TractCloud‐FOV: Deep Learning‐Based Robust Tractography Parcellation in Diffusion MRI With Incomplete Field of View

Human Brain Mapping, Journal Year: 2025, Volume and Issue: 46(5)

Published: April 1, 2025

ABSTRACT Tractography parcellation classifies streamlines reconstructed from diffusion MRI into anatomically defined fiber tracts for clinical and research applications. However, scans often have incomplete fields of view (FOV) where brain regions are partially imaged, leading to partial, or truncated tracts. To address this challenge, we introduce TractCloud‐FOV, a deep learning framework that robustly parcellates tractography under conditions FOV. We propose novel training strategy, FOV‐Cut Augmentation (FOV‐CA), in which synthetically cut tractograms simulate spectrum real‐world inferior FOV cutoff scenarios. This data augmentation approach enriches the set with realistic streamlines, enabling model achieve superior generalization. evaluate proposed TractCloud‐FOV on both two real‐life datasets significantly outperforms several state‐of‐the‐art methods all testing terms streamline classification accuracy, generalization ability, tract anatomical depiction, computational efficiency. Overall, achieves efficient consistent

Language: Английский

---

Patterns of Cerebellar–Cortical Structural Covariance Mirror Anatomical Connectivity of Sensorimotor and Cognitive Networks

Human Brain Mapping, Journal Year: 2025, Volume and Issue: 46(1)

Published: Jan. 1, 2025

ABSTRACT The cortex and cerebellum are densely connected through reciprocal input/output projections that form segregated circuits. These circuits shown to differentially connect anterior lobules of the sensorimotor regions, Crus I II prefrontal regions. This differential connectivity pattern leads hypothesis individual differences in structure should be related, especially for To test this hypothesis, we examined covariation between volumes lateral cognitive measures cortical thickness (CT) surface area (SA) across whole brain a sample 270 young adults drawn from HCP dataset. We observed patterns cerebellar–cortical covariance differed networks. Anterior motor showed greater with regions cortex, while frontal temporal Interestingly, cerebellar volume predominantly negative relationships CT positive SA. Individual SA thought largely under genetic control is more malleable by experience. suggests may stable feature, whereas affected development. Additionally, similarity metrics revealed gradual transition lobules, consistent evidence functional gradients within cerebellum. Taken together, these findings known structural cortex. They also shed new light on possibly differing area. Finally, our interactive specialization framework which proposes structurally functionally develop concert.

Language: Английский

---

Towards an informed choice of diffusion MRI image contrasts for cerebellar segmentation

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: March 11, 2025

Abstract The fine-grained segmentation of cerebellar structures is an essential step towards supplying increasingly accurate anatomically informed analyses, including, for example, white matter diffusion magnetic resonance imaging (MRI) tractography. Cerebellar tissue typically performed on structural data, such as T1-weighted while connectivity between segmented regions mapped using MRI tractography data. Small deviations in to data co-registration may negatively impact analyses. Reliable brain directly helps circumvent inaccuracies. Diffusion enables the computation many image contrasts, including a variety microstructure maps. While multiple methods have been proposed MRI, little attention has paid systematic evaluation performance different available input contrasts task. In this work, we evaluate and compare MRI-derived Specifically, include spherical mean (diffusion-weighted average) b0 (non-diffusion-weighted local signal parameterization (diffusion tensor kurtosis fit maps), contrast that most commonly employed We train popular deep-learning architecture publicly dataset (HCP-YA), leveraging region labels from atlas-based SUIT pipeline. By training testing diffusion-MRI-derived inputs, find computed b=1000 s/mm 2 shell provides stable across metrics significantly outperforms are traditionally used machine learning MRI. Key points provide evidence about dMRI structure deep neural network. improved performance. easy compute can be retrospective clinical

Language: Английский

---

Editorial overview: Role of the cerebellum in cognition and behavior

Current Opinion in Behavioral Sciences, Journal Year: 2024, Volume and Issue: 60, P. 101444 - 101444

Published: Sept. 11, 2024

Language: Английский

---

Multiscale Gradients of Corticopontine Structural Connectivity

Published: Oct. 17, 2024

The cerebellum’s involvement in a range of cognitive, emotional, and motor processes has become increasingly evident. Given the uniformity cerebellar cortex’s cellular architecture its contributions to varied are thought be partially mediated by patterns reciprocal connectivity with rest brain. A better understanding these connections is therefore fundamental disentangling contribution cognition behavior. While have been studied extensively non-human animals using invasive methods, we limited knowledge humans. current work reconstructed corticopontine projection, first segment downstream between cerebral cortices, diffusion MRI tractography human in-vivo whole brain data an independent higher resolution postmortem brainstem dataset. Dimensionality reduction was used characterize pattern cortical projections pons as two overlapping gradients that were consistent across participants datasets: medial lateral core belt. Our findings align done advance our this connection humans – providing valuable context growing body research, offering insights into impacts damage along pathway, informing clinical interventions.

Language: Английский

---