This repo contains data and analysis scripts reported in the manuscript Successful Cardiac Resynchronization Therapy Reduced Negative Septal Work in Patient-Specific Models of Dyssynchronous Heart Failure.
doi:
It contains two directories, a Patient-Specific Modeling directory and a Simplified Work Estimates directory. Documentation for each directory is below.
The methods used for simulating the patient-specific models are explained in detail in our paper "Patient-specific models of cardiac biomechanics". The patient-specific models for each patient (BiV1 to BiV8) are available in the data repository. Each folder has results from two simulation runs: LBBB baseline (Parameter 1) and CRT (Parameter 4), using the optimized activation delay for each patient. Each simulation folder contains the Continuity file for each patient with the patient-specific parameters, the complete deformed meshes for the 10th beat of simulation, and the processed stress and strain data for the 10th beat. There is also a processed file called iphases.mat that contains the timing for the opening and closing of the aortic and mitral valves.
This directory contains data and scripts for generating simplified myocardial work measurements from finite element meshes of patient-specific LV endocardium. These geometric models for ech patient are stored in the LV Geometric Models folder and include a time-series of patient-specific geometries throughout the cardiac cycle. Scripts for generating and anayzing data are stores in the scripts folder. Datasets are stored in the data folder.
These myocardial work measurements approximate the regional stress-strain loop area along the LV endocardial surface.
Regional strain is computed in CalculateTriStrain.m. This script inputs the meshes from the LV Geometric Models folder and computes strain as the change in area of each finite element over the course of the cardiac cycle. Regional stran is stored as RS_CT.mat This script also computes and stores AHA segment-based segmental strain for each patient in data file seg_strain_allpats.mat.
In this study, we evaluated five different myocardial work estimates that utilize the same regional strain measurement, but explore different approximation of regional stress.
1. Left heart catheterization-based LV pressure-strain area (PLHCSA)
Scripts needed to compute work estimates: CalculateTriMW.m
Data files needed: RS_CT.mat, WorkPSMLBBBCRT.mat
Scripts where necessary data is generated: PrincipalCurvatureAnalysis2.mat
File where results are stored: MWCT_tris.mat
Description: PLHCSA is computed in CalculateTriMW.m. This script inputs the regional strain results RS_CT.mat and the patient-specific LV pressure waveforms recorded from pre-CRT left-heart catheterization. These waveforms are stored in WorkPSMLBBBCRT.mat, which stores the pre-CRT PSM results. PLHCSA is computed as the area of the LV pressure-regional strain loop area for each element on the mesh, and the result is stored in MWCT_tris.mat.
This script also computes AHA segment-based segmental PLHCSA and PSM-derived work. These results are stored as variables segMWCT_allpats and segWork_allpats respectively in data file all_seg_work_all_pats.mat.
2. End-diastolic wall stress-strain area (WSEDSA)
3. Time-varying wall stress-strain area (WSTVSA)
Scripts needed to compute work estimates: calculateSimpleWork.m
Data files needed: Laplace_measurements.mat
Scripts where necessary data is generated: PrincipalCurvatureAnalysis2.mat
File where results are stored: Laplace_work_all_patches.mat
Description: These two work estimates leverage the law of Laplace to compute regional wall stress, which states that wall stress is proportional to cavity pressure, wall thickness, and cavity radius. WSEDSA applied regional shape information at end-diastole, while WSTVSA applied regional shape information for the whole cardiac cycle. These wall stresses are computed in calculateSimpleWork.m using the function calculatePatchWorkEstimates. (WSEDSA) is stored as variable MWCTLP and (WSTVSA) is stored as variable MWCTLPTV in the data file Laplace_work_all_patches.mat.
To calculate these two approximations, regional radius and wall thickness information is needed. Both variables are generated in script PrincipalCurvatureAnalysis2.mat. Regional radius is computed as the regional effective radius, which is calculated for each AHA segment by fitting an ellipsoid to the patient-specific geometric model. The effective radius is stored as variable r_eff_allpats. Regional wall thickness information was deried from the patient-specific geometric model. Wall thickness data is stored in the Segmental Wall Thickness Measurements folder and saved as variable thickness_data. Both variables are saved in data file Laplace_measurements.mat
The PrincipalCurvatureAnalysis2.mat script also computes AHA segment-based segmental WSEDSA and WSTVSA. These results are stored as variables segMWCTLP and segMWCTLPTV respectively in data file segMWCT_effrad_allpats.mat.
4. Generic pressure-strain area (PgenSA)
5. Scaled generic pressure-strain area (Pgen,scaledSA) Scripts needed to compute work estimates: calculateSimpleWork.m
Data files needed: genericLVPs.mat, Generic Pressure Strain Estimate/generic_LVP.csv
Scripts where necessary data is generated: evaluateLVPestimates.m
File where results are stored: GenericLVP_work_all_patches.mat
Description: These two work estimates use two different generic LV pressure waveforms as a surrogate for stress to evaluate the extent of patient-specific information needed to obtain a clinically useful myocardial work approximation. PgenSA uses a generic LV pressure waveform with no patient-specific information, while Pgen,scaledSA uses a generic LV pressure waveform scaled to patient-specific peak pressure. The digitized generic waveform is stored as Generic Pressure Strain Estimate/generic_LVP.csv. The waveforms are then computed for each patient in script evaluateLVPestimates.m and stored as variables genericLVP and genericLVP_ESP. These waveforms were then used to approximate work for each patch on the LV in script calculateSimpleWork.m. PgenSA and Pgen,scaledSA were stored as variables MWCTgenericLVP and MWCTgenericLVP_patES respectively in data file GenericLVP_work_all_patches.mat.
The evaluateLVPestimates.mat script also computes AHA segment-based segmental PgenSA and Pgen,scaledSA. These results are stored as variables segMWCT_genericLVP and segMWCT_genericLVP_patES respectively in data file segMWCT_genericLVP_allpats.mat.
Scripts needed to compare PSM-based work and simple work estimates: CRT_responder_analysis2.mat
Data files needed: WorkPSMLBBBCRT.mat, MWCT_tris.mat, Laplace_work_all_patches.mat, GenericLVP_work_all_patches.mat, PatchAreas.mat
Description: Comparison of PSM-based work and simplfied work of CRT responders and non-responders at baseline was computed in script CRT_responder_analysis2.mat. This script compares the baseline coefficient of variance of work (COVW), the LV fraction performing negative work (VfLVNW and SfLVNW), and the septal fraction performing negative work (VfSTNW and SfSTNW) between responders and non-responders. In the manuscript, COVW results are shown in Figure 5 and LV and septal negative work fraction resutls are shown in Figure 9. These results are also reported in Table 4.
Agreement between PSM-based work and simplified work estimates as described in the simplified work supplement was computed in the paper_results_figures.mat script. In this script, we compared model-based segmental peak strain and simplified segmental peak strain, and model-based segmental work and simplified segmental work across all patients. The segmental strain comparison is shown is Figure S1 and Table S1 of the supplement. The semgental work comparison across all patients is shown in Figure S2 and and for one representative patient in Figure S3. All segmental work comparisons are reported in Table S3.