Advanced Single-Slice Rebinning (ASSR) 3D and 4D Image Reconstruction for Small Cone Angles Cone-angle Γ = 6 Cone-angle Γ = 14 Cone-angle Γ = 8 Defrise phantom fokus trajector First practical solution to the cone-beam problem in medical CT Reduction of 3D data to D slices Commerciall implemented as AMPR ASSR is recommended for up to 64 slices Do not confuse the transmission algorithm ASSR with the emission algorithm SSRB! Kachelrieß M, Schaller S, Kalender WA. Med Phs 000; 7(4): The Reconstruction Plane For each reconstruction position α R minimie the mean deviation of the R plane and the spiral segment around. α R R : n r c = 0 sinγ cosϕ n = sinγ sinϕ cosγ Resulting mean deviation at : R F R at R : M n γ α R 3 intersections for each R-plane mean d ' mean d Kachelrieß M, Schaller S, Kalender WA. Med Phs 000; 7(4): τ d dfiltering in the Image Domain No in-plane interpolations Interpolation along d Arbitrar d-filter width primar, tilted images Kachelrieß M, Schaller S, Kalender WA. Med Phs 000; 7(4): d,,ξ R final, transaial images Seite 7 7

8 Comparison to Other Approimate Algorithms 180 LI d=1.5mm Π d=64mm Patient Images with ASSR MFR d=64mm ASSR d=64mm High image qualit High performance Use of available D reconstruction hardware 100% detector usage Arbitrar pitch Sensation 16 at 0.5 s rotation mm collimation pitch cm in 9 s 1.4 GB rawdata 1400 images Bruder H, Kachelrieß M, Schaller S. SPIE Med. Imag. Conf. Proc., 3979, 000 CT-Angiograph Sensation 64 spiral scan with mm and s CTA, Sensation 16 at Data courtes of Dr. Michael Lell, Erlangen, German Etended Parallel Backprojection (EPBP) Feldkamp-Tpe Reconstruction Approimate Similar to D reconstruction: 3D and 4D Feldkamp-Tpe Image Reconstruction for Large Cone Angles volume row-wise filtering of the rawdata followed b backprojection True 3D volumetric backprojection along the original ra direction Compared to ASSR: larger cone-angles possible lower reconstruction speed requires 3D backprojection hardware Trajectories: circle, sequence, spiral Scan modes: standard, phase-correlated Rebinning: aimuthal + longitudinal + radial Feldkamp-tpe: convolution + true 3D backprojection 100% detector usage Fast and efficient ra 3D backprojection Kachelrieß et al., Med. Phs. 31(6): , 004 Seite 8 8

9 Kmo l β 3-fold C longitudinall rebinned detector 4-fold C+B The complicated pattern of overlapping data will become even more complicated with phase-correlation. 5-fold C Individual voel-bvoel weighting and normaliation. C: Area used for convolution B: Area used for backprojection Kachelrieß et al., Med. Phs. 31(6): , 004 ECG Spiral EPBP Std p = The 180 Condition Spiral EPBP Std p = 1.0 ϑ dϑ w(ϑ ) = π 180 in 3 segments and r w(ϑ + kπ ) = 1 Spiral ASSR Std p = 1.0 k The (weighted) contributions to each object point must make up an interval of 180 and weight 1. Kachelrieß et al., Med. Phs. 31(6): , 004 EPBP Std EPBP CI, 0% K-K 56 slices (0/300) EPBP CI, 50% K-K Advantages of Multi-Slice Spiral CT Image qualit independent of scan parameters Increase (up to a factor of M) of scan speed of -resolution New applications CT angiograph dnamic studies virtual endoscop cardiac CT Toda, complete anatomical regions are routinel scanned with MSCT within a few seconds with isotropic sub-millimeter spatial resolution. Patient eample, 30.6 mm, -FFS, p=0.3, trot=0.375 s. Seite 9 9

10 Motion Artifacts of the Heart Cardiac CT Periodic motion Snchronisation needed (ECG, Kmogram, others) Prospective Gating Phase-correlated reconstruction = Retrospective Gating Single-phase (partial scan) approaches, e.g. 180 MCD Bi-phase approaches, e.g. ACV (Flohr et al.) Multi-phase Cardio Interpolation methods, e.g. 180 MCI (gold-standard) Generations Single-slice spiral CT: 180 CD, 180 CI (introduced 1996 * ) Multi-slice spiral CT: 180 MCD, 180 MCI (introduced 1998 * ) Cone-beam spiral CT: ASSR CD, ASSR CI (introduced 000 * ) Wide cone-beam CT: EPBP (introduced 00 * ) * Med. Phs. 5(1) 1998, Med. Phs. 7(8) 000, Proc. Full 3D 001, Med. Phs. 31(6) 004 Snchroniation with the Heart Phase Allowed data ranges phase width c t eff = width / heart rate e.g. 15% / 60bpm = 150ms Heart motion Maimum Pitch for Full Phase Selectivit Voel illumination must eceed one motion ccle Table increment per motion ccle must not eceed collimation p f H t rot R R R R 0 t rot t rot 3t rot 4t rot 5t rot 6t rot Snc-Signal ECG, Kmogram,... t E.g. t rot = 0.5 s and f H = 60 bpm implies p < 0.5 The smaller the pitch value the more segments can be combined Width, and thus t eff, corresponds to the FWTM of the phase contribution profile. Kachelrieß et al., Radiolog 05(P):15, (1997) Partial Scan Reconstruction Multi-Segment Reconstruction Use one segment of 180 +δ data of phase-coherent data for a selected heart phase Table position 1. Detector. Detector 3. Detector 4. Detector Combine n segments to obtain 180 +δ of phase-coherent data for a selected heart phase Table position 1. Detector. Detector 3. Detector 4. Detector Time Time Heartbeat 1 Heartbeat Heartbeat 3 Heartbeat 1 Heartbeat Heartbeat 3 1 Partial scan data (180 + fan angle) Effective scan time t eff t rot / t eff 00 ms at t rot = 0.4 s 3 1 Partial scan data (180 + fan angle) Effective scan time t eff 48 ms tp ms at t rot = 0.4 s Kachelrieß, Ulheimer, Kalender, Med. Phs. 7(8): (000) Kachelrieß, Ulheimer, Kalender, Med. Phs. 7(8): (000) Seite 10 10

11 Multi-Threaded CT, Dual Source CT Siemens SOMATOM Definition dual source cone-beam spiral CT at the IMP Volume Zoom, 4.5 mm, 0.5 s, 1998 Multi-segment 180 MCI reconstruction, 90 bpm Sensation 64, mm, 0.33 s, 004 Data courtes of Stephan Achenbach CT Basics From Single-Slice to Cone-Beam Spiral CT Technolog Basic parameters Detector concepts, tube technolog Scan trajectories, scan modes Algorithms D filtered backprojection Spiral -interpolation ASSR and EPBP (cone-beam recon.) Phase-correlated CT (e.g. cardiac CT) Image qualit and dose Spatial resolution (PSF, SSP, MTF) Relation of noise, dose and resolution Dose values (CTDI, patient dose) Dose reduction techniques CT-value / HU water air spong. bone lungs µ ( r) µ CT( r) = µ Water compact bone fat Water kidne 1000 HU What is Displaed? pancreas blood liver ImpactX.oc V Spatial Resolution 1 out In-plane resolution -resolution 0 1 in 0.5 mm 0.4 mm out 0 1 in 0.4 mm out 0 1 in Std. scan, / UHR scan, / Std. or UHR scan, / (0, 5000) (0, 1000) (-750, 1000) Sensation 64, collimation: mm Seite 11 11

12 Spatial Resolution In-plane resolution -resolution Spatial Resolution 3 Point Spread Function (PSF), Slice Sensitivit Profile (SSP) 0.4 mm 0.3 mm 0.5 mm ImpactX.oc V FWHM 0.6 mm Standard (no FFS) FWHM = 1.3 S Double - sampling (FFS) FWHM = 1.0 S FWTM Std. scan, / UHR scan, / Std. or UHR scan, / FWHM = S eff = effective slice thickness = freel selectable parameter during image recon. Sensation 64, collimation: mm Sensation 64, collimation: mm Tricks to Improve Resolution Sharp reconstruction kernels Lowest possible S eff Decrease the sie of the detector piels Oversampling FFS αffs Detector quarter offset Use of detector combs However, image noise becomes crucial! Dependencies of IQ and Dose Image qualit is determined b spatial resolution and contrast resolution (image noise) Image noise decreases with the square-root of dose σ Dose increases with the fourth power of the spatial resolution for a given object and image noise ( σ / µ ) Noise relative to the background (=1/SNR) 1 D 1 mas eff e µ d µ R Fourth power of the resolution element sie Dose Calculator Patient Dose in CT Tpical Values for 16-Slice Scanners Head Thora Abdomen Pelvis Scan range / mm Scan time / s Collimation / mm Eff. mas / mas Critical organ Brain Lung Stomach Colon Organ dose / msv Eff. dose / msv Eff. dose /.1 msv Routine protocols, 10 kv, male phantom. Demo version of ImpactDose available at Seite 1 1

13 Strategies for Dose Reduction Standard Displa Potential reasons for an increase: Higher volume coverage Multiphasic eaminations More eaminations Higher spatial resolution New special applications Potential was to decrease dose: New displa techniques Advanced reconstruction (MAF) Automatic eposure control (AEC) Optimied spectra Dose training (dose tutor) 0,5 0,5 0,5 mm3 C = 50 HU, W = 400 HU Inside Stor (Overeposure), Elvgren, 1959 Tube Current Modulation Sliding Thin Slab (STS) Displa (attenuation: 000) 0,5 0,5 10 mm3 C = 50 HU, W = 400 HU (attenuation: 50) Constant tube current: High, inhomogeneous noise. σ piel = const. σ projection,n n Tube Current Modulation Dose Reduction b Tube Current Modulation Rule of thumb: The number of quanta reaching the center of the patient should be constant for all view angles (attenuation: 000) (attenuation: 50) Constant total mas! Conventional scan: 37 mas Modulated tube current: Low, homogeneous noise. σ piel = const. Online current modulation: 166 mas 53% dose reduction on average for the shoulder region 49% dose reduction in this case σ projection,n Kalender WA et al. Med Phs 1999; 6(11):48-53 n Seite 13 13

14 Multidimensional Adaptive Filtering (MAF) Automatic Eposure Control (AEC) (-dependent + angular dependent tube current modulation) Standard CT C 40/W 500 a) Rawdata based Local smoothing of nois data (less than 5% modification) No loss of spatial resolution Efficient Noise reduction can be equivalentl converted to dose reduction rel. image noise rel. image noise rel. tube current rel. tube current AEC transaial filtering β a) Low attenuation: b) High attenuation: pmaf ( β, α, b) = C 40/W 500 b) α Filter width = 0 Filter width > 0 dβ dα db f β ( β β ) f α (α α ) f b (b b ) p ( β, α, b ) 34% mas reduction with AEC at constant image qualit for that specific case Kachelrieß M, Watke O, Kalender WA. Med Phs 001; 8: Standard 180 MFI Standard 180 MFI Noise image (standard 180 MFI) 100% 1% data modified Adaptive 180 MAF Adaptive 180 MAF Noise image (adaptive 180 MAF) 51% 180 MAF relative to 180 MFI: Noise left: 61% center: 63% right: 60% upper: 100% Difference image Dose 37% 40% 36% 100% Difference images Resolution 97% 97% 97% 100% Noise in the shoulder region tpicall reduced to 50%...70%. collimation 4 1 mm, d = 5 mm, (C=0 / W=500) collimation 4 1 mm, d = 5 mm, (C=0 / W=500) Summar Thank You! CT technolog is further evolving towards more slices faster rotation times higher spatial resolution CT algorithms reconstruct cone-beam data for an trajector perform phase-correlated imaging (4D) significantl reduce artifacts (beam hardening, truncation ) CT dose is becoming more and more an important issue (also in the US) is being reduced b manufacturers efforts (e.g. MAF, AEC) can be most significantl reduced b user training 3d ll.org Seite 14 14