Ph.D Thesis

Ph.D StudentBachner Noa
SubjectThe Regional Function of the Left Ventricle as Assessed
by Echocardiography in Health and Acute
Myocardial Infarction
DepartmentDepartment of Biomedical Engineering
Full Thesis textFull thesis text - English Version


Diagnosis of left ventricular (LV) function after myocardial infarction (MI) is performed by eyeballing the echocardiography cines, evaluating the ejection fraction and scoring the regional wall motion of the myocardial segments. This estimation is subjective and relies on the Echocardiographer’s experience. Hence, automatic tool for wall motion estimation, based on 2 dimensional speckle tracking echocardiography (STE), was developed.

In this study, the STE commercial program was enhanced by applying a 3 dimensional wavelet de-noising to the myocardial velocities, instead of the built-in smoothing process. Following validation with software-implemented and mechanical phantoms versus imposed values, short-axis cines were obtained from 50 rats, comparing the results of the commercial and enhanced layer-specific (LS) programs. After validation, LS strain parameters of normal rats (n=110) were compared to those of humans (n=120). Subsequently, 13 rats underwent left anterior descending artery occlusion for 30 minutes followed by reperfusion. Short-axis scans at apical and papillary muscles levels were obtained at baseline, 10 minutes and 24 hours after reperfusion. The scans were post-processed by the LS program. Thereafter, animals were sacrificed, and the MI size was determined by Triphenyltetrazolium Chloride method.

The accuracy of rotation measurements using software-implemented phantom yielded errors of 7.5%, 2.9% and 3.4%, for inner, middle and outer layers, respectively. Analysis of a shrinking/expanding mechanical phantom yielded strain errors of 3%, 5% and 7% for the three layers. Bland-Altman analysis, performed on the rat results (n=50), showed agreement between the commercial and enhanced programs. The comparative study of LV function of rats versus humans showed that the longitudinal strain at all layers is equal in rats and humans, while the circumferential strain is larger in humans (P<0.001). The longitudinal, circumferential and radial segmental strains were different for rats and humans (P<0.05). Additionally, while LV twist is equal in rats and humans, in rats the apical rotation is larger (P<0.01) and basal rotation is smaller (P<0.001). Following MI in the rat model, the largest change versus baseline occurred in peak systolic circumferential strain (transmural: sensitivity 100%, specificity 96%, non-transmural: sensitivity 88%, specificity 84% at the apex).

LS strain analysis, based on STE measurements, is feasible in the rat model. Moreover, it is possible to detect either transmural or non-transmural MI in this model. When evaluating LV function of rats, it is recommended to measure the global longitudinal strain and LV twist, since no scaling is required when converting these parameters and inferring them to humans.