Full Professor of Medical Physics · University of Rome Tor Vergata · Martinos Center, MGH / Harvard Medical School
BACKGROUND
Biography
Nicola Toschi is Full Professor of Medical Physics and Head of the Medical Physics Section at the University of Rome Tor Vergata. He is also a research investigator at the Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital and Harvard Medical School. His research combines physical modelling, biomedical imaging and artificial intelligence to study physiological mechanisms and develop quantitative methods for biomedical measurement and data analysis. His work spans MRI and PET, brain connectivity, neuromodulation, nanomedicine and machine learning for the analysis of images and neural signals.
He trained in physics at Imperial College London and Tor Vergata, in mathematical modelling at the University of Oxford, and in natural sciences at Ludwig Maximilian University of Munich and the Max Planck Institute of Psychiatry. He subsequently specialised in medical physics at Tor Vergata. He coordinates international research collaborations and contributes to medical, postgraduate and doctoral education, including training in neurotechnology and artificial intelligence.
Thrombotic events (TEs) occur in up to 40% of patients with vacuoles, E1 enzyme, X-linked, autoinflammatory, and somatic (VEXAS) syndrome, but data on its clinical-genomics features and anticoagulation strategies are limited. To gain more insight into this, we conducted a two-step study evaluating the prevalence and outcome of TE in VEXAS. First, among 1086 patients followed for TEs, 198 were men aged >40 years with unprovoked thrombosis and no known thrombophilia; 21 also had at least one VEXAS-compatible feature and underwent UBA1 exon 3 testing. No UBA1 mutation was detected in these 21 patients. Next, we leveraged our Italian VEXAS network, and we accrued 87 molecularly confirmed Italian VEXAS cases (median age 70 years). Any history of TE was documented in 43/87 patients (49%), deep vein thrombosis being the most common (71%). Because follow-up varied, incident thrombosis was analyzed using a time-to-first-event framework from molecular VEXAS diagnosis, with death without prior TE treated as a competing event. Among 49 patients without prior/concomitant TE, five developed incident post-diagnosis TE; the 24-month cumulative incidence was 18.3%. Thrombophilia testing revealed a 15% co-occurrence, including heterozygous Factor V Leiden, Factor II G20210A, and anti-cardiolipin antibodies. Treatments comprised direct oral anticoagulants (DOACs) (51%), low molecular weight heparin (LMWH) (28%), Fondaparinux (14%), and vitamin K antagonists (AVKs) (7%). Notably, 27% experienced multiple TEs, of which 22% occurring despite anticoagulation during disease flares. Our findings provide an updated cartography of VEXAS-related TE, suggesting early screening for thrombophilia in these patients to inform both personalized anticoagulation and disease-control strategies.
Recurrent neural networks are studied along two largely parallel tracks: as machine-learning models evaluated by task performance and as computational-neuroscience models of cortical circuits evaluated by dynamical realism. Reservoir computing offers a meeting point, yet the link between dynamical regime and computational performance has not been systematically mapped in biologically constrained spiking architectures. We treat the Brunel balanced excitatory-inhibitory network as a reservoir and characterize separation capacity (kernel quality) and transient memory (corrected linear memory capacity, validated by non-parametric mutual information) across the full phase diagram. The analysis uses a four-state Markov source whose Shannon entropy rate is set in closed form by a single parameter at fixed marginal entropy. Both capabilities increase monotonically with the inhibitory ratio g, remaining jointly highest in the asynchronous irregular regime, with diminishing increments consistent with eventual saturation; the synchronous irregular regime, despite a network timescale three orders of magnitude longer, supports neither. Memory further requires sparse input coupling: dense coupling collapses the driven timescale and erases memory in every regime. Inhibitory balance thus emerges as a unified architectural control parameter, providing a quantitative design criterion for cortical-circuit modeling and reservoir computing applications.
Recent advancements in neural decoding have shown promising results in reconstructing visual experiences from brain activity. However, existing approaches focus primarily on decoding within a single dataset or subject, which limits generalization across various sources of neuroimaging. In this work, we propose a novel framework for the decoding of visual stimuli between subjects and between data sets, integrating neural recordings from multiple publicly available fMRI datasets. To address inherent intersubject and interdataset variability, we introduce a contrastive learning-based alignment strategy using image embeddings from a pre-trained IP-Adapter model. Our approach learns a shared latent space by aligning subject-specific neural representations with image features, enabling generalized decoding across both subjects and datasets. In addition, we propose a simple yet effective data augmentation method using ridge regression. This method synthesizes realistic fMRI-like signals from novel images by predicting voxel activity and injecting learned noise distributions, thus enhancing training diversity and model robustness. To the best of our knowledge, while several recent studies have explored cross-subject decoding, we extend recent cross-subject decoding efforts by training a single unified framework jointly across multiple public fMRI datasets and subjects, enabling cross-dataset transfer in addition to cross-subject generalization. We distinguish this multi-dataset unified training setting, where each dataset contributes training data, from a stricter leave-one-dataset-out transfer setting in which the target dataset is excluded from source pretraining and used only for lightweight alignment-layer adaptation. Empirically, our unified model achieves strong semantic reconstruction across datasets (e.g., up to 94.8% CLIP similarity on NSD (AUG) and 0.403 SSIM on BOLD5000 after lightweight finetuning), demonstrating robust cross-subject and cross-dataset transfer.
Characterizing the information content of intracortical signals during visual processing is a central challenge in systems neuroscience. We address the problem of decoding visual information from high-density intracortical recordings in primates, using the THINGS Ventral Stream Spiking Dataset. We systematically evaluate the effects of model architecture, training objectives, and data scaling on decoding performance. Results show that decoding accuracy is jointly driven by non-linearity and selective temporal aggregation, rather than heavier sequence modelling in this data regime. A simple model combining temporal attention with a shallow MLP achieves up to 70% top-1 image retrieval accuracy, outperforming linear baselines as well as recurrent and convolutional approaches. Scaling analyses reveal predictable diminishing returns with increasing input dimensionality and dataset size. Building on these findings, we design a modular generative decoding pipeline that combines low-resolution latent reconstruction with semantically conditioned diffusion, generating plausible images from 200 ms of brain activity. This framework provides principles for brain-computer interfaces and semantic neural decoding.
Implantable and wearable devices require antennas that are both miniaturized and efficient, yet conventional designs are constrained by narrow bandwidth and orientation sensitivity. We report overtone ultrawideband magnetoelectric (OUWB-ME) antennas that exploit higher-order acoustic modes in polished silicon substrates to achieve a 22.6-gigahertz -10-decibel bandwidth and overtone capability in the 3- to 4-gigahertz range. Packaged into "μBots," these magnetoelectric heterostructures bonded with silver nanoparticle inks maintain stable operation under biological loading. In vitro assays confirm the biocompatibility of aluminum nitride and the protective role of parylene encapsulation for iron-gallium. Ex vivo rat and human tissues reshape transmission and reflection spectra, with reproducible frequency windows near 3.3 and 3.9 gigahertz. μBots enable real-time audiovisual telemetry using software-defined radios and exhibit compatibility with 7-tesla magnetic resonance imaging. By combining wideband response, robustness to misalignment, and biocompatible packaging, OUWB-ME μBots provide a scalable platform for wireless bio-integrated communication and telemetry.