Integrative System Framework for Noninvasive Understanding of Myocardial Tissue Electrophysiology Linwei Wang, Ken C.L.
Download ReportTranscript Integrative System Framework for Noninvasive Understanding of Myocardial Tissue Electrophysiology Linwei Wang, Ken C.L.
Integrative System Framework for Noninvasive Understanding of Myocardial Tissue Electrophysiology Linwei Wang, Ken C.L. Wong, Pengcheng Shi Computational System Biomedicine Laboratory B. Thomas Golisano College of Computing and Information Sciences Rochester Institute of Technology CPP, INI, Cambridge, July 2009 From patient observations to personalized electrophysiology • Noninvasive observations on specific patient: – Structural: tomographic image – Functional: projective ECG/BSPM • In clinical settings, it really has been a sophisticated pattern recognition process to decipher the information. • How can models offer help? – Better models lead to more appropriate constraints in data analysis. • Physiological plausibility vs. algorithmic/computational feasibility • Volumetric? – Ultimately, models have to be personalized to be truly meaningful. Integrative system perspective Prior Knowledge: Models • Built up over many years • General population System Modeling Patient Observations Individual Subject Personalized Information Recovery • Electrical function • System dynamics • Tissue property • System • Latent substrate observations Wang et al. IEEE-TBME, 2009 (in press). • Subject-specific information • Noisy, sparse, incomplete Data Acquisition • BSP sequence • Tomographic images Physiological model constrained statistical framework • System perspective to recover personalized cardiac electrophysiology – (phenomenal) Model constrained data analysis: prior knowledge guides a physiologicallymeaningful understanding of personal data – Data driven model personalization: patient data helps to Information Recovery System Modeling Data Acquisition Volumetric myocardial representation • Ventricular wall: point cloud Slice segmentation Surface mesh Volume representation • Fiber structure: mapped from Auckland model Surface registration Surface fiber structure 3D fiber structure Surface body torso representation • (Isotropic and homogeneous volume conductor) Patient’s images Surface model Cardiac electrophysiological system Volumetric TMP dynamics model Personalized 3DBEM mixed hearttorso model TMP-to-BSP mapping System dynamics: volumetric TMP activity • Diffusion-reaction system: 2-variable ordinary differential equation •uMeshfree representation U and 1computation t (Du) f1(u,v) v f 2 (u,v) t t M KU f1 (U,V) V f 2 (U,V) t - u: excitation variable: TMP - v: recovery variable: current - D: diffusion tensor System observation: TMP-to-BSP mapping • Quasi-static electromagnetism Poisson’s equation • Mixed meshfree and boundary element methods Governing equation Direct solution method Meshfree+BEM 2 (r) (D(r)u(r)) c( ) ( ) 1 4 ( t t Surface integral: BEM (r)q* ( ,r)dt Di (r) u(r) r dt | r | n HU t t (r) * r ( ,r)dt n 1 (Di (r)u(r))dt ) | r | Volume integral: meshfree State space system representation U 1 t M KU f1(U,V,) V f 2 (U,V,) X UT t TMP activity: Y TMP-to BSP mapping: T Uncertainty:, Nonlinear dynamic model HU VT Local linearization Temporal discretization State Xk Fd (Xk1, k1 ) k equation: Parameter: k k1 k Measurement Y HX k k k equation: Prediction Correction Large-scale & high-dimensional system U,V is of dimension 2000-3000 Monte Carlo integration Sequential data assimilation • Combination of unscented transform (UT) and Kalman filter: unscented Kalman filter Prediction: UT (MC integration + deterministic sampling) Preserve intact model nonlinearity Black-box discretization Correction: KF update Computational feasibility TMP estimator: reconstructing TMP from BSP Initialization Uˆ 0 Pˆ u0 k = k+1 Ensemble generation (unscented transform) n n ˆ ˆ ˆ k1,i 0 Uk1 Uk1 ( )Pu 2 k1 Correction (KF update) Prediction (MC integration) ( k|k1,i , k|k1,i ) F˜d ( k1,i,Vk1) ˆ U K u (Y HU ) U k k k k k Pˆ uk (I K uk H)Puk Uk W im k|k1,i ,Vk W im k|k1,i n n i 0 n c i 0 Filter Gain K uk Puk HT (HPuk HT Rvk )1 i 0 P W i ( k|k1,i U )( k|k1,i Uk )T Q u uk k k Parameter estimator: reconstructing model parameters Initialization ˆ 0 Pˆ 0 Ensemble generation (unscented transform) k1,i0 2n ˆ ˆ (n )Pˆ k1 k1 k1 k = k+1 Prediction (MC integration) ˆ , ) H ˜ F˜d (U k|k1,i k1 k1,i Correction(KF update) ˆ ˆ K (Y Y ) k k1 k k k Pˆ k P k K k Pyk K T k k|k1,i k|k1,i ˆ Q Pk P k1 k Yk W im k|k1,i 2n k yk Filter Gain i 0 P W i ( Y )(k|k1,i Yk )T R k i 0 2n ˆ )( Y )T P W c ( 2n yk c i 0 k|k1,i i k k1,i k1 k|k1,i k Kk Pk yk Pyk 1 Nonlinear measurement model Experiments (PhysioNet.org): electrocardiographic imaging of myocardial infarction • Four post-MI patients personalized heart-torso structures MRI Cardiac: 1.33×1.33×8mm Whole-body: 1.56×1.56×5mm BSP 123 electrodes, QRST @ 2KHz sampling Gd-enhanced MRI gold standard Goals and procedures • Quantitative reconstruction of tissue property and electrical functioning Tissue excitability TMP dynamics Procedures Initialization – TMP estimation with general normal model Simultaneous estimation of TMP and excitability Identify arrhythmogenic substrates (imaging + quantitative evaluation) Localize abnormality in TMP and excitability Investigate the correlation of local abnormality between TMP and excitability Result: case II • Infarct location: septal-inferior basal-middle LV Simulated normal TMP dynamics Estimated TMP dynamics Result: case II -Black contour: abnormal TMP dynamics -Color: recovered tissue excitability • Location, extent, and 3D complex shape of infarct tissues • Correlation of abnormality between electrical functions and tissue property – Abnormal electrical functioning occurs within infarct zone – Border zone exhibits normal electrical functioning Result: case II Delayed activation: 4-9 Infarct: 3-14 Delay enhanced MRI registered with epicardial electrical signals H. Ashikaga, Am J Physiol Heart Circ Physiol, 2005 Result: case I • Infarct: septal-anterior basal LV, septal middle LV Result: case III • Infarct: inferior basal-middle LV, lateral middle-apical LV Result: case V • Infarct: anterior basal LV, septal middle-apical LV Quantitative validation - EP: Percentage of infarct in ventricular mass - CE: Center of infarct, labeled by segment - Segments: A set of segments which contain infarct - SO: Percentage of correct identification compared to gold standard Comparison with existent results - EPD: difference of EP from gold standard - CED: difference of CE from gold standard - Dawoud et al: epicardial potential imaging - Farina et al: optimization of infarct model - Mneimneh et al: pure ECG analysis Conclusion • Personalized noninvasive imaging of volumetric cardiac electrophysiology Noninvasive observations Personalized volumetric cardiac electrophysiology Latent substrates