To begin, acquire ultrashort echo time MRI images of the lung during free breathing. Import the data and k-spaced trajectories into MATLAB. Discard the first 1000 projections to ensure that the data reaches a steady state magnetization.
Next, perform the image reconstruction using a non-uniform fast Fourier transform to a matrix size of 96 by 96 by 96. Use approximately 200 projections corresponding to 0.6 to 0.8 seconds worth of data. Then reconstruct and store images from all coil elements as well as the final coil combined image.
In the coil combined image, select a coronal slice that clearly shows the diaphragm. Once the coronal slice is selected, view the individual coil images for this slice and select one or two coil elements that best show the diaphragm. Now reconstruct only the data from the coil elements using a sliding window to generate images with approximately 0.5 second temporal resolution.
Use the first 200 projections to reconstruct an image using a non-uniform fast Fourier transform and store only the diaphragm slice. Shift by 100 projections and reconstruct an additional image storing the diaphragm slice. Now, select a line over the diaphragm in the first of the sliding window images.
Visualize respiratory motion by viewing this respiratory navigator for all projections. Determine the location of the diaphragm for all respiratory navigators and use this location to label projections as belonging to a given respiratory bin. Then identify the bin with the highest number of projections corresponding to the end expiration and choose it for reconstruction.
Use an exponential filter to provide a weight of one to projections within the primary bin and a sharply reducing weight to projections within different respiratory bins. Next, use the Berkeley advanced reconstruction toolbox to reconstruct a high resolution image at the desired respiratory bin. Calculate density compensation weights using iterative density combination.
Scale the density compensation weights by the soft-gating weights. Then scale data based on the density compensation and soft-gating weights. Now perform a basic non-uniform fast Fourier transform to facilitate coil combination.
Convert the non-uniform fast Fourier transform image into gridded k-space for coil combination. Then generate a coil combination matrix and use it to combine coils for both the raw data and the gridded k-space and estimate coil sensitivities. Afterward using the weighted density compensation, coil combined data and coil sensitivity maps, perform parallel imaging compressed sense reconstruction.
Images generated at end expiration using both image-based and k-space based gating showed clear visualization of the diaphragm with image-based gating demonstrating superior motion compensation. Soft-gating enhanced the sharpness of the inspiration images reducing under sampling artifacts compared to hard-gating. Both image-based and k-space-based gating successfully detected respiratory waveforms during regular breathing, with image-based gating yielding clearer results under irregular breathing conditions.
