Gaussian Process Regressions for Inverse Problems and Parameter Searches in Models of Ventricular Mechanics
Patient specific models of ventricular mechanics require the optimization of their many parameters under the uncertainties associated with imaging of cardiac function. We present a strategy to reduce the complexity of parametric searches for 3-D FE models of left ventricular contraction. The study e...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2018-08-01
|
| Series: | Frontiers in Physiology |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/article/10.3389/fphys.2018.01002/full |
| _version_ | 1811278660011819008 |
|---|---|
| author | Paolo Di Achille Ahmed Harouni Svyatoslav Khamzin Svyatoslav Khamzin Olga Solovyova Olga Solovyova John J. Rice Viatcheslav Gurev |
| author_facet | Paolo Di Achille Ahmed Harouni Svyatoslav Khamzin Svyatoslav Khamzin Olga Solovyova Olga Solovyova John J. Rice Viatcheslav Gurev |
| author_sort | Paolo Di Achille |
| collection | DOAJ |
| description | Patient specific models of ventricular mechanics require the optimization of their many parameters under the uncertainties associated with imaging of cardiac function. We present a strategy to reduce the complexity of parametric searches for 3-D FE models of left ventricular contraction. The study employs automatic image segmentation and analysis of an image database to gain geometric features for several classes of patients. Statistical distributions of geometric parameters are then used to design parametric studies investigating the effects of: (1) passive material properties during ventricular filling, and (2) infarct geometry on ventricular contraction in patients after a heart attack. Gaussian Process regression is used in both cases to build statistical models trained on the results of biophysical FEM simulations. The first statistical model estimates unloaded configurations based on either the intraventricular pressure or the end-diastolic fiber strain. The technique provides an alternative to the standard fixed-point iteration algorithm, which is more computationally expensive when used to unload more than 10 ventricles. The second statistical model captures the effects of varying infarct geometries on cardiac output. For training, we designed high resolution models of non-transmural infarcts including refinements of the border zone around the lesion. This study is a first effort in developing a platform combining HPC models and machine learning to investigate cardiac function in heart failure patients with the goal of assisting clinical diagnostics. |
| first_indexed | 2024-04-13T00:38:56Z |
| format | Article |
| id | doaj.art-698d3ba690fd43a58fc1f1611d130aed |
| institution | Directory Open Access Journal |
| issn | 1664-042X |
| language | English |
| last_indexed | 2024-04-13T00:38:56Z |
| publishDate | 2018-08-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Physiology |
| spelling | doaj.art-698d3ba690fd43a58fc1f1611d130aed2022-12-22T03:10:14ZengFrontiers Media S.A.Frontiers in Physiology1664-042X2018-08-01910.3389/fphys.2018.01002355607Gaussian Process Regressions for Inverse Problems and Parameter Searches in Models of Ventricular MechanicsPaolo Di Achille0Ahmed Harouni1Svyatoslav Khamzin2Svyatoslav Khamzin3Olga Solovyova4Olga Solovyova5John J. Rice6Viatcheslav Gurev7Healthcare and Life Sciences Research, IBM T.J. Watson Research Center, Yorktown Heights, NY, United StatesIBM Research Almaden, San Jose, CA, United StatesUral Federal University, Yekaterinburg, RussiaInstitute of Immunology and Physiology, Ural Branch of the Russian Academy of Sciences (UB RAS), Yekaterinburg, RussiaUral Federal University, Yekaterinburg, RussiaInstitute of Immunology and Physiology, Ural Branch of the Russian Academy of Sciences (UB RAS), Yekaterinburg, RussiaHealthcare and Life Sciences Research, IBM T.J. Watson Research Center, Yorktown Heights, NY, United StatesHealthcare and Life Sciences Research, IBM T.J. Watson Research Center, Yorktown Heights, NY, United StatesPatient specific models of ventricular mechanics require the optimization of their many parameters under the uncertainties associated with imaging of cardiac function. We present a strategy to reduce the complexity of parametric searches for 3-D FE models of left ventricular contraction. The study employs automatic image segmentation and analysis of an image database to gain geometric features for several classes of patients. Statistical distributions of geometric parameters are then used to design parametric studies investigating the effects of: (1) passive material properties during ventricular filling, and (2) infarct geometry on ventricular contraction in patients after a heart attack. Gaussian Process regression is used in both cases to build statistical models trained on the results of biophysical FEM simulations. The first statistical model estimates unloaded configurations based on either the intraventricular pressure or the end-diastolic fiber strain. The technique provides an alternative to the standard fixed-point iteration algorithm, which is more computationally expensive when used to unload more than 10 ventricles. The second statistical model captures the effects of varying infarct geometries on cardiac output. For training, we designed high resolution models of non-transmural infarcts including refinements of the border zone around the lesion. This study is a first effort in developing a platform combining HPC models and machine learning to investigate cardiac function in heart failure patients with the goal of assisting clinical diagnostics.https://www.frontiersin.org/article/10.3389/fphys.2018.01002/fullLV mechanicsFEMinfarct modelunloaded configurationkriginginverse optimization |
| spellingShingle | Paolo Di Achille Ahmed Harouni Svyatoslav Khamzin Svyatoslav Khamzin Olga Solovyova Olga Solovyova John J. Rice Viatcheslav Gurev Gaussian Process Regressions for Inverse Problems and Parameter Searches in Models of Ventricular Mechanics Frontiers in Physiology LV mechanics FEM infarct model unloaded configuration kriging inverse optimization |
| title | Gaussian Process Regressions for Inverse Problems and Parameter Searches in Models of Ventricular Mechanics |
| title_full | Gaussian Process Regressions for Inverse Problems and Parameter Searches in Models of Ventricular Mechanics |
| title_fullStr | Gaussian Process Regressions for Inverse Problems and Parameter Searches in Models of Ventricular Mechanics |
| title_full_unstemmed | Gaussian Process Regressions for Inverse Problems and Parameter Searches in Models of Ventricular Mechanics |
| title_short | Gaussian Process Regressions for Inverse Problems and Parameter Searches in Models of Ventricular Mechanics |
| title_sort | gaussian process regressions for inverse problems and parameter searches in models of ventricular mechanics |
| topic | LV mechanics FEM infarct model unloaded configuration kriging inverse optimization |
| url | https://www.frontiersin.org/article/10.3389/fphys.2018.01002/full |
| work_keys_str_mv | AT paolodiachille gaussianprocessregressionsforinverseproblemsandparametersearchesinmodelsofventricularmechanics AT ahmedharouni gaussianprocessregressionsforinverseproblemsandparametersearchesinmodelsofventricularmechanics AT svyatoslavkhamzin gaussianprocessregressionsforinverseproblemsandparametersearchesinmodelsofventricularmechanics AT svyatoslavkhamzin gaussianprocessregressionsforinverseproblemsandparametersearchesinmodelsofventricularmechanics AT olgasolovyova gaussianprocessregressionsforinverseproblemsandparametersearchesinmodelsofventricularmechanics AT olgasolovyova gaussianprocessregressionsforinverseproblemsandparametersearchesinmodelsofventricularmechanics AT johnjrice gaussianprocessregressionsforinverseproblemsandparametersearchesinmodelsofventricularmechanics AT viatcheslavgurev gaussianprocessregressionsforinverseproblemsandparametersearchesinmodelsofventricularmechanics |