University of Modena and Reggio Emilia
Archivio istituzionale della ricerca - Università di Modena e Reggio EmiliaNot a member yet
116400 research outputs found
Sort by
Histological Brain Imaging Super-resolution with Frequency-guided Diffusion Models
No full textHigh-resolution histological imaging provides essential detail for quantitative brain modeling, yet acquiring whole-brain data at micrometer scale remains technically and economically challenging. This work introduces Brain-SR, a diffusion-based super-resolution framework designed to reconstruct high-resolution cortical sections from low-resolution BigBrain data. Building upon the InvSR paradigm, our method performs resolution enhancement in the latent space of a pretrained variational autoencoder, guided by a task-specific noise-predictor network. A key contribution is a frequency-domain supervision term that compares the magnitude spectra of predicted and target patches, enforcing spectral consistency while remaining robust to local misalignments. Quantitative evaluations demonstrate that Brain-SR achieves substantial improvements in LPIPS (-27%) and FID (-58%) compared to baseline diffusion Super-Resolution, while spectral analysis confirms accurate recovery of the frequency distribution. The resulting reconstructions preserve neuronal structures consistent with high-resolution references, offering a practical step toward large-scale, morphologically faithful brain histology reconstruction. The code is publicly available to support reproducibility: https://github.com/AImageLab-zip/Brain-SR
Ellero e Finzi: i destini incrociati di due penalisti italiani, tra Unità nazionale e oppressione fascista
No full textTh is essay off ers a historical and critical refl ection on Pietro Ellero
and Marco Finzi, whose archives and library were respectively at the center of the PRIN-PNRR research project entitled “CriArcLi - Italian professors of Criminal law’s archives and libraries: models, digitisation, and public engagement.” The reflections collected here are based on the contents of the reports presented during the fi nal conference of the aforementioned research project, which was held in Trento from October 1st to 3rd 2025, and whose contributions are collected in this same volume
Artificial intelligence predicts GBA1 mutated status in Parkinson’s Disease patients
Get PDFBackground: GBA1 variants are the major genetic risk factor for Parkinson’s Disease (PD) and
account for 5–30% of PD cases depending on the population and age at onset of the disease.
Objectives: The aim of this study was to assess whether Artificial Intelligence (AI) could predict GBA1-mutated genotype in PD (GBA1-PD). Particularly, the main objective was to identify a Machine Learning (ML) model capable of accurately providing a pre-test estimate of GBA1-mutated status, relying on the clinical and demographic variables with the highest predictive value.
Methods: A cohort of GBA1-PD patients has been paired with non-mutated PD (NM-PD). The dataset comprised 58 GBA1-PD and 58 NM-PD, for each of whom 124 features were recorded. A Leave-One-Out cross-validation method was employed for testing and SHapley Additive exPlanations (SHAP) for examine each feature’s contribution. XGBoost resulted the most effective ML model for this supervised classification task.
Results: Through AI, we developed a model based on four specific clinical features with significant impact in predicting GBA1-mutated genotype with an accuracy of 73%, reaching 94% in a subset of patients where the model has a SHAP confidence level greater than 80%. These variables included family history and scores for cognitive (MDS-UPDRS 1.1) and motor impairment (MDS-UPDRS 3.8a and 3.8b and rigidity subscore). Conclusions: This study underlies the potential of AI in enhancing targeted genetic screening in PD, especially in clinical settings where resources are limited. Main limitations of this study are the modest sample size and lack of external validation. Further studies on larger, independent cohorts are needed to refine the predictive model
Ultrasound-guided infiltration of the pudendal nerve: a technical approach for neuropathic pain management
No full textPurpose: Pudendal neuropathy is a debilitating condition often underdiagnosed due to its complex clinical presentation and overlapping symptoms with other pelvic disorders. This review aims to provide an updated synthesis of anatomical, clinical, and technical aspects of ultrasound-guided pudendal nerve infiltration, highlighting its diagnostic and therapeutic relevance in neuropathic pelvic pain. Methods: A narrative analysis was conducted of the pudendal nerve’s anatomy, etiologies of neuropathy, clinical manifestations, diagnostic tools, and image-guided intervention strategies. Particular attention was dedicated to high-resolution ultrasound (HRUS) for anatomical visualization and to the technical considerations underlying perineural injection procedures. Results: Perineural infiltration of the pudendal nerve provides immediate, partial pain relief after anesthetic administration, with the addition of corticosteroids contributing to sustained relief. Ultrasound guidance minimizes complications and has proven superior to traditional landmark-based approaches. The technique demonstrates feasibility, safety, and reproducibility in clinical practice, for both diagnostic and therapeutic purposes. However, clinical results have varied, and repeated sessions or integration with multimodal strategies, including physical therapy, drug therapy, and lifestyle modifications, have often been required. Conclusion: Ultrasound-guided pudendal nerve infiltration is a minimally invasive, safe, and effective technique for both diagnosis and treatment of pudendal neuropathy. While offering rapid pain relief and confirming nerve involvement, infiltrations should be considered as part of a broader multimodal management strategy. Future directions include standardization of protocols, refinement of imaging guidance, and exploration of novel injectates or regenerative therapies to optimize long-term outcomes
Approcci terapeutici basati su cellule staminali neurali e neurostimolazione per curare l’epilessia del lobo temporale: sfide e risultati
No full textL’epilessia del lobo temporale (TLE) è il tipo di epilessia più comune negli adulti, rappresentando il 50%- 70% dei casi. Inoltre, la TLE è la forma di epilessia farmacoresistente più diffusa, con il 30-40% di pazienti non responsivi ai medicinali. In questi casi, la resezione chirurgica del fuoco epilettico risulta la prima opzione terapeutica; tuttavia, non tutti i pazienti farmacoresistenti sono candidati idonei per l’intervento e tra i pazienti operati non sempre si osserva una totale remissione delle crisi, sottolineando la necessità di esplorare trattamenti alternativi.
Il presente studio fa parte di HERMES, un progetto UE che mira a ripristinare funzionalmente i circuiti neuronali nel modello pilocarpina di TLE tramite bioibridi intelligenti, ottenuti combinando organoidi, impianti neuromorfici e intelligenza artificiale. Il CA3 ventrale (vCA3), un'area altamente epilettogena e danneggiata nella TLE, è stato selezionato come target del trattamento. Invece di impiantare organoidi intatti che avrebbero causato danni data la posizione profonda del vCA3 nel cervello, abbiamo generato organoidi in vivo iniettando cellule staminali neurali (NSC). I ratti epilettici sono stati sottoposti a chirurgia stereotassica per l’iniezione di acido ibotenico in vCA3, per eliminare le cellule epilettiche. Successivamente, le NSC, capaci di differenziarsi in neuroni eccitatori, inibitori e astrociti, sono state infuse da sole o con alginato, un biopolimero che in vitro crea un idrogel a supporto della sopravvivenza e maturazione cellulare. Inoltre abbiamo dimostrato che l’idrogel riduce la neuroinfiammazione in vivo modulando l’attivazione microgliale. Le analisi di immunofluorescenza hanno rivelato uno scarso differenziamento delle NSC a prescindere dalla presenza dell’idrogel. Per valutare il potenziale neurogenico delle NSC in vivo, sono state effettuate iniezioni bilaterali in ratti non epilettici sia in vCA3 che nel giro dentato dorsale (dDG), la nicchia naturale delle NSC. Pur osservando una maggiore sopravvivenza in dDG, le NSC non sono maturate e il loro numero si è ridotto significativamente nel tempo.
Nonostante i numerosi sforzi, la formazione di organoidi in vivo è risultata inefficace, determinando il passaggio da un approccio basato su bioibridi a un paradigma di neurostimolazione a circuito chiuso. Gli esperimenti sono stati condotti utilizzando un sistema di ingegneria neuromorfica sviluppato presso l’Università di Aarhus, che consente la stimolazione in base all’analisi in tempo reale di dati elettrografici, con l’obiettivo di prevenire le crisi. Due settimane dopo l’iniezione di pilocarpina, i ratti sono stati impiantati con un elettrodo per la stimolazione nel subiculum ventrale e con elettrodi di registrazione sia profondi che epidurali. Gli effetti della stimolazione sono stati valutati tramite gravità, frequenza e durata delle crisi in due ratti epilettici che hanno completato il protocollo. Nel ratto 1, la gravità delle crisi è diminuita drasticamente durante la stimolazione, con totale scomparsa di crisi convulsive. Tuttavia la frequenza è aumentata da 6,7 a 40,1 crisi al giorno, con una lieve riduzione della durata nella fase di washout. Il ratto 2 ha mostrato una riduzione della gravità solo nel washout, con crisi convulsive diminuite dal 100% al 66,9%, frequenza passata da 1 a 17,5 crisi al giorno e durata invariata. L’analisi dei fuochi epilettici ha evidenziato cambiamenti nelle zone di insorgenza delle crisi durante e dopo la stimolazione.
Questo lavoro evidenzia le sfide e il potenziale di terapie alternative per la TLE basate su cellule staminali e neurostimolazione avanzata, sottolineando la necessità di ulteriori perfezionamenti in termini di sopravvivenza cellulare e controllo delle crisi.Temporal lobe epilepsy (TLE) is the most frequently diagnosed form of epilepsy in adults, accounting for 50-70% of cases. Despite the number of anti-seizure medications available on the market, TLE represents the most prevalent form of drug-resistant epilepsy (DRE), with 30-40% of patients unresponsive to the pharmacological approach. In these cases, surgical resection of the epileptic focus represents the first-line therapeutic option; however, not all DRE patients are good candidates for surgery, and not all operated patients reach a seizure-free condition, suggesting the necessity to explore other treatment alternatives.
The present work is part of HERMES, an EU-funded project aimed at functionally repairing neuronal circuits in the pilocarpine model of TLE using intelligent biohybrids, obtained by the combination of organoids, neuromorphic implants, and artificial intelligence. Ventral CA3 (vCA3), a highly epileptogenic and damaged zone in TLE, has been selected as the target area of the treatment. Instead of implanting intact organoids that would cause massive damage due to the vCA3 location in the deepest part of the brain, we foresaw the in vivo generation of organoids starting from fluid pre-tissue and, therefore, injecting neural stem cells (NSCs). Subsequently, NSCs, capable of differentiating into excitatory and inhibitory neurons as well as into astrocytes, were infused either alone or in combination with alginate, a biopolymer shown in vitro to create a hydrogel scaffold that supports stem cell survival and maturation. In addition, we demonstrated that the hydrogel reduced neuroinflammation in vivo by modulating microglial activation. Immunofluorescence analyses revealed poor NSC differentiation, regardless of the presence of the hydrogel. To assess NSC neurogenic potential in vivo, bilateral injections were performed in non-epileptic rats in both vCA3 and dorsal dentate gyrus (dDG), the natural NSC niche. Although NSCs demonstrated enhanced survival in dDG, they did not mature, and their number declined significantly over time.
Despite extensive efforts, the organoid formation in vivo was ineffective, resulting in the transition from a biohybrid-based approach to a closed-loop neurostimulation paradigm. The experiments were conducted using a neuromorphic engineering system developed at Aarhus University that allows stimulation to be delivered when the system triggers it based on real-time analysis of electrographic data, to prevent seizures. Two weeks after pilocarpine injection, rats were implanted with one stimulating electrode in the ventral subiculum and multiple recording depth and epidural electrodes. Closed-loop stimulation effects were assessed by evaluating seizure severity, frequency, and duration in two epileptic rats completing the protocol. In rat 1, seizure severity decreased dramatically during stimulation, with convulsive seizures disappearing completely. Seizure frequency unexpectedly increased from 6.7 to 40.1 per day, with a slight reduction in duration during the washout phase. Rat 2 showed a reduction in seizure severity only during washout, with convulsive seizures decreasing from 100% to 66.9%, frequency rising from 1 to 17.5 seizures per day, while seizure duration remained unchanged. Analysis of epileptic foci revealed changes in the onset zones during and after stimulation.
Overall, this work highlights the challenges and potential of stem cell therapy and advanced neurostimulation to develop alternative treatments for TLE, emphasizing the need for further refinements in terms of cellular survival and seizure control
Partial and Lamb waves in non-local elasticity with kernel modification
No full textWe study dispersion of Rayleigh-Lamb (R-L) waves in an infinite isotropic strip within the theory of non-local elasticity with kernel modification. Within this approach, the set of constitutive boundary conditions (CBCs) embedded in the attenuation functions contain the set of natural boundary conditions (BCs) of the problem and this feature, besides avoiding nonphysical BCs, warrants that the problem is well-posed. We show that, contrast to local elasticity, the dispersion equation emerges from imposing the equations of motion, given that the BCs are automatically satisfied by the very choice of the attenuation functions. Similarly to local elasticity, the problem naturally decouples into symmetric and anti-symmetric partial modes, although this feature is not obvious here and crucially depends on certain symmetry properties of the kernels. We prove that symmetric and anti-symmetric kernels may be constructed directly, to avoid solving the full problem, and we show how these kernels relate to the original. Explicit dispersion relations for symmetric and anti-symmetric partial waves are obtained, that reveal the size-dependent deviation from the classical predictions. Overall, results reproduce the general features already observed in local elasticity, such as the convergence of the fundamental modes the Rayleigh speed and of the higher modes to the bulk wave speeds, although these are no longer constants. Yet, both fundamental modes, and especially the symmetric one, significantly depart from the local theory, which fact has important consequences on the corresponding asymptotic model for non-local beams
Standardizing microbiome research: interlaboratory validation of SOPs for sample preparation and DNA extraction from food and environmental ecosystems
No full textMicrobiome research has expanded rapidly, however, lack of standardized and validated protocols for microbiome sampling and DNA extraction has hindered the reproducibility and comparability of studies. The SUS-MIRRI.IT project aimed to prepare and validate Standard Operating Procedures (SOPs) for microbiome analysis across diverse ecosystems, including fermented foods, soils, waters, and more. To validate these protocols, 15 Italian research units (RUs) participated in an interlaboratory trial on 120 samples (liquid and solid fermented foods, waters, and soils). Metataxonomic sequencing was performed using 16S rRNA gene amplicon sequencing to assess the reproducibility of the protocols. The interlaboratory trial involved distributing homogenized samples to participating RUs and evaluating performance both between and within RUs. This was done by comparing results obtained from DNA extraction and amplicon-based sequencing
DOLFIN: Balancing Stability and Plasticity in Federated Continual Learning
No full textFederated continual learning (FCL) enables models to learn new tasks across multiple distributed clients, protecting privacy and without forgetting previously acquired knowledge. However, current methods face challenges balancing performance, privacy preservation, and communication efficiency. We introduce a Distributed Online LoRA for Federated INcremental learning methodDOLFIN, a novel approach combining Vision Transformers with low-rank adapters designed to efficiently and stably learn new tasks in federated environments. Our method leverages LoRA for minimal communication overhead and incorporates Dual Gradient Projection Memory (DualGPM) to prevent forgetting. Evaluated on CIFAR-100, ImageNet-R, ImageNet-A, and CUB-200 under two Dirichlet heterogeneity settings,DOLFINconsistently surpasses six strong baselines in final average accuracy while matching their memory footprint. Orthogonal low-rank adapters offer an effective and scalable solution for privacy-preserving continual learning in federated settings