1. Comparative Study of Myocardium Tissue Based on Gradient Features
Simona Moldovanu, Luminita Moraru, Bibicu Dorin
Physics Department, University of Galati, 47 Domneasca St., Galati, Romania
Echocardiography as imagistic tool is wide used in observing the clinical manifestations and features of the heart and it provides efficacious and lucrative information about the heart heath and function. Taking the advantage that is a cheaper and fast method, the echocardiography technique becomes a commonly and functionally ways to diagnose and describe the most heart problem. However, being a technique that uses ultrasonic wave propagation in human tissues its image quality is affected by noise. From this reason the ultrasound investigation of the myocardial function is a challenge because any visual assessment of heart is subjective and requires extensive training.
Figure 2 Ultrasound gray-scale images of the heart. From the left: apical four-chamber view
(4CH), apical two-chamber view (2CH), apical long-axis view (3CH) and a short axis view at the basal level of the heart.
Figure 1 Definition of imaging planes of the heart. Segmental analysis of left ventricular walls based on schematic views. The upper figures show the apical long-axis views: four-chamber view (4CH), two-chamber view (2CH) and apical long-axis view (3CH). The lower figures illustrate the short-axis view of the heart at basal, mid and apical level. Reproduced with permission R. M. Lang at al., 2005
Abstract- The analysis of echocardiography texture images related to muscle fibre orientation is a modality to characterize the myocardial pathology. In this article, our goal is to explain the potential clinical applicability of the proposed method in processing echocardiography images. The difference of gradient texture features between apical two-chamber (A2C) and parasternal short axis (SAX) views in basal inferior zone was investigated. Additionally the standard deviation of the magnitude and the direction gradient of the edges across the end-systole and end-diastole frames were computed. Myocardium tissue was estimated analysing 104 ROIs (regions of interest) not less than pixel. The ROIs were automatically cropped from each echocardiography of 26 healthy patients in basal inferior zone. The same tissue has been viewed from different angles and in order to determine the standard error of the mean between classes A2C and SAX the t-test method was applied. A test system for an automatic comparison of data was implemented.
Subclasses
Standard error of the mean between subclasses
A2C / SAX gradient magnitude
0.022
SAX /A2C gradient magnitude
0.021
A2C / SAX gradient direction
0.135
SAX / A2C gradient direction
0.147
Experimental Tools
Experiment environment was Intel (R) Core (TM) 2 Duo CPU T5900, 2.20 with 3G RAM
104 ROIs samples was used in our work
t-test statistical approach
A2C / SAX gradient magnitude
SAX /A2C gradient magnitude
A2C / SAX gradient direction
SAX / A2C gradient direction
The clinical study involves 26 healthy patients with median age of 58 years
The US images used in this analysis are acquired from scanning systems, VIVID E9 and GE HORTEN MOK WAY using curvilinear probe with transducer frequency of 3.5 MHz.
Fig. 3. The block diagram of the test system. Legend: ROIs-A2C are the regions of interest from healthy myocardium of the apical two-chamber class, ROIs-SAX are the regions of interest from healthy myocardium of the parasternal short axis class
Conclusion
This paper deals with the echocardiography texture images analysis in order to optimize and facilitate communication between information contained into cardiac images and clinical applications. To study the difference between two classes of ROIs which depicted the same tissue viewed from two different angles the t-test is carried out. In accordance with the standard error of the mean values calculated using both the standard deviation magnitude of the gradient and of the direction for A2C and SAX views in basal inferior zone classes and with the data presented in Table 1 we concluded that the standard deviation magnitude of the gradient and the direction for the studied classes do not overlap and t-test analysis provides the standard error of the mean values between features of the classes. The source of these errors is the different orientation edges in the classes A2C and SAX and noise from echocardiography images. Usually, the t-test systems find the errors between the groups belonging to the same structure.
Fig. 4. (a)Template layout for SAX view (b) Template layout for A2C view.
Fig. 5. (a) A scatter plot of the SAX gradient magnitude and A2C gradient magnitude; (b) A scatter plot of the SAX gradient direction and A2C gradient direction.
The authors would like to thank the Project SOP HRD-EFICIENT 61445/2009 and the Project SOP HRD -TOP ACADEMIC - 107/1.5/S id of University Dunarea de Jos of Galati, Romania.