Deep Learning Methods for Identification of White Matter Fiber Tracts: Review of State-of-the-Art and Future Prospective

Changing the Paradigm for Tractography Segmentation in Neurosurgery: Validation of a Streamline-Based Approach

In glioma surgery, maximizing the extent of resection while preserving cognitive functions requires an understanding of the unique architecture of the white matter (WM) pathways of the single patient and of their spatial relationship with the tumor. Tractography enables the reconstruction of WM pathways, and bundle segmentation allows the identification of critical connections for functional preservation. This study evaluates the effectiveness of a streamline-based approach for bundle segmentation on a clinical dataset as compared to the traditional ROI-based approach. We performed bundle segmentation of the arcuate fasciculus, of its indirect anterior and posterior segments, and of the inferior fronto-occipital fasciculus in the healthy hemisphere of 25 high-grade glioma patients using both ROI- and streamline-based approaches. ROI-based segmentation involved manually delineating ROIs on MR anatomical images in Trackvis (V0.6.2.1). Streamline-based segmentations were performed in Tractome, which integrates clustering algorithms with the visual inspection and manipulation of streamlines. Shape analysis was conducted on each bundle. A paired t-test was performed on the irregularity measurement to compare segmentations achieved with the two approaches. Qualitative differences were evaluated through visual inspection. Streamline-based segmentation consistently yielded significantly lower irregularity scores (p < 0.001) compared to ROI-based segmentation for all the examined bundles, indicating more compact and accurate bundle reconstructions. Qualitative assessment identified common biases in ROI-based segmentations, such as the inclusion of anatomically implausible streamlines or streamlines with undesired trajectories. Streamline-based bundle segmentation with Tractome provides reliable and more accurate reconstructions compared to the ROI-based approach. By directly manipulating streamlines rather than relying on voxel-based ROI delineations, Tractome allows us to discern and discard implausible or undesired streamlines and to identify the course of WM bundles even when the anatomy is distorted by the lesion. These features make Tractome a robust tool for bundle segmentation in clinical contexts.

Assessment of U-Net in the segmentation of short tracts: Transferring to clinical MRI routine

Accurately studying structural connectivity requires precise tract segmentation strategies. The U-Net network has been widely recognized for its exceptional capacity in image segmentation tasks and provides remarkable results in large tract segmentation when high-quality diffusion-weighted imaging (DWI) data are used. However, short tracts, which are associated with various neurological diseases, pose specific challenges, particularly when high-quality DWI data acquisition within clinical settings is concerned. Here, we aimed to evaluate the U-Net network ability to segment short tracts by using DWI data acquired in different experimental conditions. To this end, we conducted three types of training experiments involving 350 healthy subjects and 11 white matter tracts, including the anterior, posterior, and hippocampal commissure, fornix, and uncinated fasciculus. In the first experiment, the model was exclusively trained with high-quality data of the Human Connectome Project (HCP) dataset. The second experiment focused on images of healthy subjects acquired from a local hospital dataset, representing a typical clinical routine acquisition. In the third experiment, a hybrid training approach was employed, combining data of the HCP and local hospital datasets. Then, the best model was also tested in unseen DWIs of 10 epilepsy patients of the local hospital and 10 healthy subjects acquired on a scanner from another company. The outcomes of the third experiment demonstrated a notable enhancement in performance when contrasted with the preceding trials. Specifically, the short tracts within the local hospital dataset achieved Dice scores ranging between 0.60 and 0.65. Similar intervals were obtained with HCP data in the first experiment, and a substantial improvement compared to the scores between 0.37 and 0.50 obtained with the local hospital dataset at the same experiment. This improvement persisted when the method was applied to diverse scenarios, including different scanner acquisitions and epilepsy patients. These results indicate that combining datasets from different sources, coupled with resolution standardization strengthens the neural network ability to generalize predictions across a spectrum of datasets. Nevertheless, short tract segmentation performance is intricately linked to the training composition, to validation, and to testing data. Moreover, curved tracts have intricate structural nature, which adds complexities to their segmenting. Although the network training approach tested herein has provided promising results, caution must be taken when extrapolating its application to datasets acquired under distinct experimental conditions, even in the case of higher-quality data or analysis of long or short tracts.

NeurostimML: a machine learning model for predicting neurostimulation-induced tissue damage

Objective. The safe delivery of electrical current to neural tissue depends on many factors, yet previous methods for predicting tissue damage rely on only a few stimulation parameters. Here, we report the development of a machine learning approach that could lead to a more reliable method for predicting electrical stimulation-induced tissue damage by incorporating additional stimulation parameters.Approach. A literature search was conducted to build an initial database of tissue response information after electrical stimulation, categorized as either damaging or non-damaging. Subsequently, we used ordinal encoding and random forest for feature selection, and investigated four machine learning models for classification: Logistic Regression, K-nearest Neighbor, Random Forest, and Multilayer Perceptron. Finally, we compared the results of these models against the accuracy of the Shannon equation.Main Results. We compiled a database with 387 unique stimulation parameter combinations collected from 58 independent studies conducted over a period of 47 years, with 195 (51%) categorized as non-damaging and 190 (49%) categorized as damaging. The features selected for building our model with a Random Forest algorithm were: waveform shape, geometric surface area, pulse width, frequency, pulse amplitude, charge per phase, charge density, current density, duty cycle, daily stimulation duration, daily number of pulses delivered, and daily accumulated charge. The Shannon equation yielded an accuracy of 63.9% using akvalue of 1.79. In contrast, the Random Forest algorithm was able to robustly predict whether a set of stimulation parameters was classified as damaging or non-damaging with an accuracy of 88.3%.Significance. This novel Random Forest model can facilitate more informed decision making in the selection of neuromodulation parameters for both research studies and clinical practice. This study represents the first approach to use machine learning in the prediction of stimulation-induced neural tissue damage, and lays the groundwork for neurostimulation driven by machine learning models.

A systematic review of automated methods to perform white matter tract segmentation

White matter tract segmentation is a pivotal research area that leverages diffusion-weighted magnetic resonance imaging (dMRI) for the identification and mapping of individual white matter tracts and their trajectories. This study aims to provide a comprehensive systematic literature review on automated methods for white matter tract segmentation in brain dMRI scans. Articles on PubMed, ScienceDirect [NeuroImage, NeuroImage (Clinical), Medical Image Analysis], Scopus and IEEEXplore databases and Conference proceedings of Medical Imaging Computing and Computer Assisted Intervention Society (MICCAI) and International Symposium on Biomedical Imaging (ISBI), were searched in the range from January 2013 until September 2023. This systematic search and review identified 619 articles. Adhering to the specified search criteria using the query, "white matter tract segmentation OR fiber tract identification OR fiber bundle segmentation OR tractography dissection OR white matter parcellation OR tract segmentation," 59 published studies were selected. Among these, 27% employed direct voxel-based methods, 25% applied streamline-based clustering methods, 20% used streamline-based classification methods, 14% implemented atlas-based methods, and 14% utilized hybrid approaches. The paper delves into the research gaps and challenges associated with each of these categories. Additionally, this review paper illuminates the most frequently utilized public datasets for tract segmentation along with their specific characteristics. Furthermore, it presents evaluation strategies and their key attributes. The review concludes with a detailed discussion of the challenges and future directions in this field.

NeurostimML: A machine learning model for predicting neurostimulation-induced tissue damage

Objective

The safe delivery of electrical current to neural tissue depends on many factors, yet previous methods for predicting tissue damage rely on only a few stimulation parameters. Here, we report the development of a machine learning approach that could lead to a more reliable method for predicting electrical stimulation-induced tissue damage by incorporating additional stimulation parameters.

Approach

A literature search was conducted to build an initial database of tissue response information after electrical stimulation, categorized as either damaging or non-damaging. Subsequently, we used ordinal encoding and random forest for feature selection, and investigated four machine learning models for classification: Logistic Regression, K-nearest Neighbor, Random Forest, and Multilayer Perceptron. Finally, we compared the results of these models against the accuracy of the Shannon equation.

Main Results

We compiled a database with 387 unique stimulation parameter combinations collected from 58 independent studies conducted over a period of 47 years, with 195 (51%) categorized as non-damaging and 190 (49%) categorized as damaging. The features selected for building our model with a Random Forest algorithm were: waveform shape, geometric surface area, pulse width, frequency, pulse amplitude, charge per phase, charge density, current density, duty cycle, daily stimulation duration, daily number of pulses delivered, and daily accumulated charge. The Shannon equation yielded an accuracy of 63.9% using a k value of 1.79. In contrast, the Random Forest algorithm was able to robustly predict whether a set of stimulation parameters was classified as damaging or non-damaging with an accuracy of 88.3%.

Significance

This novel Random Forest model can facilitate more informed decision making in the selection of neuromodulation parameters for both research studies and clinical practice. This study represents the first approach to use machine learning in the prediction of stimulation-induced neural tissue damage, and lays the groundwork for neurostimulation driven by machine learning models.