Automatic planning of MR-guided transcranial focused ultrasound treatment for essential tremor.

Introduction: Transcranial focused ultrasound therapy (tcFUS) offers precise thermal ablation for treating Parkinson's disease and essential tremor. However, the manual fine-tuning of fiber tracking and segmentation required for accurate treatment planning is time-consuming and demands expert knowledge of complex neuroimaging tools. This raises the question of whether a fully automated pipeline is feasible or if manual intervention remains necessary. Methods: We investigate the dependence on fiber tractography algorithms, segmentation approaches, and degrees of automation, specifically for essential tremor therapy planning. For that purpose, we compare an automatic pipeline with a manual approach that requires the manual definition of the target point and is based on FMRIB software library (FSL) and other open-source tools. Results: Our findings demonstrate the high feasibility of automatic fiber tracking and the automated determination of standard treatment coordinates. Employing an automatic fiber tracking approach and deep learning (DL)-supported standard coordinate calculation, we achieve anatomically meaningful results comparable to a manually performed FSL-based pipeline. Individual cases may still exhibit variations, often stemming from differences in region of interest (ROI) segmentation. Notably, the DL-based approach outperforms registration-based methods in producing accurate segmentations. Precise ROI segmentation proves crucial, surpassing the importance of fine-tuning parameters or selecting algorithms. Correct thalamus and red nucleus segmentation play vital roles in ensuring accurate pathway computation. Conclusion: This study highlights the potential for automation in fiber tracking algorithms for tcFUS therapy, but acknowledges the ongoing need for expert verification and integration of anatomical expertise in treatment planning.

Deep Learning for Ultrasound Image Formation: CUBDL Evaluation Framework and Open Datasets.

Deep learning for ultrasound image formation is rapidly garnering research support and attention, quickly rising as the latest frontier in ultrasound image formation, with much promise to balance both image quality and display speed. Despite this promise, one challenge with identifying optimal solutions is the absence of unified evaluation methods and datasets that are not specific to a single research group. This article introduces the largest known international database of ultrasound channel data and describes the associated evaluation methods that were initially developed for the challenge on ultrasound beamforming with deep learning (CUBDL), which was offered as a component of the 2020 IEEE International Ultrasonics Symposium. We summarize the challenge results and present qualitative and quantitative assessments using both the initially closed CUBDL evaluation test dataset (which was crowd-sourced from multiple groups around the world) and additional in vivo breast ultrasound data contributed after the challenge was completed. As an example quantitative assessment, single plane wave images from the CUBDL Task 1 dataset produced a mean generalized contrast-to-noise ratio (gCNR) of 0.67 and a mean lateral resolution of 0.42 mm when formed with delay-and-sum beamforming, compared with a mean gCNR as high as 0.81 and a mean lateral resolution as low as 0.32 mm when formed with networks submitted by the challenge winners. We also describe contributed CUBDL data that may be used for training of future networks. The compiled database includes a total of 576 image acquisition sequences. We additionally introduce a neural-network-based global sound speed estimator implementation that was necessary to fairly evaluate the results obtained with this international database. The integration of CUBDL evaluation methods, evaluation code, network weights from the challenge winners, and all datasets described herein are publicly available (visit https://cubdl.jhu.edu for details).

Deep learning-based reconstruction of ultrasound images from raw channel data.

PURPOSE: We investigate the feasibility of reconstructing ultrasound images directly from raw channel data using a deep learning network. Starting from the raw data, we present the network the full measurement information, allowing for a more generic reconstruction to form, as compared to common reconstructions constrained by physical models using fixed speed of sound assumptions. METHODS: We propose a U-Net-like architecture for the given task. Additional layers with strided convolutions downsample the raw data. Hyperparameter optimization was used to find a suitable learning rate. We train and test our deep learning approach on plane wave ultrasound images with a single insonification angle. The dataset includes phantom as well as in vivo data. RESULTS: The images produced by our method are visually comparable to ones reconstructed with the conventional delay and sum algorithm. Deviations between prediction and ground truth are likely to be related to speckle noise. For the test set, the mean absolute error is [Formula: see text] for the phantom images and [Formula: see text] for the in vivo data. CONCLUSION: The result shows the feasibility of our approach and opens up new research directions regarding information retrieval from raw channel data. As the networks reconstruction performance is limited by the quality of the ground truth images, using other ultrasound reconstruction technique or image types as target data would be of interest.

A focused ultrasound treatment system for moving targets (part I): generic system design and in-silico first-stage evaluation.

BACKGROUND: Focused ultrasound (FUS) is entering clinical routine as a treatment option. Currently, no clinically available FUS treatment system features automated respiratory motion compensation. The required quality standards make developing such a system challenging. METHODS: A novel FUS treatment system with motion compensation is described, developed with the goal of clinical use. The system comprises a clinically available MR device and FUS transducer system. The controller is very generic and could use any suitable MR or FUS device. MR image sequences (echo planar imaging) are acquired for both motion observation and thermometry. Based on anatomical feature tracking, motion predictions are estimated to compensate for processing delays. FUS control parameters are computed repeatedly and sent to the hardware to steer the focus to the (estimated) target position. All involved calculations produce individually known errors, yet their impact on therapy outcome is unclear. This is solved by defining an intuitive quality measure that compares the achieved temperature to the static scenario, resulting in an overall efficiency with respect to temperature rise. To allow for extensive testing of the system over wide ranges of parameters and algorithmic choices, we replace the actual MR and FUS devices by a virtual system. It emulates the hardware and, using numerical simulations of FUS during motion, predicts the local temperature rise in the tissue resulting from the controls it receives. RESULTS: With a clinically available monitoring image rate of 6.67 Hz and 20 FUS control updates per second, normal respiratory motion is estimated to be compensable with an estimated efficiency of 80%. This reduces to about 70% for motion scaled by 1.5. Extensive testing (6347 simulated sonications) over wide ranges of parameters shows that the main source of error is the temporal motion prediction. A history-based motion prediction method performs better than a simple linear extrapolator. CONCLUSIONS: The estimated efficiency of the new treatment system is already suited for clinical applications. The simulation-based in-silico testing as a first-stage validation reduces the efforts of real-world testing. Due to the extensible modular design, the described approach might lead to faster translations from research to clinical practice.

Characterisation and reduction of the EEG artefact caused by the helium cooling pump in the MR environment: validation in epilepsy patient data.

The EEG acquired simultaneously with fMRI is distorted by a number of artefacts related to the presence of strong magnetic fields, which must be reduced in order to allow for a useful interpretation and quantification of the EEG data. For the two most prominent artefacts, associated with magnetic field gradient switching and the heart beat, reduction methods have been developed and applied successfully. However, a number of artefacts related to the MR-environment can be found to distort the EEG data acquired even without ongoing fMRI acquisition. In this paper, we investigate the most prominent of those artefacts, caused by the Helium cooling pump, and propose a method for its reduction and respective validation in data collected from epilepsy patients. Since the Helium cooling pump artefact was found to be repetitive, an average template subtraction method was developed for its reduction with appropriate adjustments for minimizing the degradation of the physiological part of the signal. The new methodology was validated in a group of 15 EEG-fMRI datasets collected from six consecutive epilepsy patients, where it successfully reduced the amplitude of the artefact spectral peaks by 95 ± 2 % while the background spectral amplitude within those peaks was reduced by only -5 ± 4 %. Although the Helium cooling pump should ideally be switched off during simultaneous EEG-fMRI acquisitions, we have shown here that in cases where this is not possible the associated artefact can be effectively reduced in post processing.

Reduction of EEG artefacts induced by vibration in the MR-environment.

The EEG acquired simultaneously with functional magnetic resonance imaging (fMRI) is distorted by a number of artefacts related to the presence of strong magnetic fields. In order to allow for a useful interpretation of the EEG data, it is necessary to reduce these artefacts. For the two most prominent artefacts, associated with magnetic field gradient switching and the heart beat, reduction methods have been developed and applied successfully. Due to their repetitive nature, such artefacts can be reduced by subtraction of the respective template retrieved by averaging across cycles. In this paper, we investigate additional artefacts related to the MR environment and propose a method for the reduction of the vibration artefact caused by the cryo-cooler compression pumps system. Data were collected from the EEG cap placed on an MR head phantom, in order to characterise the MR environment related artefacts. Since the vibration artefact was found to be repetitive, a template subtraction method was developed for its reduction, and this was then adjusted to meet the specific requirements of patient data. The developed methodology successfully reduced the vibration artefact by about 90% in five EEG-fMRI datasets collected from two epilepsy patients.