Limits...
Quantification of myocardial perfusion with self-gated cardiovascular magnetic resonance.

Likhite D, Adluru G, Hu N, McGann C, DiBella E - J Cardiovasc Magn Reson (2015)

Bottom Line: The gated and the self-gated datasets were then quantified with standard methods.Regional myocardial blood flow estimates (MBFs) obtained using self-gated systole (0.64 ± 0.26 ml/min/g), self-gated diastole (0.64 ± 0.26 ml/min/g), and ECG-gated scans (0.65 ± 0.28 ml/min/g) were similar.Based on the criteria for interchangeable methods listed in the statistical analysis section, the MBF values estimated from self-gated and gated methods were not significantly different.

View Article: PubMed Central - PubMed

ABSTRACT

Background: Current myocardial perfusion measurements make use of an ECG-gated pulse sequence to track the uptake and washout of a gadolinium-based contrast agent. The use of a gated acquisition is a problem in situations with a poor ECG signal. Recently, an ungated perfusion acquisition was proposed but it is not known how accurately quantitative perfusion estimates can be made from such datasets that are acquired without any triggering signal.

Methods: An undersampled saturation recovery radial turboFLASH pulse sequence was used in 7 subjects to acquire dynamic contrast-enhanced images during free-breathing. A single saturation pulse was followed by acquisition of 4-5 slices after a delay of ~40 msec. This was repeated without pause and without any type of gating. The same pulse sequence, with ECG-gating, was used to acquire gated data as a ground truth. An iterative spatio-temporal constrained reconstruction was used to reconstruct the undersampled images. After reconstruction, the ungated images were retrospectively binned ("self-gated") into two cardiac phases using a region of interest based technique and deformably registered into near-systole and near-diastole. The gated and the self-gated datasets were then quantified with standard methods.

Results: Regional myocardial blood flow estimates (MBFs) obtained using self-gated systole (0.64 ± 0.26 ml/min/g), self-gated diastole (0.64 ± 0.26 ml/min/g), and ECG-gated scans (0.65 ± 0.28 ml/min/g) were similar. Based on the criteria for interchangeable methods listed in the statistical analysis section, the MBF values estimated from self-gated and gated methods were not significantly different.

Conclusion: The self-gated technique for quantification of regional myocardial perfusion matched ECG-gated perfusion measurements well in normal subjects at rest. Self-gated systolic perfusion values matched ECG-gated perfusion values better than did diastolic values.

No MeSH data available.


Related in: MedlinePlus

A line profile through a self-gated systolic slice with and without deformable registration. The suppression of the residual cardiac motion by use of deformable registration is visible.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC4325943&req=5

Fig4: A line profile through a self-gated systolic slice with and without deformable registration. The suppression of the residual cardiac motion by use of deformable registration is visible.

Mentions: After binning the datasets into near-systole and near-diastole, some cardiac motion was still present. Deformable registration was used to reduce the cardiac motion. This registration was complicated by the changes in contrast uptake, such that registering all of the images to a single time frame generally worked poorly. Thus a model-based deformable registration technique [25] was used instead of registering to a single frame or to neighboring time frames. The model-based deformable registration involved two steps: generation of model images, and registration of the self-gated images to these model images. The generation of model images was done by using a compartment model, as described in [26]. The model did not support motion or deformation, so these model images were “still” and acted as reference images - the self-gated images with cardiac motion were then registered to them individually at each time frame. In this way the contrast was similar between the source and target images that were being registered. The registration for each image used a symmetric image normalization method that maximized the cross-correlation within the space of diffeomorphic maps [27]. The software used for registration was the ‘Advanced Normalization Tools’ [23,28]. The sets of parameters for the deformable registration were tuned manually by testing different parameters on a few randomly chosen slices. The set of parameters selected were: Step size for transformation model = 0.25, sigma(deformation field) = 2, sigma(similarity field) = 10. For speed, the ANTs made use of a 3 level image pyramid with a maximum of 100 iterations at the coarsest resolution, 100 iterations at the next coarsest and 10 iterations at full resolution [28]. Figure 4 shows a line profile for a slice from a dataset before and after model-based deformable registration.Figure 5


Quantification of myocardial perfusion with self-gated cardiovascular magnetic resonance.

Likhite D, Adluru G, Hu N, McGann C, DiBella E - J Cardiovasc Magn Reson (2015)

A line profile through a self-gated systolic slice with and without deformable registration. The suppression of the residual cardiac motion by use of deformable registration is visible.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC4325943&req=5

Fig4: A line profile through a self-gated systolic slice with and without deformable registration. The suppression of the residual cardiac motion by use of deformable registration is visible.
Mentions: After binning the datasets into near-systole and near-diastole, some cardiac motion was still present. Deformable registration was used to reduce the cardiac motion. This registration was complicated by the changes in contrast uptake, such that registering all of the images to a single time frame generally worked poorly. Thus a model-based deformable registration technique [25] was used instead of registering to a single frame or to neighboring time frames. The model-based deformable registration involved two steps: generation of model images, and registration of the self-gated images to these model images. The generation of model images was done by using a compartment model, as described in [26]. The model did not support motion or deformation, so these model images were “still” and acted as reference images - the self-gated images with cardiac motion were then registered to them individually at each time frame. In this way the contrast was similar between the source and target images that were being registered. The registration for each image used a symmetric image normalization method that maximized the cross-correlation within the space of diffeomorphic maps [27]. The software used for registration was the ‘Advanced Normalization Tools’ [23,28]. The sets of parameters for the deformable registration were tuned manually by testing different parameters on a few randomly chosen slices. The set of parameters selected were: Step size for transformation model = 0.25, sigma(deformation field) = 2, sigma(similarity field) = 10. For speed, the ANTs made use of a 3 level image pyramid with a maximum of 100 iterations at the coarsest resolution, 100 iterations at the next coarsest and 10 iterations at full resolution [28]. Figure 4 shows a line profile for a slice from a dataset before and after model-based deformable registration.Figure 5

Bottom Line: The gated and the self-gated datasets were then quantified with standard methods.Regional myocardial blood flow estimates (MBFs) obtained using self-gated systole (0.64 ± 0.26 ml/min/g), self-gated diastole (0.64 ± 0.26 ml/min/g), and ECG-gated scans (0.65 ± 0.28 ml/min/g) were similar.Based on the criteria for interchangeable methods listed in the statistical analysis section, the MBF values estimated from self-gated and gated methods were not significantly different.

View Article: PubMed Central - PubMed

ABSTRACT

Background: Current myocardial perfusion measurements make use of an ECG-gated pulse sequence to track the uptake and washout of a gadolinium-based contrast agent. The use of a gated acquisition is a problem in situations with a poor ECG signal. Recently, an ungated perfusion acquisition was proposed but it is not known how accurately quantitative perfusion estimates can be made from such datasets that are acquired without any triggering signal.

Methods: An undersampled saturation recovery radial turboFLASH pulse sequence was used in 7 subjects to acquire dynamic contrast-enhanced images during free-breathing. A single saturation pulse was followed by acquisition of 4-5 slices after a delay of ~40 msec. This was repeated without pause and without any type of gating. The same pulse sequence, with ECG-gating, was used to acquire gated data as a ground truth. An iterative spatio-temporal constrained reconstruction was used to reconstruct the undersampled images. After reconstruction, the ungated images were retrospectively binned ("self-gated") into two cardiac phases using a region of interest based technique and deformably registered into near-systole and near-diastole. The gated and the self-gated datasets were then quantified with standard methods.

Results: Regional myocardial blood flow estimates (MBFs) obtained using self-gated systole (0.64 ± 0.26 ml/min/g), self-gated diastole (0.64 ± 0.26 ml/min/g), and ECG-gated scans (0.65 ± 0.28 ml/min/g) were similar. Based on the criteria for interchangeable methods listed in the statistical analysis section, the MBF values estimated from self-gated and gated methods were not significantly different.

Conclusion: The self-gated technique for quantification of regional myocardial perfusion matched ECG-gated perfusion measurements well in normal subjects at rest. Self-gated systolic perfusion values matched ECG-gated perfusion values better than did diastolic values.

No MeSH data available.


Related in: MedlinePlus