MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Building 3D Dynamic Keyhole Library with Compressive Sensing and Parallel Imaging Reconstruction

D Lam*, B Lewis, H Gach, S Mutic, T Kim, Washington University in St. Louis, St. Louis, MO

Presentations

(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose:
Motion-free high-spatial resolution library is a key component for real-time 3D dynamic keyhole in MRgRT. The aim of this study is to develop a procedure for a motion-free high-spatial resolution library using a compressed-sensing approach in image reconstruction and processing for real-time 3D MRgRT.

Methods:
In 0.35T MRgRT systems, 3D MR images were acquired using 3D TrueFISP (FOV: 350 x 350 mm, matrix: 150 x 150 x 8, voxel: 2 x 2 x 2 mm3, acquisition time: 5 s) from a motion phantom and a human subject. To reduce motion during imaging, the phantom was posed at the designated position and the human subject was in breath-hold using respiratory visual guidance system. The keyhole library is reconstructed using compressed sensing with parallel imaging. Specifically, coil sensitivity is estimated by the ESPIRiT calibration method, followed by compressed-sensing reconstruction from the input k-space and coil sensitivity with total variation constraint to achieve reduced noise, high contrast images. In the study, under-sampled data were used to improve the temporal resolution of the library. The performance of the method was compared with the Fourier-transformed images

Results:
The proposed method can generate a high spatial, high resolution 3D reconstruction from a low spatial, high temporal resolution 3D MRI dataset in the phantom. SNR, CNR of the reconstructed images were =8.1 dB and =16.2 dB, compared to FFT reconstructed images. SSMI was 96% for the 70% data with the reference to 100% of the acquired data. In the human dataset, the templates in 3 different planes reconstructed by the presented method show much less noise and more contrast then the zero filled FFT method.

Conclusion:
3D dynamic keyhole with compressed sensing total variation reconstruction could obtain better image quality in less acquisition time.

Keywords

MRI, K-space

Taxonomy

IM- MRI : Image Reconstruction

Contact Email