Classical and Learned MR to Pseudo-CT Mappings for Accurate Transcranial Ultrasound Simulation

Abstract

Model-based treatment planning for transcranial ultrasound therapy typically involves mapping the acoustic properties of the skull from an X-ray computed tomography (CT) image of the head. Here, three methods for generating pseudo-CT (pCT) images from magnetic resonance (MR) images were compared as an alternative to CT. A convolutional neural network (U-Net) was trained on paired MR-CT images to generate pCT T images from either T1-weighted or zero-echo time (ZTE) MR images (denoted tCT and zCT, respectively). A direct mapping from ZTE to pCT was also implemented (denoted cCT). When comparing the pCT and ground-truth CT images for the test set, the mean absolute error was 133, 83, and 145 Hounsfield units (HU) across the whole head, and 398, 222, and 336 HU within the skull for the tCT, zCT, and cCT images, respectively. Ultrasound simulations were also performed using the generated pCT images and compared to simulations based on CT. An annular array transducer was used targeting the visual or motor cortex. The mean differences in the simulated focal pressure, focal position, and focal volume were 9.9%, 1.5 mm, and 15.1% for simulations based on the tCT images; 5.7%, 0.6 mm, and 5.7% for the zCT; and 6.7%, 0.9 mm, and 12.1% for the cCT. The improved results for images mapped from ZTE highlight the advantage of using imaging sequences, which improves the contrast of the skull bone. Overall, these results demonstrate that acoustic simulations based on MR images can give comparable accuracy to those based on CT.

Figure 1. The input to the neural network is one or multiple adjacent transverse MR slices, acquired either using ZTE or T1w sequences.

The output of the neural network is multiplied by a head mask and compared to the registered CT inside the mask.

Figure 2. Density plot showing the correlation between the pCT and ground truth CT images for the test set.

The white lines show y = x.

Figure 3. Example of pCT and error maps for a single subject in the test set.

The pseudo-CTs are registered to the ground truth CT before comparison.

Figure 4. Middle sagittal slice for the learned pseudo-CT images for the 14 test subjects.

The mean absolute error (MAE) value reported is calculated for the volume inside the head mask. A small number of the learned images mapped from T1w MR images have discontinuous skull boundaries.

Figure 5

Differences in the focal pressure, focal position, and focal volume for acoustic simulations using different pseudo-CT images against simulations using a ground truth CT. Results are divided into two sets, targeting the visual cortex (V1) and motor cortex (M1).

Figure 6

Sound speed maps and simulated acoustic field for a subject from the test set for targets in the (a) visual cortex (V1) and (b) motor cortex (M1). The difference plots show the difference against the ground truth generated using the real CT images. The V1 and M1 results show sagittal and coronal slices through the spatial peak pressure, respectively.