Objectives Emerging XX-ray Fluoroscopic Guidance Technologies for Challenging Cardiovascular Interventions Michael A. Speidel University of Wisconsin - Madison AAPM 2009 Annual Meeting 1. Review the demands and limitations of x-ray fluoroscopy (XRF) in guided cardiac interventions - Lack of tissue contrast and depth information - X-ray dose concerns 2. Understand the principles of Inverse Geometry XRF - Scanning-Beam Digital X-ray (SBDX) prototype system - Reduction of patient x-ray dose - 3D tracking of catheter devices 3. Discuss x-ray fluoroscopy combined with 3D roadmaps - Visualization of 3D soft tissue targets - Endocardial stem cell therapy X-Ray Fluoroscopic (XRF) Guidance Basic demands on a guidance system in the cardiac cath lab: 1. XX-ray Guidance in Cardiac Interventions 1. Real time continuous feedback 2. High spatial, temporal resolution 3. Device position relative to anatomy 4. Simple to set up and use 5. Compatible with catheter devices XRF meets these requirements well in many types of interventions Coronary Angioplasty 1 Lack of Tissue Contrast and Depth Focus Ablation of Atrial Fibrillation Left Atrium Device: RF ablation catheter Pulm. veins Target: line around pulmonary vein ostia Target anatomy lacks contrast Catheter position difficult to determine relative to 3D target Endomyocardial Cell Therapy Device: injection catheter Target: viable peri-infarct zone Left ventricle X-ray Radiation Dose in the Cath Lab Deterministic risk of skin injury ( > 2 Gy to skin) Case reports of skin injury, 1996-2001 Coronary intervention vs. Cardiac RF ablation Koenig, T. et al. AJR 177, 3-11 (2001). Chida, K. et al. AJR 186, 774-778 (2006). Coronary angiography & intervention: Cardiac radiofrequency ablation: TIPS placement: Neuroradiologic intervention: Other: 44 11 6 2 3 PCI RF Abl. Fluoro time (min): 37 +/- 23 121 +/- 63 Cine runs (#): 35 +/- 17 18 +/- 12 Max skin dose (Gy): 1.45 +/- 0.99 0.64 +/- 0.55 Stochastic risk of cancer induction Infarct zone Requires delineation of soft tissue based on functional status Experimental procedure Obesity and Radiation Dose in RF ablation of Atrial Fibrillation Ector, J. et al. JACC 50, 234-242 (2007). BMI < 25 25-30 ≥ 30 n 28 41 16 Age 48 +/- 10 51 +/- 7 46 +/- 10 Calc. Effective Lifetime Attributable Risk Dose (mSv) of cancer incidence 15.2 +/- 7.9 1/1000 26.8 +/- 11.6 1/633 39.0 +/- 14.7 1/405 Guidance Solutions for the Cath Lab 2. Inverse Geometry XRF Pursue non-fluoroscopic technologies E.g. Electroanatomic mapping systems (EAM) - 3D tracking of specialized catheters - Point-by-point endocardial surface mapping - Cardiac ablation guidance ScanningScanning-Beam Digital XX-ray (SBDX) Prototype Operating Principles Dose Reduction Catheter Tracking Or seek to modify / enhance XRF guidance by: 1) Reducing x-ray dose while maintaining image quality 2) Adding 3D context to the live image display 2 SBDX Operating Principles Photon-counting Detector Array Real-time Reconstructor ~40,000 images in 1/15 sec 15-30 fps 16 planes Dose Reduction Principles 1. Beam scanning and large airgap reduces detected x-ray scatter 1500 3-7% SF Thick CdTe 2. Thick CdTe detector maintains high DQE at high source kVp 1000 500 0 -50 0 50 -50 0 50 25-50% SF Thin CsI X-ray beam Multi-hole Collimator Collimator Transmission Target 100 x 100 positions Dose Reduction Principles SBDX Prototype Performance (2006) 3. Inverse geometry reduces x-ray fluence at the patient entrance Large-area SNR 25 SBDX at 120 kVp 20 SBDX at 120 kVp SNR 15 1/r2 10 SBDX at 70 kVp 5 ~2x larger entrance field Conventional Entrance Exposure 140 II/CCD cine SBDX at equal kVp 1/r2 SBDX Conventional Electron beam 0 Entrance exposure (R/min) Scanned Focal spot 123 R/min 123 kVp II/CCD cine 120 SBDX at 120 kVp SBDX at equal kVp 100 80 60 40 18 R/min 62 kVp 9.3 R/min 20 0 18 20 22 24 26 28 30 32 phantom thickness (cm acrylic) 34 36 18 20 22 24 26 28 30 32 34 36 phantom thickness (cm acrylic) SBDX operating at equal SNR: 15% - 31% entrance exposure • Greatest dose reduction for largest phantoms SBDX Speidel, M. et al. Comparison of entrance exposure and signal to noise ratio for an SBDX prototype. Med Phys 33, 2728-2743 (2006). 3 SBDX System Development Detector Re-design 1% SFxray 6% SFxray 3% SFxray 14 Next Gen Iodine SNR 12 CineCine-quality 120 kVp, 24.3 kWp, 90% DQE 1500 120 kVp, 24.3 kWp, 71% DQE 10 8 FluoroFluoro-quality 6 Shift-and-add backprojection at multiple planes 10.6 cm x 5.3 cm area Phantom: 28 cm acrylic 16 High Speed Multiplanar Tomosynthesis ‘04 -’06 100 kVp,12.6 kWp, 62% DQE scan line Source & Detector Specs 1000 ‘98 -’03 4 2 500 70 kVp, 4.2 kWp, ~40% DQE 1996 0 0 1 2 3 4 5 6 7 8 9 0 10 -50 0 50 -50 0 50 X-ray Beam Solid Angle Ω 16 planes per frame 12 mm spacing (relative units) SNR ∝ (1 − SF ) DQE (mAs ) Ω Depth Focus Property Plane Selection Algorithm Multiplane Composite Display Rays through object originate from different spot positions Pixel-by-pixel plane selection: Plane stack In-plane High contrast, sharp appearance object position Out-of-plane: Low contrast, blurry “Score stack” Gradient filtering Display pixel from plane with highest object focus metric 4 3D Catheter Tracking Algorithm 3D Localization Generate MIP along z axis Helix of 1-mm Pt spheres Calculate center-ofmass along z Extract score vs. z distribution Z 80 Rawscore at fixed(x,y) 70 object 1 object 2 Z Y 3.0 (z) (x,y) x,y) Perform 2D connected component labeling 60 50 40 30 threshold 20 10 Z error: -0.56 +/- 0.65 mm 2.0 354 402 450 498 Planeposition Position ZZ (mm) Plane (mm) Output is a set of (x,y,z) coordinates for each image frame sphere size 2σ 1σ 1.0 0.0 -1.0 -2.0 -3.0 0 0 X Tracking Accuracy & Precision SBDX Prototype Geometry Z-coordinate Error (mm) Segmentation Score Image Stack Tracking Simulation Study (2008) 5 10 15 20 25 30 Source power (kWp) 12 mm plane-to-plane spacing 28 cm acrylic, 120 kVp Stationary helix Speidel M. et al. Frame-by-frame 3D catheter tracking methods for an inverse geometry [...] Proc SPIE 6913 (2008) Tracking Phantom Study 3D Tracking Demonstration 3M chest phantom Ablation catheter in trans-septal sheath AngiogramSam cardiac chamber phantom Linear stage for catheter pullback 10 mm/sec pullback rate Catheter sheath Fiducials with screw mounts SBDX source Tracking performed in software using stored detector images 15 frame/sec SBDX imaging 1850 photons/mm2 at isocenter 5 Comparison with CT Inverse Geometry XRF & 3D Tracking SBDX Tracked tip to sheath centerline: 1.0 +/- 0.8 mm (Tip diameter = 2.5 mm) 82% of tracked positions inside sheath volume catheter tip confining sheath Well-suited to long, complex cardiac interventions Real-time 3D tracking at end diastole Fluoroscopy at 15% skin dose rate 470 ring in sheath ABL tip (in focus) Left Atrium XYThresh=4, ZThresh=2 Frame 15 480 tracked tip positions sheath volume 460 CS tip 450 440 lasso cath. 430 CS tip (blurred) Z 410 Y CT scan Z plane = 427 mm ABL tip 420 X Endocardial target lasso 400 40 20 0 -20 -40 40 20 0 RF ablation catheter -20 Tracking works with standard catheters, any number of elements, and uses a single gantry angle, automatically registered to XRF system without calibration Targeted CellCell-based Therapy for MI 3. XRF / 3D Roadmap Fusion Laboratory of Amish Raval, Raval, M.D. UWUW-Madison Cardiology Stem cell therapy may improve left ventricle function after recent myocardial infarction (acute MI) [1] Endomyocardial Cell Therapy Device: injection catheter Left ventricle Direct endomyocardial cell injection requires guidance system beyond XRF in order to: 1) Target peri-infarct region 2) Avoid perforating friable infarct XRF / 3D MRI fusion enables device & target visualization while minimizing tissue contact [2] Target: Viable peri-infarct zone Avoid: Infarct [1] Segers, V. and Lee, R. Stem-cell therapy for cardiac disease. Nature 451, 937-942 (2008). [2] de Silva, Gutierrez, et al. X-ray fused with magnetic resonance imaging to target endomyocardial injections. Circulation 114 (2006). 6 BiBi-plane XRF / 3D Fusion System Conventional Bi-plane Fusion Display Control Display XRF / 3D Fusion Procedure MRI Scanner X-Ray Fluoroscopy Segmentation Workstation Slice Contours Combine with 3D XRF Model PC Workstation Portable Fusion System Frame grabber (Helios eA, Matrox) DICOM MR or CT data 3D Surface Generation C-arm Calibration (one-time) Projection Matrices Custom fusion software (C++) Surface Projection and Overlay Live Video Gantry Orientation Manual Adjustments Frames Frame Grabber Fusion Display Tomkowiak M. et al. Multimodality image merge to guide catheter based injection of biologic agents. RSNA, Chicago, IL, 2008. Porcine Study: Segmentation Pre-intervention MRI LV Endocardial contour Epicardial contour 3D Model Red: LV endocardium Yellow: infarct volume Porcine Study: Registration Manual Registration to Internal Anatomy Biplane Ventriculogram (end diastole, end expiration) bSSFP scan DHE scan Infarct contour End diastole, end expiration Frontal plane Lateral plane 7 Porcine Study: Injections Porcine Study: Targeting Accuracy Post-procedure: Cath lab: Frontal plane Lateral plane Virtual 3D marker Biplane XRF / 3D Fusion MRI Necropsy Yellow: infarct Orange: injection Bullseye display 6 animal studies: Study time: 24 +/- 12 min Injected mixture iodinated contrast : intra procedure myo. staining iron oxide (SPIO) : MRI visualization of injections tissue dye : for post procedure necropsy D1 injection point to infarct perimeter Targeting accuracy depends on the quality of: Fusion System Development Desired features: - Respiratory and patient motion compensation - Ability to re-check registration throughout procedure - Cardiac gating - Automation, to the extent it is safe and reliable (gantry calibration) (depends on modality) (landmarks) MRI and X-ray fusion method feasible and safe for targeted injections to the peri-infarct region - No myocardial perforation - Targeting error ~ MR slice thickness & in-plane resolution Portability and vendor-independence 28 injection sites: D2 – D1 = 3.6 +/- 2.3 mm Supposed distance vs. Actual distance XRF / MRI Roadmap Fusion - Modeling of XRF system - Segmentation of 3D images - Registration of 3D surface to live x-ray D2 Automated device and anatomic landmark tracking - Conventional XRF tracking Ultrasound EAM systems (2D imaging) (3D imaging) (3D points) - Inverse Geometry XRF (tomosynthesis, 3D tracking) 8 Emerging Fluoroscopic Technologies XRF Guidance: Advantages and limitations - High quality, real-time imaging - Device compatibility - Simple, easy use Conclusion - Poor 3D visualization of devices and endocardial targets - Radiation dose in long procedures Inverse geometry XRF: Unique design and capabilities - Narrow scanning x-ray beam - Inverted x-ray field geometry - High speed multiplane tomosynthesis Low dose fluoroscopy 3D tracking capability XRF / 3D Fusion: 3D anatomy & function in the cath lab - Enables novel cardiac interventions - Non-contact visualization of function - 3D soft tissue anatomy Acknowledgements Financial support for this work was provided by: University of Wisconsin Amish Raval, M.D. Andrew Klein, M.D. Douglas Kopp, M.D. NHLBI R01 HL084022 NovaRay Medical, Inc. Michael Van Lysel, Ph.D. Michael Tomkowiak, M.S. Karl Vigen, Ph.D. Timothy Hacker, Ph.D. Larry Whitesell TripleRing Technologies, Inc. Joseph Heanue, Ph.D. Augustus Lowell, Ph.D. Brian Wilfley, Ph.D. Thank you! 9