Model-Based decomposition of myocardial strains: activation time and contractility mapping Borut Kirn Department of Biomedical Engineering University of Maastricht The Netherlands In collaboration Institute of Physiology University of Ljubljana Slovenia

Detection of cardiac motion MRI-taggingUS - speckle tracking

Coordinated contraction Circumferential strain (ε cc )

Left bundle brench block (LBBB) and Ischemia Discoordinated contraction LBBB LBBB + Ischemia

Clinical problem In cardiac resynchronization therapy patients are selected upon: QRS duration (LBBB) Heart failure indices (LV dilatation, low EF) 30% of patients show no benefit + 20% no reduction of LV dilatation Can we improve patients selection and PM positioning using mechanical indices?

Normal conduction Time[ms] P Q R S Short QRS duration

Conduction during LBBB Prolonged QRS duration Time[ms] P Q R S

Ischemia

onset of shortening Mechanical indices of asynchrony

Onset of shortening time is not activation time Early activated regions are not detected Activation time is only one component of dyscoordination However,

Aims Design of a model to simulate local circumferential shortening (ε cc ) for different –activation time (Act) –contractility (Con) Mapping by inverse use of model fit to ε cc –map of Act –map of Con

CircAdapt model of heart and circulation Dynamic(t) Compliances Inertias Non-linear Modeling of circulation - lumped model in modules: chambers, tubes, valves Arts T et al. Am J Physiol. 2005;288:H1943-H1954 Adaptation of modules to load, i.e. wall mass, cavity size

Characteristics of CircAdapt Modular architecture Structured parameter data base

C i r c A d a p t CavityMech Tubes SarcMech (Act, Con) Timing Valves RepSarc SarcMech (Act, Con) SarcMech (Act, Con) SarcMech (Act, Con) SarcMech (Act, Con) SarcMech (Act, Con) SarcMech (Act, Con) SarcMech (Act n, Con n ) Act, Con Act n-2, Con n-2 Act n-1, Con n-1 Act n, Con n CircAdapt with Multi-segment myofiber

Sarcomere element : Passive Active n n-1 Assumption: Equal stress in all regions of ventricular wall

Solving inverse problem INPUT PARAMETERS: MODEL RESULT: Act 1, Act 2 … Act 160 Con 1, Con 2 … Con 160 ε cc,1, ε cc,2 …. ε cc,160 compare measured ε cc simulated ε cc MAPS:

Sarcomere length Solution 1: 159 sarc.el. with Act=0; Con=1 1 sarc.el. with Act=?; Con=? Influence of Activation time:Influence of Contractility: time Sarcomere length

Solution 2: linear decomposition

Reconstruction of maps, using MRI-tagging measurement on a dog model: ischemia induced by ligation apex base septum anterior Contractility (normalized)

Reconstruction of maps, using MRI-tagging measurement on a dog model: ischemia induced by ligation apex base septum anterior Activation time [ms]

Activation timeContractility Reconstruction of maps, using MRI-tagging measurement on human: LBBB

Activation timeContractility Reconstruction of maps, using MRI-tagging measurement on human: LBBB + Ischemic cardiomyopathy

Activation timeContractility Reconstruction of maps, using MRI-tagging measurement on human: healthy

healthy +ischemia +LBBB Activation time Contractility Reconstruction of maps, using MRI-tagging measurement on a dog model: healthy, +ischemia, +LBBB

Conclusions The circumferential strain as a function of time (ε cc ) in asynchronous contracting left ventricle (LV) was modelled as a long fibre around the LV, consisting of a series of fibre segments, each having its own activation time (Act) and contractility (Con). Applying the model inversely, the measured maps of regional ε cc were converted to maps of activation time and contractility. Obtained maps were in agreement with clinical diagnosis of LBBB and ischemia in animal experiments and in patients.

Colaborators Theo Arts, Joost Lumens, Tammo Delhaas, Wilco Kroon and Frits Prinzen