Biology subjects

Amorosino, G.

Publications and source records attributed to Amorosino, G..

8 recordsLinked to original sources

bioRxiv2026

Benchmarking Generalizability in Deep Learning-Based White Matter Tract Segmentation

Summary paragraphWhite matter tracts (WMTs) are the brains structural foundation for information transfer, underlying essential cognitive and behavioral functions. While diffusion MRI and tractography enable non-invasive mapping of these pathways, automated segmentation often lacks generalizability across diverse data sources. We conducted a systematic, cross-dataset evaluation of four state-of-the-art deep learning architectures, benchmarking their performance across independent datasets with varying acquisition protocols and populations. CNN-based models such as TractSeg achieved the highest within-domain accuracy, but performance dropped sharply under domain shift, most severely when we applied adult-trained models to pediatric data. To address this degradation, we introduce Ensemble White Matter Tract Segmentation (EWMTS), which combines complementary models to partially recover accuracy under domain shift, although performance still falls short of within-domain levels. By openly releasing this benchmark and a reproducible processing pipeline, we provide the neuroimaging community with a framework to develop and benchmark segmentation models across the heterogeneity of real-world neuroimaging data.

neuroscience↗

bioRxiv2026

Short-Term Developmental Trajectories of Dorsal-Ventral Pathways and Their Relationships with First-Grade Learning

The first year of formal schooling is a year of foundational reading and math learning, and individual differences emerging within this single year predict academic achievement decades later. Yet, how brain changes throughout this critical year relate to individual differences in reading and math learning remains uncharacterized. In this pre-registered study (https://osf.io/97ybe), we acquired monthly both behavioral assessments of reading- and math-learning, and diffusion-weighted MRI scans to measure white matter microstructure, across the first-grade year. Behavioral learning trajectories follow either a sigmoid for reading or an inverted-U for math. Month-to-month microstructural changes in the right middle longitudinal fasciculus predicted corresponding changes in math performance, but not in reading. Findings highlight white matter microstructure as a dynamic substrate of early math learning, and reveal a more general principle: rapid changes in white-matter microstructure during the foundational learning window may be associated with distinct academic domains. FundingR01 HD114489

neuroscience↗

bioRxiv2026

Human white matter tracts aligned with canonical functional brain networks

White matter tracts form the structural backbone of large-scale brain networks, yet their relation to functional organization remains poorly defined. Although major pathways are well characterized anatomically, their relationship to distributed cognitive systems has not been systematically established. Here, we constructed a population-level white matter tract termination atlas and projected tract endpoints onto the cortical surface to characterize their spatial organization. We integrated this atlas with large-scale meta-analytic decoding to derive functional profiles for individual tracts. Functional decoding revealed that white matter tracts exhibited distinct and biologically interpretable cognitive signatures. Hierarchical clustering of these profiles further showed that tracts organize into coherent ensembles defined by shared functional associations. These tract ensembles recapitulated canonical intrinsic brain networks across multiple cortical atlases, including a notable ensemble that demonstrated alignment with the default mode, salience, and frontoparietal control networks, corresponding to the core architecture of the triple-network model of cognitive control. This finding identified a candidate structural backbone linking distributed functional systems implicated across neuropsychiatric conditions. Together, these results demonstrate that white matter architecture is organized according to large-scale functional principles and establish a tract-to-network framework for linking structural connectivity to cognition.

neuroscience↗

bioRxiv2026

A retinotopic wiring principle of the human brain

Summary paragraphA central challenge in neuroscience is to understand how anatomical connectivity links neural representations to behavior. In sensory systems, cortical areas encode structured maps of the external world, yet it remains unclear whether and how this representational geometry constrains long-range connectivity in the living human brain. Here, we show that structural connectivity in the human visual cortex preserves retinotopic organization, revealing a retinotopic wiring principle; cortical regions representing the same locations in visual space preferentially connect. Across more than 1,700 participants spanning ages 2 to 88 years, we further show that well-known perceptual asymmetries correspond to systematic asymmetries in connectivity within early visual cortex, but not in connections linking visual cortex to the rest of the brain. We introduce a scalable approach to map retinotopic connectivity when tractography alone is unreliable. Together, these findings demonstrate that anatomical connectivity preserves the geometry of neural representations, providing a general principle linking brain structure, function, and behavior.

neuroscience↗

bioRxiv2025

Supervised white matter bundle segmentation in glioma patients with transfer learning

In clinical settings, the virtual dissection of white matter tracts represents an informative source of information for monitoring neurological conditions or to support the planning of a treatment. The automation of this task using data-driven approaches, and in particular deep learning models, provides promising evidence of good accuracy when carried out on healthy individuals. However, the lack of large clinical datasets and the profound differences between healthy and clinical populations hinder the translation of these results to patients. Here, we investigated for the first time the effectiveness of transfer learning in adapting a deep learning architecture trained on a healthy population to glioma patients. Importantly, we provided the first thorough characterization of domain shift and its complexity, distinguishing systematic (i.e. measurement and pre-processing related) from tumor specific components. Our results suggest that (i) models trained on a large normative healthy population have a significant performance drop when the inference is carried out on patients; (ii) transfer learning can be an effective strategy to overcome the shortage of clinical data and to manage the systematic shift; (iii) fine tuning of the learning model cannot accommodate large white matter deformations induced by the tumor. The results were coherent across the five white matter bundles and three different input modalities tested, highlighting their robustness and generalizability. Our work provides valuable insights for advancing automated white matter segmentation in clinical populations and enhancing clinical transfer learning applications. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=72 SRC="FIGDIR/small/659452v1_ufig1.gif" ALT="Figure 1"> View larger version (16K): org.highwire.dtl.DTLVardef@1c2bbbcorg.highwire.dtl.DTLVardef@1b3555eorg.highwire.dtl.DTLVardef@a9bb4dorg.highwire.dtl.DTLVardef@147dc5e_HPS_FORMAT_FIGEXP M_FIG C_FIG HighlightsO_LIWhite matter bundle segmentation is essential for planning glioma tumor resection. C_LIO_LIWe trained a deep learning model on a healthy dataset to segment white matter bundles. C_LIO_LIWe observed a sub-optimal performance on glioma patients due to domain shift. C_LIO_LITransfer learning mitigated domain shift between healthy and clinical populations. C_LIO_LIWe highlighted domain shift domains solved/not solved by transfer learning. C_LI

neuroscience↗

bioRxiv2023

Supervised White Matter Bundle Segmentation in Glioma Patients with Transfer Learning

In clinical neuroscience, the segmentation of the main white matter bundles is propaedeutic for many tasks such as preoperative neurosurgical planning and monitoring of neuro-related diseases. Automating bundle segmentation with data-driven approaches and deep learning models has shown promising accuracy in the context of healthy individuals. The lack of large clinical datasets is preventing the translation of these results to patients. Inference on patients data with models trained on healthy population is not effective because of domain shift. This study aims to carry out an empirical analysis to investigate how transfer learning might be beneficial to overcome these limitations. For our analysis, we consider a public dataset with hundreds of individuals and a clinical dataset of glioma patients. We focus our preliminary investigation on the corticospinal tract. The results show that transfer learning might be effective in partially overcoming the domain shift.

neuroscience↗

bioRxiv2022

How Does White Matter Registration Affect Tractography Alignment?

Tractography is a powerful method to represent the structural connectivity of the brain white matter. Nevertheless, the comparison of these data structures between two individuals is still an open challenge because of their complexity, e.g. digital representation of millions of fibers as polylines. The scientific community spent a meaningful effort to develop new methods of white matter registration aiming to take advantage of diffusion MRI models. Despite the effort to improve the registration of the white matter, little is known about the effect of the registration on tractogram alignment. The main issue for an empirical evaluation is the lack of ground truth, e.g. a sample of data where the correct alignment is validated by experts. This work aims to overcome this drawback by proposing an evaluation framework based on the matching of homologous fiber structures, e.g. known neuroanatomical bundles. The contribution is a quantitative comparison of how the most representative methods of white matter registration affect tractogram alignment.

neuroscience↗

bioRxiv2022

Connectivity by the Frontal Aslant Tract (FAT) explains local functional specialization of the superior and inferior frontal gyri in humans while choosing predictive over reactive strategies: a tractography-guided TMS study

Predictive and reactive behaviors represent two mutually exclusive strategies for successfully completing a sensorimotor task. It is thought that predictive actions are based on the medial premotor system, in the superior frontal gyrus (SFG) and reactive stimulus-response behaviors rely on a lateral premotor system, in the inferior frontal gyrus (IFG). The frontal aslant tract (FAT), a white matter tract connecting SFG and IFG, is a possible neural substrate of the predictive/reactive interactions. We used diffusion-weighted imaging (DWI) of 17 male and female healthy human volunteers, to dissect 3 sub-bundles of fibers belonging to the left FAT (bundles 1, 2 and 3), arising ventrally from 1) the ventral precentral gyrus, 2) midway between the PCG and pars opercularis (POp) and 3) the POp and terminating dorsally in 3 different parts of the SFG, in a caudal-rostral order. We applied online transcranial magnetic stimulation (TMS) to 6 spots, corresponding to the medial and lateral terminations of bundles 1-3 during the fixed-duration set period of a delayed reaction task, that can be solved using a predictive (anticipatory) strategy or with a reactive strategy. Results showed that TMS changed the frequency of predictive/reactive strategies only when applied over 2 spots, the SFG and IFG terminations of bundle 2. Importantly, the effects of TMS were opposite when applied to the SFG or to the IFG. Our data show that the SFG and the IFG have opposite roles in producing predictive or reactive behavior and that reciprocal integration or competition is probably mediated by the FAT. Significance StatementAs is well-known by athletes at starting blocks, interaction with the world can occur with a predictive strategy (anticipating a GO-signal) or a reactive strategy (waiting for the GO-signal to be manifest) and they are mutually exclusive. Here we showed, by using non-invasive brain stimulation (TMS), that two specific cortical regions in the superior frontal gyrus (SFG) and the inferior frontal gyrus (IFG) have opposite roles in facilitating a predictive or a reactive strategy. Importantly these two very distant regions but with highly interconnected functions are specifically connected by a small white matter bundle, which probably mediates the competition between predictive and reactive strategies. More generally, we show that the implementing anatomical connectivity in TMS studies strongly reduces spatial noise.

neuroscience↗