Objectives Cynthia K. Rigsby Children s Memorial Hospital Chicago, IL CMR volumetric analysis Techniques Normalized data Sources of error CMR phase contrast flow analysis Techniques What we can do with the numbers Sources of error Functional and flow analysis: the power of combined data CMR Volumetric Analysis Assessment of ventricular volumes and function paramount for diagnosis and follow up of congenital heart disease MR functional analysis validated Excellent correlation between MR volumetry and actual cavity volume by cast model Gold standard for volumetric analysis Rehr R. Radiology 1985; 156:717. Cranney GB. Circulation 1990; 82:154. Sakuma H. Radiology 1993; 188: 377. Pennell D. Eur H Journal 2004; 25: 1940. CMR Volumetric Analysis Technique ECG gated cine SSFP SSFP image contrast depends on T2/T1 tissue ratio Independent of through plane blood flow Good myocardial to blood pool contrast Good border detection Accurate assessment of volumes CMR Volumetric Analysis Short axis standard imaging plane Standard MR imaging planes not cardiac planes Need to account for cardiac rotation in all planes to achieve cardiac imaging planes Different methods Cardiac imaging plane acquisition Axial scout 2 chamber 1
Cardiac imaging plane acquisition 2 chamber pseudo 4 chamber Cardiac imaging plane acquisition short axis Cardiac imaging plane acquisition true 4 chamber Cardiac imaging plane acquisition 4 chamber parallel to AV valve 12 slices short axis 2 chamber parallel to AV valve Volumetric Analysis RV and LV endocardial and epicardial systolic and diastolic contours RV and LV EDV, ESV Stroke volume Cardiac output Ejection fraction Ventricular mass Functional analysis 8 yo s/p arterial switch for D-TGV LV stroke volume slightly greater than RV stroke volume due to mild aortic insufficiency LVEF 56%; RVEF 57%; normal volumes and mass Normal septal motion Alfakih K. JMRI 2003; 18:25 2
Value of normalized data Need for normalized data 9 month old pulmonary htn RVEDV 50 ml; 160 ml/m2 19 yo TOF repair RVEDV 123 ml; 171 ml/m2 Relation of heart and body size is allometric Relationship changes from infancy to adulthood Somatic growth Steady rise in ventricular EDV, ESV and RV, LV mass by age for girls and boys Volumes 10 15% greater for boys than girls Biventricular EF constant throughout growth, not gender-specific Published data normalized to BSA Several different BSA calculations Sarikouch S. Circ Cardiovasc Imaging 2010; 3:65. 8-20 yrs Helbing WA. J. Magn. Reson. Imaging 2009;29:552 559. 8-17 yrs Need for normal data 120 healthy adults (20-80 yrs) SSFP short axis imaging Normalized RV mass and EDV and ESV decrease with age RV EF increases with age Absolute and normalized RV volumes and mass significantly larger in males Increased BSA associated with increased RV volume and mass Need for normal data Younger children Acquisition of normal data difficult as Sedation or anesthesia required May alter physiology Difficult to scan normal volunteers Small patients stress SSFP parameters May need to use GRE sequence depending on system and coils Volumetric data extremely important Decisions on intervention or reintervention based on volumetric data Maciera A. Eur H Journal 2006; 27:2879. CMR Volumetric Analysis Technique Study evaluating RV volume and reproducibility Comparison of LV and RV volumes SAX and axial SSFP acquisitions 20 normal volunteers (32-60 yrs; mean 45.4 yrs) Statistically significant differences with smaller mean RV EDV, ESV on axial acquisition Correlation between RV and LV stroke volume similar leading to similar EF s Axial measurements more reproducible Alfakih K. JMRI 2003; 18:25. Sources of error Basal slices Enlarged RV Difficult to locate AV valve plane on basal ventricular SAX images Tricuspid valve plane may not be parallel to mitral valve plane Utilize other planes to help localize AV valves 3
Sources of error Include RV basal of AV valve in imaging CMR Volumetric Analysis Technique Axial acquisition TV, MV profiled Pulmonary valve may be difficult to assess, but RVOT volume generally smaller cross sectional area Difficult with GRE reduced myocardial/blood pool due to inplane flow Fewer issues with SSFP Inferior wall difficult to assess CMR Volumetric Analysis Technique 46 patients with repaired TOF Axial to short axis acquisition End diastolic and end systolic RV and LV volumes Two investigators Interobserver variance smaller for RV EDV, RV ESF, LV EDV, and LV ESF with axial acquisition Volumetric data more reproducible with axial acquisition No gold standard volume comparison Fratz. Am J Cardiol. 2009; 103: 1764. Sources of error GRE imaging GRE SAX good signal as inflow is perpendicular to acquisition Signal depends on inflow May be more difficult to detect borders Ventricular volumes higher, myocardial mass lower on SSFP Consider when measuring volumes at 3T and in infants if using GRE Hudsmith LE. J Magn Reson Imaging. 2006; 24:312. Pennell D. Eur H Journal 2006; 27: 2879. Sources of error Inspiration vs expiration Inspiration LV CO 4.3 l/min/m2 LVEF 50% RV CO 3.7 l/min/m2 RVEF 54% Correlation between thoracic excursion (inspiration or expiration) and LVSV during normal breathing in the same subject measured by real time asc ao PC MRI Expiration LV CO 5.0 l/min/m2 LVEF 55% RV CO 4.9 l/min/m2 RVEF 59% van den Hout R J.Radiology 2003;229:513-519 au = arbitrary units Think about right heart 4
Sources of error Papillary muscles Include or not? Adult study Normals and patients Statistically significant differences in stroke volumes, RV and LV ESV, LVEDV No statistically significant difference in RVEDV or EF Increased analysis time for papillary muscle inclusion (25 min) vs not (13 min) Tools need to allow CMR Volumetric Analysis Technique Be consistent in acquisition method and post processing Know the limitations Sievers B. JCMR. 2004; 6 No. I: 9-16. Phase contrast (PC) velocity imaging Direct non invasive measurement of blood flow velocity, direction, flow rate Proven tool Valve regurgitation and stenosis Relative lung blood flow Shunt fraction (Qp:Qs) Assessment of collateral flow Myocardial wall motion Spins moving along a magnetic field gradient acquire phase shift relative to stationary spins Two velocity encoding gradients applied Phase shift proportional to moving spin velocity Lotz J et al. Radiographics 2002;22:651-671 MR data acquisition yields information on magnitude and phase of each voxel Conventional MRI Phase information discarded PC MR Phase information used to calculate velocity per voxel yielding a velocity image Two images for every acquisition PC velocity imaging Flow direction White forward Black reverse Gray stationary Speckle no data Velocity Direction and value Flow volume Velocity Vessel area 5
Phase contrast velocity imaging Validation in vitro and in vivo Accuracy depends on acquisition technique Many sources of potential error Data not able to all be acquired in one heart beat on most systems Retrospective ECG gating Data acquisition continuous At acquisition completion, data assigned to intervals of the cardiac cycle No data missed Prospective ECG gating Acquisition triggered at beginning of cardiac cycle Paused to wait for next trigger at end diastole Cardiac cycle at constant length Data may not be collected at end diastole Less arrhythmia sensitive Courtesy Siemens Optimize misregistration artifact and temporal resolution Minimize TR and TE Image perpendicular to vessel for evaluation of flow Capture vessel cross section +/- 15 degrees tolerated Adequate temporal resolution 30 true frames/beat Especially important for velocity Adequate spatial resolution 4 pixels across vessel of interest 4-8 mm slice thickness Vessel in center of FOV Lotz. RadioGraphics 2002; 22:651. Gatehouse. Eur Radiol. 2005; 15:1272. Greil. JMRI. 2002; 15: 47. Appropriate venc Velocity within vessel + 15% Increased noise with higher venc Aliasing Velocity scout 200 300 PC velocity imaging Free breathing acquisition Multiple signal averages Segmented k-space Round vessel No aliasing 350 6
PC velocity imaging flow curve Phase contrast imaging max velocity/gradient Regurgitant fraction Backward flow/forward flow 0.2 l/min/2.8 l/min 6% regurgitant fraction Peak velocity R F R In plane phase contrast Determine maximum velocity Image along vessel of interest Look for aliasing Use for phase contrast cross sectional imaging Adequate temp res Gradient = 4v 2 17 yo TGA, ASO with LPA stent and RPA stenosis Maximum velocity 3.2 m/s PC velocity imaging Phase offset errors Significant inaccuracies can occur due to Eddy currents and other effects that result in a spatially varying, but temporally stable artificial background phase Stationary tissue shows apparent velocity Increases when increased distance from center of FOV Result in significant errors in low flow parameters such as valvular regurgitation PV velocity imaging Phase offset errors Phase offset errors tested on 3 major vendors systems Phantom study; worst case scenario Greatest uncorrected velocity averaged 2.7 cm/s All systems exceeded allowable error 0.6 cm/s All systems need correction Gatehouse et al. Journal of Cardiovascular Magnetic Resonance 2010, 12:5 Gatehouse et al. Journal of Cardiovascular Magnetic Resonance 2010, 12:5 PC velocity imaging Methods of background correction ROI in stationary tissue adjacent to vessel and subtract any non-stationary velocity from vessel No stationary tissue adjacent to heart or great vessels ROI at distance problematic because offset errors vary spatially Measure phase offset directly by repeating PC imaging sequence with a stationary phantom Requires time Offset directly proportional to vessel area PC velocity imaging Methods of background correction Spatially dependent background phase correction as part of image reconstruction Based on identification of most stationary pixels, least squares fit performed Calculates surface representing the background phase for all pixels in the image Correction applied to all PC images of the time series Hard to quantify as no gold standard in vivo Tend to correlate better with ventricular stroke vols Wolff. JCMR. 2007; 9: 681. Courtesy of Siemens. 7
Example of background correction Example of background correction Pulmonary valve 2.8/35 = 8% 2.1/36 = 6% Example of background correction Pulmonary valve Example of background correction Aortic valve Net flow 8.6 l/min Net flow 7.9 l/min 52/81 = 64% 54/78 = 69% Flow sensitive 4D MRI PC MRI sequences account for a significant amount of a pediatric CMR study time 2-13 PC MRI acquisitions/case Techniques to shorten acquisition time Spiral trajectory with parallel imaging 4D volumetric acquisition Steeden JA. Radiology. 2011; 260: 79. 3 dimensional spatially encoded, three directionally encoded time resolved cine velocity acquisition ECG and respiratory navigator gated Spatial res 2 mm 3 ; temporal res 4-8 TR 6-20 minutes for thoracic aortic volume 8-20 min for whole heart coverage Scan time depends on HR and efficiency of respiratory control Significant post-processing computing needed and not commercially available Can retrospectively assess any location in imaging volume Markl et al. Journal of Cardiovascular Magnetic Resonance 2011, 13:7 8
4D MR Data 3D Flow Visualization Courtesy of Dr Michael Markl Markl M, et al. J Magn Reson Imaging 2007; 25:824-831 Bock J, et al. Magn Reson Med 2010; 63:330-338 Courtesy of Dr Michael Markl PC-MRA & 3D Particle-Traces Aortic Aneurysm: Effect on blood flow in entire aorta velocity [m/s] Post coarctation repair Courtesy of Dr Michael Markl Frydrychowicz A, Markl M et al. J Cardiovasc Magn Reson 2008;10(1):30 t = 100ms t = 180ms t = 300ms t = 380ms Courtesy of Dr Michael Markl Frydrychowicz A, et al. J Cardiovasc Magn Reson 2008;10(1):30 Fontan Circulation direct return of venous blood to PA single ventricle extracardiac TCPC Courtesy of Dr Michael Markl Frydrychowicz A, et al. Circulation 2008;118:e16-17 Markl M, et al. Eur J Cardiothorac Surg. 2011;39 206-212 Markl M, et al. Eur J Cardiothorac Surg. 2011;39 206-212 9
4D viewer Functional and flow analysis: the power of combined data Case examples PAPVC Ebstein anomaly Complex CHD/single ventricle physiology 11 yo with murmur PAPVC PAPVC 3 yo Ebstein anomaly Ebstein anomaly By echo, tiny PFO with trivial mostly left to right shunting 10
Heterotaxy/asplenia with R dominant CAVC, DORV, right arch, TAPVC, hepatic veins to the LA s/p PA band, s/p oversewing of PV and bidirectional Glenn. Quantitative CMR CMR volumetric analysis Techniques Normalized data Sources of error CMR phase contrast flow analysis Techniques What we can do with the numbers Sources of error Functional and flow analysis: the power of combined data 11