Marina Grisoli

Background and Objective: Deep learning (DL) is increasingly used to predict overall survival (OS) in glioblastoma (GB), yet it rarely outperforms conventional machine learning (ML), and models are compared only through aggregate metrics that do not show which patients they mispredict.Methods: We evaluated a clinical and a radiomic Cox model and four pretrained DL architectures (two brain-specific, two general medical) on an institutional cohort (n=297) and the public UCSF-PDGM cohort (n=350). Discrimination was measured with the concordance index (C-index) under within-cohort cross-validation, cross-cohort comparison of model rankings, and bidirectional external validation. We further introduced a censoring-aware error profiling procedure based on Martingale residuals, which extends to survival prediction the analysis of hidden failure subgroups developed for classification, and characterized mispredicted patients in the model embeddings and by clinical and tumor characteristics.Results: The C-index was 0.60–0.68 for all models; ML performed as well as DL, and general medical pretraining matched brain-specific pretraining. Under external validation, the C-index fell to 0.56–0.63 in both directions. DL models agreed with each other no more than with the ML baselines on which patients they mispredicted, and the error groups were not separable inthe embedding space. In the institutional cohort, underestimated-risk patients had larger T2-FLAIR tumors and were more often MGMT-unmethylated (q < 0.05); this profile was not found in the public cohort. Saliency maps were spatially diffuse, with tumor activation only 1.2 times that of the surrounding brain.Conclusions: The proposed framework compares survival models, quantifies how much DL adds over traditional ML, and adds transparency about which patients are mispredicted and which characteristics are associated with the error. The procedure is model-agnostic, applies to any survival model, and makes model comparison informative for individual patients rather than only on average.

Background: Spinocerebellar ataxia type 1 (SCA1) is a rare, autosomal dominant neurodegenerative disorder characterized by progressive cerebellar and brainstem degeneration. Previous studies have shown that spinal cord atrophy is also a key aspect of SCA1 neuropathology. Magnetic resonance imaging (MRI)‐based spinal cord measurements could, therefore, serve as staging or monitoring biomarkers. However, previous findings were limited to cross‐sectional analyses, and longitudinal changes remain unexplored. Objectives: This study investigates both cross‐sectional and longitudinal cervical spinal cord alterations in SCA1 mutation carriers compared to healthy controls, evaluating the utility of this biomarker. Methods: Baseline and 1‐year MRI changes were assessed in 40 controls, 16 preataxic and 58 symptomatic SCA1 mutation carriers. T1‐weighed images were processed with FreeSurfer and the Spinal Cord Toolbox. We related Z‐scores to disease duration, calculated standardized response means, and analyzed clinico‐genetic associations using a linear mixed model. Results: The three groups differed significantly in cross‐sectional area (CSA) at all levels at baseline. Over time, CSA at levels C1 and C2 decreased in the preataxic and symptomatic groups compared to controls. Preataxic carriers already showed pronounced spinal cord atrophy, whereas pontine changes emerged later in the disease course. Standardized response means were highest for CSA at C2 in preataxic stage, whereas this was pontine volume at symptomatic stage, indicating region‐specific biomarkers across disease stages. Conclusion: Cervical spinal cord atrophy is an early and progressive feature of SCA1, detectable before clinical onset, providing a promising imaging biomarker for early disease stages. Our findings suggest a caudal to rostral progression of atrophy in SCA1. © 2026 The Author(s). Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.

Cerebellar ataxias are a rare group of disorders manifesting with motor incoordination and cognitive-affective deficits of variable severity. Although neurogenetic has revealed multiple mutations, the study of ataxias still relies on clinical evaluation, while the underlying neural network changes remain unclear. It has been argued that the less severe symptoms in congenital (like Joubert syndrome, JS) than in slowly progressive (SP) ataxias reflect a different interplay of alteration and compensation but direct evidence is still lacking. Moreover, it is unclear why, in front of a wide heterogeneity of molecular alterations, SPs show common clinical symptoms. To address these questions, we created brain digital twins for each participant by combining volumetry, graph theory analysis of structural and functional connectivity, and dynamical simulations using the virtual brain. We studied 8 JS (3 females, 21±6years), 8 SP (3 females, 20±5years) and 11 healthy controls (HC; 5 females, 21±years).Volumetry quantified atrophy, graph metrics (centrality, segregation and integration) characterized topology, and neural dynamical simulations estimated excitation/inhibition balance, providing anatomo-physiological parameters within the somatomotor (SMN) and ventral attention (VAN) networks. Anatomo-physiological parameters were correlated with clinical/neuropsychological scores, and unsupervised clustering was applied to assess whether network features can discriminate between JS and SP beyond clinical classification. MRI morphometry confirmed selective vermis reduction in JS and a widespread cerebellar atrophy in SP compared to HC. In both ataxia groups, SMN and VAN showed reduced volume and structural connectivity but with different patterns of topological and dynamical alterations. In the SMN of SP, reduced centrality and excitation/inhibition balance depressed information transfer through the network. In the VAN of JS, reduced centrality, segregation, and integration, were detrimental but coexisted with a higher number of functional core nodes and an increased large-scale excitatory coupling, supporting compensatory reorganisation in extracerebellar nodes. Clustering confirmed that SMN better differentiates SP, whereas VAN better clusters JS. Importantly, anatomo-physiological parameters of network volume, topology, and dynamics correlated with patients' motor and cognitive performance. In conclusion, primary cerebellar damage secondarily impacts large-scale brain networks, altered in both ataxia groups but compensated only in JS. Similar clinical symptoms in SP reflects the similarity of network changes, while differential involvement of SMN and VAN in JS and SP reflects the connectivity pattern of the lesioned areas inside these large-scale brain circuits. Importantly, anatomo-physiological parameters are sufficient to explain individual motor and cognitive performance, offering a basis for improved patient profiling and personalized therapies.

Background Emotional lability (EL), also known as pathological laughter and crying, is a common but understudied symptom in motor neuron diseases (MND): amyotrophic lateral sclerosis and primary lateral sclerosis. This study aimed to investigate the prevalence of EL in MND and to explore the independent frequency components of laughter and crying in relation to motor, cognitive, neuropsychiatric, and neuroimaging factors. Methods A total of 198 incident MND patients were enrolled. The Centre of Neurological Study-Lability Scale was used to measure EL. Associations between EL and motor function, mood, neuropsychological variables, and structural MRI were examined, with cortical thinning measured on a subset of 48 patients. Results EL was identified in 36% of patients showing more severe motor functional disabilities, heightened depressive and anxiety symptoms and behavioral changes than those without EL. Women exhibited more severe EL and altered mood with frequent crying episodes than men. EL was strongly correlated with bulbar involvement. Crying episodes were associated with mood disorders, while laughter correlated with disinhibition and emotional regulation difficulties. EL had a specific association with the thinning of frontal regions, including the right pars orbitalis, which was also linked to altered emotional and behavioral regulation. Conclusion These findings underscore the role of corticobulbar and frontal pathways in EL pathophysiology. The study highlights the distinct mechanisms underlying pathological crying and laughter and their independency from general cognitive decline. It emphasizes the need for clinicians to recognize EL as an independent symptom, necessitating targeted management strategies to improve patient outcomes and support caregivers.

Background: Magnetic resonance‐guided focused ultrasound (MRgFUS) targeting the thalamic ventral intermediate nucleus (VIM) is an innovative treatment for drug‐refractory essential tremor (ET). The relationship between lesion characteristics, dentate‐rubro‐thalamic‐tract (DRTT) involvement and clinical benefit remains unclear.Objectives: To investigate whether clinical outcome is related to lesion volume and/or its overlap with the DRTT. To compare the reliability of probabilistic versus deterministic tractography in reconstructing the DRTT and improving VIM targeting.Methods: Forty ET patients who underwent MRgFUS thalamotomy between 2019 and 2022 were retrospectively analyzed. Clinical outcomes and adverse effects were recorded at 1/6/12 months after the procedure. The DRTT was generated using deterministic and probabilistic tractography on preoperative diffusion‐tensor 3 T‐images and location and volume of the lesion were calculated.Results: Probabilistic tractography identified both decussating (d‐DRTT) and non‐decussating (nd‐DRTT) components of the DRTT, whereas the deterministic approach only identified one component overlapping with the nd‐DRTT. Despite the lesions predominantly intersecting the medial portion of the d‐DRTT, with a significantly greater overlap in responder patients, we observed only a non‐significant correlation between tremor improvement and increased d‐DRTT‐lesion overlap (r = 0.22, P = 0.20). The lesion volume demonstrated a significant positive correlation with clinical improvement at 1‐day MRI (r = 0.42, P < 0.01).Conclusion: Variability in the reconstructed DRTT position relative to the lesion center of mass, even among good responders, suggests that this fiber bundle is unlikely to be considered the sole target for a successful MRgFUS thalamotomy in ET. Indirect individualized targeting allows for more precise and reproducible identification of actual treatment coordinates than the direct method.

Objective: Spinocerebellar ataxia type 2 (SCA2) is a rare, inherited neurodegenerative disease characterised by progressive deterioration in both motor coordination and cognitive function. Atrophy of the cerebellum, brainstem, and spinal cord are core features of SCA2, however the evolution and pattern of whole-brain atrophy in SCA2 remain unclear. We undertook a multi-site, structural magnetic resonance imaging (MRI) study to comprehensively characterize the neurodegeneration profile of SCA2. Methods: Voxel-based morphometry analyses of 110 participants with SCA2 and 128 controls were undertaken to assess groupwise differences in whole-brain volume. Correlations with clinical severity and genotype, and cross-sectional profiling of atrophy patterns at different disease stages, were also performed. Results: Atrophy in SCA2 relative to controls was greatest (Cohen’s d >2.5) in the cerebellar white matter (WM), middle cerebellar peduncle, pons, and corticospinal tract. Very large effects ( d >1.5) were also evident in the superior cerebellar, inferior cerebellar, and cerebral peduncles. In cerebellar grey matter (GM), large effects ( d >0.8) mapped to areas related to both motor coordination and cognitive tasks. Strong correlations (| r |>0.4) between volume and disease severity largely mirrored these groupwise outcomes. Stratification by disease severity showed a degeneration pattern beginning in cerebellar and pontine WM in pre-clinical subjects; spreading to the cerebellar GM and cerebro-cerebellar/corticospinal WM tracts; then finally involving the thalamus, striatum, and cortex in severe stages. Interpretation: The magnitude and pattern of brain atrophy evolves over the course of SCA2, with widespread, non-uniform involvement across the brainstem, cerebellar tracts, and cerebellar cortex; and late involvement of the cerebral cortex and striatum.

Amide Proton Transfer-weighted (APTw) imaging is a molecular MRI technique used to quantify protein concentrations in gliomas, which have heterogeneous components with varying cellularity and metabolic activity. This study aimed to assess the correlation between the component-specific APT signal of the neoplasm and WHO grade, molecular profile and survival status. Sixty-one patients with adult-type diffuse gliomas were retrospectively analyzed. APT values were semi-automatically extracted from tumor solid and, whenever present, necrotic components. APT values were compared between groups stratified by WHO grade, IDH-mutation, MGMT promoter methylation and 1- and 2-year survival status using Wilcoxon rank-sum test, adjusting for multiple comparisons. Overall survival (OS) was analyzed in the subgroup of 48 patients with grade 4 tumors using Cox proportional-hazards models. Random-effects models were used to assess inter-subject heterogeneity of the mean APT values in each tumor component. APT values of the solid component significantly differed between patients with grades 2–3 and 4 tumors (mean 1.58 ± 0.50 vs. 2.04 ± 0.56, p = 0.028) and correlated with OS after 1 year (1.81 ± 0.58 in survivors vs. 2.17 ± 0.51 in deceased patients, p = 0.030). APT values did not differ by IDH-mutation, MGMT methylation, and 2-year survival status. Within grade 4 glioma patients, higher APT kurtosis of the solid component was a negative prognostic factor (hazard ratio = 1.60, p = 0.040). Mean APT values of the necrosis showed high inter-subject variability, although most necrotic tumors were grade 4 and IDH wildtype. In conclusion, APTw imaging in the solid component provided metrics associated with glioma grade and survival status but showed weak correlation with IDH-mutation and MGMT promoter methylation status, in contrast to previous works. Further research is needed to understand APT signal variability within the necrotic component of high-grade gliomas.

This study aimed to evaluate the intensity of head pain associated with sonication. In particular, we were interested in identifying predictive factors of high-intensity head pain as it could be a source of interference with the procedure. Furthermore, we explored eventual dissociative or post-traumatic consequences related to high-intensity pain during MRgFUS.

Background: Magnetic resonance‐guided focused ultrasound (MRgFUS) thalamotomy is increasingly used to treat drug‐resistant essential tremor (ET). Data on MRgFUS thalamotomy in dystonic tremor (DT) are anecdotal.Objectives: To investigate efficacy, safety, and differences in target coordinates of MRgFUS thalamotomy in DT versus ET.Methods: Ten patients with DT and 35 with ET who consecutively underwent MRgFUS thalamotomy were followed for 12 months. Although in both groups the initial surgical planning coordinates corresponded to the ventralis intermediate (Vim), the final target could be modified intraoperatively based on clinical response.Results: Tremor significantly improved in both groups. The thalamic lesion was significantly more anterior in DT than ET. Considering both ET and DT groups, the more anterior the lesion, the lower the odds ratio for adverse events.Conclusions: MRgFUS thalamotomy is safe and effective in DT and ET. Compared to classical Vim coordinates used for ET, more anterior targeting should be considered for DT.