ESTRO 2024 - Abstract Book

S3969

Physics - Image acquisition and processing

ESTRO 2024

In order to demonstrate the impact of view angle tilting in case of moderate B0 inhomogeneities two sets of experiments were performed. A commercially available Philips head phantom was scanned on a 1.5T MR system (Ingenia Ambition, Philips Medical System Netherlands). Two sets of 2D TSE images were acquired (FOV 230x230x42 mm, acquisition voxel size 1x1x3mm, readout bandwidth 869.0 Hz/mm, scan time 1min27s with and without VAT) with opposing readout directions. The same acquisitions were repeated with view angle tilting added. Artificial B0 variations were simulated by adding a B0 gradient of 0.2mT/m along the AP direction. A 3D Dixon based B0 map was acquired (FOV 230x230x42 mm, acquisition voxel size 2x2x3mm) to characterize the B0 inhomogeneities. For both sets of images, a difference image was created between the opposing readout directions to visualize the effect of residual geometric distortion. A similar experiment was repeated on a healthy volunteer using a 3T MR system (MR7700, Philips Medical System Netherlands). Images of the brain were acquired using two sets of 2D TSE images (FOV 232x208x72mm, acquisition voxel size 1x1x3mm, readout bandwidth 434.5 Hz/mm, scan time 2min 15s with and without VAT) with opposing readout directions. The same acquisitions were repeated with view angle tilting added. No artificial B0 variations were added. A 3D Dixon based B0 map was acquired (FOV 232x208x72 mm, acquisition voxel size 2x2x3mm) to characterize the B0 inhomogeneities. For both sets of images, a difference image was created between the opposing readout directions to visualize the effect of residual geometric distortion. Based on the B0 map and the known readout bandwidth, a distortion map was calculated.

Results:

Figure 1 shows the comparison between images acquired with and without VAT using the standard Philips head phantom. The difference image clearly shows that VAT is able to drastically reduce the induced geometric distortion due to the artificially added B0 gradient compared to the acquisitions without VAT added without a visible impact on image quality or scan time.

Figure 1: Comparison of phantom images without VAT (top row) and with VAT (bottom row). First column: Images acquired with readout direction pointing in the posterior direction, Second column: Images acquired with readout direction pointing in the anterior direction, third column: Difference image between readout in posterior and anterior direction, fourth column: distortion map based on an acquired B0 map. The in-vivo brain example in Figure 2 demonstrates that VAT is equally effective in the presence of common B0 inhomogeneities encountered in the human body, ranging from -1.5mm to +1mm. Specifically close to the sinuses and inner ear, the difference in geometric distortion between the cases with and without VAT can be appreciated.