Pointwise encoding time reduction with radial acquisition (PETRA) to pseudo-CT (computer tomography) mapping for...

Ultrasound simulations used for treatment planning in transcranial ultrasound stimulation (TUS) currently depend on converting computed tomography (CT) images to acoustic properties. Aiming to reduce subject exposure to ionising radiation, recent literature focused on generating pseudo-CT images from different magnetic resonance (MR) imaging sequences. Here, the pointwise encoding time reduction with radial acquisition (PETRA) sequence was used because of its stronger ability to image bones. A dataset of paired CT and PETRA scans was acquired separately for human subjects and ex vivo skulls. Principal component analysis performed on all bone voxels uncovered the affine relationship converting PETRA images to pseudo-CTs. Acoustic simulations based on CTs and pseudo-CTs were performed in k-Plan and k-wave. For in vivo data, the mean absolute errors in focal position, peak pressure, and focal volume were 0.48±0.25 mm, 5.1±4.0% and 5.8±4.4%. The similarity of acoustic fields from ex vivo CTs/pseudo-CTs was compared with experimental data. The corresponding errors were 1.5 ± 1.2 mm, 17.1 ± 14.6% and 32.5 ± 22.5% for CT and 1.6 ± 1.4 mm, 18.2 ± 17.1% and 19.2 ± 15.9% for pseudo-CT. The small errors for in vivo data and the similarity of the experimental errors for both methods validate the use of PETRA-derived pseudo-CT as an alternative to CT, ensuring precise acoustic field predictions. The conversion tool is available at https://github.com/ucl-bug/petra-to-ct.