Magnetic Resonance Derived Myocardial Strain Assessment Using Feature Tracking

The concept of strain is the deformation of an object normalized to its original shape, which is fundamentally important in the assessment of ventricular function. This video demonstrates a novel feature tracking technique for the assessment of circumferential strain using standard cine sequences. First, a standard cardiac MRI short access sequence is acquired.

The cardiac MRI Short access Sydney sequence is then downloaded into the feature tracking software, and a data package is composed for strain analysis. After tracing an initial endocardial border at the end diastolic phase, the feature tracking software will automatically track the displacement of individual patterns representing anatomical structures throughout the entire cardiac cycle. Finally, the tracing is analyzed for circumferential strain assessment.

Absolute, as well as regional values of velocities, displacement, and strain can be derived. The main advantage of feature tracking technique over existing methods is that circumferential strain analysis can be performed without additional imaging and is simple to use and correlates well with currently available tag imaging software. Once master, this technique can be done in a matter of Minutes, though this analysis method is presented here for single short access slices.

At a mid-level, it has the potential to do a complete assessment of the ventricle. This is accomplished by using also levels at the base and the apex, and in addition, long axis slices to measure the longitudinal deformation. The first step in the procedure is to acquire a standard cardiac MRI short access sequence using a standard MRI machine with cardiac software.

This is part of a standard protocol and no additional time consuming tagging procedure is needed. The standard short access syn sequence is stored offline and the images can be imported into the feature tracking software for circumferential strain analysis. Now let's see how to import and select data.

The standard short access cini sequence is imported into the patient database of the feature tracking software from an offline workstation or network drive. Once the short access cini sequence is available in the patient database, a data packages composed using a package composer for strain analysis. The package composer allows selection of a specific short axis slice from the short axis sequence or stack from the patient database.

This will be saved and stored in the patient database for strain analysis. In this case, the short axis slice at the mid ventricle at the level of the papillary muscle is chosen for circumferential strain analysis. Now let's see how to analyze the data.

The selected short axis slice is imported into the feature tracking analysis software, and the image contrast can be adjusted as desired. Start the analysis by initially tracing the endocardial border at the end diastolic phase when the chambers at its largest dimension tracing can be corrected or modified if needed. Before tracking is started to ensure that the initial trace is within the cavity tissue interface with a push of a button start the fully automated tracking procedure.

The individual points composing the first trace are followed in time by searching the same features that are about one point in its neighborhood in the following frames. The tracked features can be the cavity tissue boundary or anatomical elements that are different along the tissue. They're found by methods of maximum likelihood in two regions of interest between two frames.

Feature tracking analysis of the short axis results in data for velocity displacement and strain in both the radial and circumferential directions. Currently, only circumferential strain has been validated and correlates highly with currently available software for tagged imaging analysis. The short access circumferential strain can be presented segmentally based on the American Society of Echocardiography 16 or 17 segment model, or as an average of all the segments representing global mid left ventricular circumferential strain.

The circumferential strain results can then be exported into a text file for further analysis. This feature tracking technique correlates well with harmonic phase assessment of tagged images in a large Duchenne muscular dystrophy patient population with a wide range of cardiac dysfunction. Now, let's see, validation of the technique with a phantom.

A series of artificial computer generated loops have been prepared to allow testing of the image analysis procedure in simple and perfectly controlled conditions for this purpose. A phantom in a short axis projection of an ideal left ventricle is prepared. The tracking method was tested with phantoms varying in spatial resolution frame rate and epicardial type of motion.

Also, the number of points used for the initial tracing were varied and the effect tested in all cases. Errors were very small for the integral quantities, radius and strain, and slightly larger for the differential quantities, velocity and strain rate. This was expected because the derivative operator amplifies errors.

The quality of the results degrades with reduced spatial resolution becoming very low. However, time resolution does not significantly affect results as long as a minimum of about 25 frames per second is kept very high. Frame rates do not improve the results because frame by frame displacement become lower than the pixel size.

Therefore, an increase in the frame rate is of little or no utility when it is not accompanied by an increase in the spatial resolution. Now we'll show you some representative clinical cases and results the strain difference between a control subject young DMD patient with normal ejection fraction older DMD patients with normal ejection fraction, A DMD patient with abnormal ejection fraction and A DMD patient with end stage heart failure is shown. Note the global circumferential strain difference between the control subjects as well as the DMD patients at different stages of disease.

Feature tracking of left ventricular circumferential strain not only correlates highly with assessment of strain using tagged images, it was also able to define a similar ST strata as harmonic phase analysis of tagged image sequence. Circumen strain analysis by feature tracking method may help answer key questions in the field of heart failure by detecting cardiac dysfunction prior to declining in ejection fraction in both acquire and congenital disease. This may allow early intervention and assessment of therapeutic efficacy.

A problem in cardiac quantification has always been that different imaging modalities use different measurement techniques and the results are difficult to compare. The technology you have seen here is not only applicable to cardiac MRI, but also to other imaging sources like ultrasound and is therefore a great step towards comparability across modalities.