End-to-end semantic segmentation of personalized deep brain structures for non-invasive brain stimulation

Perspectives on Optimized Transcranial Electrical Stimulation Based on Spatial Electric Field Modeling in Humans

Background: Transcranial electrical stimulation (tES) generates an electric field (or current density) in the brain through surface electrodes attached to the scalp. Clinical significance has been demonstrated, although with moderate and heterogeneous results partly due to a lack of control of the delivered electric currents. In the last decade, computational electric field analysis has allowed the estimation and optimization of the electric field using accurate anatomical head models. This review examines recent tES computational studies, providing a comprehensive background on the technical aspects of adopting computational electric field analysis as a standardized procedure in medical applications. Methods: Specific search strategies were designed to retrieve papers from the Web of Science database. The papers were initially screened based on the soundness of the title and abstract and then on their full contents, resulting in a total of 57 studies. Results: Recent trends were identified in individual- and population-level analysis of the electric field, including head models from non-neurotypical individuals. Advanced optimization techniques that allow a high degree of control with the required focality and direction of the electric field were also summarized. There is also growing evidence of a correlation between the computationally estimated electric field and the observed responses in real experiments. Conclusions: Computational pipelines and optimization algorithms have reached a degree of maturity that provides a rationale to improve tES experimental design and a posteriori analysis of the responses for supporting clinical studies.

Precise and Rapid Whole-Head Segmentation from Magnetic Resonance Images of Older Adults using Deep Learning

Whole-head segmentation from Magnetic Resonance Images (MRI) establishes the foundation for individualized computational models using finite element method (FEM). This foundation paves the path for computer-aided solutions in fields, particularly in non-invasive brain stimulation. Most current automatic head segmentation tools are developed using healthy young adults. Thus, they may neglect the older population that is more prone to age-related structural decline such as brain atrophy. In this work, we present a new deep learning method called GRACE, which stands for General, Rapid, And Comprehensive whole-hEad tissue segmentation. GRACE is trained and validated on a novel dataset that consists of 177 manually corrected MR-derived reference segmentations that have undergone meticulous manual review. Each T1-weighted MRI volume is segmented into 11 tissue types, including white matter, grey matter, eyes, cerebrospinal fluid, air, blood vessel, cancellous bone, cortical bone, skin, fat, and muscle. To the best of our knowledge, this work contains the largest manually corrected dataset to date in terms of number of MRIs and segmented tissues. GRACE outperforms five freely available software tools and a traditional 3D U-Net on a five-tissue segmentation task. On this task, GRACE achieves an average Hausdorff Distance of 0.21, which exceeds the runner-up at an average Hausdorff Distance of 0.36. GRACE can segment a whole-head MRI in about 3 seconds, while the fastest software tool takes about 3 minutes. In summary, GRACE segments a spectrum of tissue types from older adults T1-MRI scans at favorable accuracy and speed. The trained GRACE model is optimized on older adult heads to enable high-precision modeling in age-related brain disorders. To support open science, the GRACE code and trained weights are made available online and open to the research community at https://github.com/lab-smile/GRACE.

Monopolar tDCS might affect brainstem reflexes: A computational and neurophysiological study

OBJECTIVE

To assess whether monopolar multi-electrode transcranial direct current stimulation (tDCS) montages might selectively affect deep brain structures through computational predictions and neurophysiological assessment.

METHODS

Electric field distribution in deep brain structures (i.e., thalamus and midbrain) were estimated through computational models simulating tDCS with two monopolar and two monopolar multi-electrode montages. Monopolar multi-electrode tDCS was then applied to healthy subject, and effects on pontine and medullary circuitries was evaluated studying changes in blink reflex (BR) and masseter inhibitory reflex (MIR).

RESULTS

Computational results suggest that tDCS with monopolar multi-electrode montages might induce electric field intensities in deep brain structure comparable to those in grey matter, while neurophysiological results disclosed that BR and MIR were selectively modulated by tDCS only when cathode was placed over the right deltoid.

CONCLUSIONS

Multi-electrode tDCS (anodes over motor cortices, cathode over right deltoid) could induce significant electric fields in the thalamus and midbrain, and selectively affect brainstem neural circuits.

SIGNIFICANCE

Multi-electrode tDCS (anodes over motor cortices, cathode over right deltoid) might be further explored to affect brainstem activity, also in the context of non-invasive deep brain stimulation.

The effect of deep brain structure modeling on transcranial direct current stimulation-induced electric fields: An in-silico study

To study transcranial direct current stimulation (tDCS) and its effect on the brain, it could be useful to predict the distribution of the electric field induced in the brain with given tDCS parameters. As a solution, simulation with realistic computational models using magnetic resonance images (MRIs) have been widely used in the fields. With the recent advance of deep learning-based segmentation techniques of the brain, questions have been raised about if tDCS-induced electric field is affected by the deep brain structures. This study aimed to investigate the effect of the deep brain structure modeling on the induced electric field. To this end, we generated models with and without the deep brain structures by using an open MRI dataset comprising tDCS parameters, electric field simulation results and in-vivo intracranial recordings in the deep brain structures. We investigated the difference between the simulation results of the two models with a statistical analysis. Our results indicated that tDCS-induced electric fields and current flow in the brain are significantly different when the deep brain structures are considered.

EA-Net: Edge-aware network for brain structure segmentation via decoupled high and low frequency features

Automatic brain structure segmentation in Magnetic Resonance Image (MRI) plays an important role in the diagnosis of various neuropsychiatric diseases. However, most existing methods yield unsatisfactory results due to blurred boundaries and complex structures. Improving the segmentation ability requires the model to be explicit about the spatial localization and shape appearance of targets, which correspond to the low-frequency content features and the high-frequency edge features, respectively. Therefore, in this paper, to extract rich edge and content feature representations, we focus on the composition of the feature and utilize a frequency decoupling (FD) block to separate the low-frequency and high-frequency parts of the feature. Further, a novel edge-aware network (EA-Net) is proposed for jointly learning to segment brain structures and detect object edges. First, an encoder-decoder sub-network is utilized to obtain multi-level information from the input MRI, which is then sent to the FD block to complete the frequency separation. Further, we use different mechanisms to optimize both the low-frequency and high-frequency features. Finally, these two parts are fused to generate the final prediction. In particular, we extract the content mask and the edge mask from the optimization feature with different supervisions, which forces the network to learn the boundary features of the object. Extensive experiments are performed on two public brain MRI T1 scan datasets (the IBSR dataset and the MALC dataset) to evaluate the effectiveness of the proposed algorithm. The experiments show that the EA-Net achieves the best performance compared with the state-of-the-art methods, and improves the segmentation DSC score by 1.31% at most compared with the U-Net model and its variants. Moreover, we evaluate the EA-Net under different noise disturbances, and the results demonstrate the robustness and superiority of our method under low-quality noisy MRI. Code is available at https://github.com/huqian999/EA-Net.

DBSegment: Fast and robust segmentation of deep brain structures considering domain generalization

Segmenting deep brain structures from magnetic resonance images is important for patient diagnosis, surgical planning, and research. Most current state-of-the-art solutions follow a segmentation-by-registration approach, where subject magnetic resonance imaging (MRIs) are mapped to a template with well-defined segmentations. However, registration-based pipelines are time-consuming, thus, limiting their clinical use. This paper uses deep learning to provide a one-step, robust, and efficient deep brain segmentation solution directly in the native space. The method consists of a preprocessing step to conform all MRI images to the same orientation, followed by a convolutional neural network using the nnU-Net framework. We use a total of 14 datasets from both research and clinical collections. Of these, seven were used for training and validation and seven were retained for testing. We trained the network to segment 30 deep brain structures, as well as a brain mask, using labels generated from a registration-based approach. We evaluated the generalizability of the network by performing a leave-one-dataset-out cross-validation, and independent testing on unseen datasets. Furthermore, we assessed cross-domain transportability by evaluating the results separately on different domains. We achieved an average dice score similarity of 0.89 ± 0.04 on the test datasets when compared to the registration-based gold standard. On our test system, the computation time decreased from 43 min for a reference registration-based pipeline to 1.3 min. Our proposed method is fast, robust, and generalizes with high reliability. It can be extended to the segmentation of other brain structures. It is publicly available on GitHub, and as a pip package for convenient usage.

Dual U-Net-Based Conditional Generative Adversarial Network for Blood Vessel Segmentation with Reduced Cerebral MR Training Volumes

Segmenting vessels in brain images is a critical step for many medical interventions and diagnoses of illnesses. Recent advances in artificial intelligence provide better models, achieving a human-like level of expertise in many tasks. In this paper, we present a new approach to segment Time-of-Flight Magnetic Resonance Angiography (TOF-MRA) images, relying on fewer training samples than state-of-the-art methods. We propose a conditional generative adversarial network with an adapted generator based on a concatenated U-Net with a residual U-Net architecture (UUr-cGAN) to carry out blood vessel segmentation in TOF-MRA images, relying on data augmentation to diminish the drawback of having few volumes at disposal for training the model, while preventing overfitting by using regularization techniques. The proposed model achieves 89.52% precision and 87.23% in Dice score on average from the cross-validated experiment for brain blood vessel segmentation tasks, which is similar to other state-of-the-art methods while using considerably fewer training samples. UUr-cGAN extracts important features from small datasets while preventing overfitting compared to other CNN-based methods and still achieve a relatively good performance in image segmentation tasks such as brain blood vessels from TOF-MRA.

Factorisation-Based Image Labelling

Segmentation of brain magnetic resonance images (MRI) into anatomical regions is a useful task in neuroimaging. Manual annotation is time consuming and expensive, so having a fully automated and general purpose brain segmentation algorithm is highly desirable. To this end, we propose a patched-based labell propagation approach based on a generative model with latent variables. Once trained, our Factorisation-based Image Labelling (FIL) model is able to label target images with a variety of image contrasts. We compare the effectiveness of our proposed model against the state-of-the-art using data from the MICCAI 2012 Grand Challenge and Workshop on Multi-Atlas Labelling. As our approach is intended to be general purpose, we also assess how well it can handle domain shift by labelling images of the same subjects acquired with different MR contrasts.

Fully automated segmentation in temporal bone CT with neural network: a preliminary assessment study

BACKGROUND

Segmentation of important structures in temporal bone CT is the basis of image-guided otologic surgery. Manual segmentation of temporal bone CT is time- consuming and laborious. We assessed the feasibility and generalization ability of a proposed deep learning model for automated segmentation of critical structures in temporal bone CT scans.

METHODS

Thirty-nine temporal bone CT volumes including 58 ears were divided into normal (n = 20) and abnormal groups (n = 38). Ossicular chain disruption (n = 10), facial nerve covering vestibular window (n = 10), and Mondini dysplasia (n = 18) were included in abnormal group. All facial nerves, auditory ossicles, and labyrinths of the normal group were manually segmented. For the abnormal group, aberrant structures were manually segmented. Temporal bone CT data were imported into the network in unmarked form. The Dice coefficient (DC) and average symmetric surface distance (ASSD) were used to evaluate the accuracy of automatic segmentation.

RESULTS

In the normal group, the mean values of DC and ASSD were respectively 0.703, and 0.250 mm for the facial nerve; 0.910, and 0.081 mm for the labyrinth; and 0.855, and 0.107 mm for the ossicles. In the abnormal group, the mean values of DC and ASSD were respectively 0.506, and 1.049 mm for the malformed facial nerve; 0.775, and 0.298 mm for the deformed labyrinth; and 0.698, and 1.385 mm for the aberrant ossicles.

CONCLUSIONS

The proposed model has good generalization ability, which highlights the promise of this approach for otologist education, disease diagnosis, and preoperative planning for image-guided otology surgery.

Influence of layered skull modeling on the frequency sensitivity and target accuracy in simulations of transcranial current stimulation

With the development of electrical stimulation technology, especially the emergence of temporally interfering (TI) stimulation, it is necessary to discuss the influence of current frequency on stimulation intensity. Accurate skull modeling is important for transcranial current stimulation (tCS) simulation prediction because of its large role in dispersing current. In this study, we simulated different frequencies of transcranial alternating current stimulation (tACS) and TI stimulation in single-layer and layered skull model, compared the electric field via error parameters such as the relative difference measure and relative magnification factor. Pearson correlation analysis and t-test were used to measure the differences in envelope amplitude. The results showed that the intensity of electric field in the brain generated by per unit of stimulation current will increase with current frequency, and the layered skull model had a better response to frequency. An obvious pattern difference was found between the electric fields of the layered and single-layer skull individualized models. For TI stimulation, the Pearson correlation coefficient between the envelope distribution of the layered skull model and the single-layer skull was only 0.746 in the individualized model, which is clearly lower than the correlation coefficient of 0.999 determined from the spherical model. Higher carrier frequencies seemed to be easier to generate a large enough brain electric field envelope in TI stimulation. In conclusion, we recommend using layered skull models instead of single-layer skull models in tCS (particularly TI stimulation) simulation studies in order to improve the accuracy of the prediction of stimulus intensity and stimulus target.

Influence of segmentation accuracy in structural MR head scans on electric field computation for TMS and tES

In several diagnosis and therapy procedures based on electrostimulation effect, the internal physical quantity related to the stimulation is the induced electric field. To estimate the induced electric field in an individual human model, the segmentation of anatomical imaging, such as magnetic resonance image (MRI) scans, of the corresponding body parts into tissues is required. Then, electrical properties associated with different annotated tissues are assigned to the digital model to generate a volume conductor. However, the segmentation of different tissues is a tedious task with several associated challenges specially with tissues appear in limited regions and/or low-contrast in anatomical images. An open question is how segmentation accuracy of different tissues would influence the distribution of the induced electric field. In this study, we applied parametric segmentation of different tissues to exploit the segmentation of available MRI to generate different quality of head models using deep learning neural network architecture, named ForkNet. Then, the induced electric field are compared to assess the effect of model segmentation variations. Computational results indicate that the influence of segmentation error is tissue-dependent. In brain, sensitivity to segmentation accuracy is relatively high in cerebrospinal fluid (CSF), moderate in gray matter (GM) and low in white matter for transcranial magnetic stimulation (TMS) and transcranial electrical stimulation (tES). A CSF segmentation accuracy reduction of 10% in terms of Dice coefficient (DC) lead to decrease up to 4% in normalized induced electric field in both applications. However, a GM segmentation accuracy reduction of 5.6% DC leads to increase of normalized induced electric field up to 6%. Opposite trend of electric field variation was found between CSF and GM for both TMS and tES. The finding obtained here would be useful to quantify potential uncertainty of computational results.

Bilateral attention decoder: A lightweight decoder for real-time semantic segmentation

The encoder-decoder structure has been introduced into semantic segmentation to improve the spatial accuracy of the network by fusing high- and low-level feature maps. However, recent state-of-the-art encoder-decoder-based methods can hardly attain the real-time requirement due to their complex and inefficient decoders. To address this issue, in this paper, we propose a lightweight bilateral attention decoder for real-time semantic segmentation. It consists of two blocks and can fuse different level feature maps via two steps, i.e., information refinement and information fusion. In the first step, we propose a channel attention branch to refine the high-level feature maps and a spatial attention branch for the low-level ones. The refined high-level feature maps can capture more exact semantic information and the refined low-level ones can capture more accurate spatial information, which significantly improves the information capturing ability of these feature maps. In the second step, we develop a new fusion module named pooling fusing block to fuse the refined high- and low-level feature maps. This fusion block can take full advantages of the high- and low-level feature maps, leading to high-quality fusion results. To verify the efficiency of the proposed bilateral attention decoder, we adopt a lightweight network as the backbone and compare our proposed method with other state-of-the-art real-time semantic segmentation methods on the Cityscapes and Camvid datasets. Experimental results demonstrate that our proposed method can achieve better performance with a higher inference speed. Moreover, we compare our proposed network with several state-of-the-art non-real-time semantic segmentation methods and find that our proposed network can also attain better segmentation performance.