FDG-PET markers of heterogeneity and different risk of progression in amnestic MCI.

INTRODUCTION: Amnestic mild cognitive impairment (aMCI) is emerging as a heterogeneous condition. METHODS: We looked at a cohort of N = 207 aMCI subjects, with baseline fluorodeoxyglucose positron emission tomography (FDG-PET), T1 magnetic resonance imaging, cerebrospinal fluid (CSF), apolipoprotein E (APOE), and neuropsychological assessment. An algorithm based on FDG-PET hypometabolism classified each subject into subtypes, then compared biomarker measures and clinical progression. RESULTS: Three subtypes emerged: hippocampal sparing-cortical hypometabolism, associated with younger age and the highest level of Alzheimer's disease (AD)-CSF pathology; hippocampal/cortical hypometabolism, associated with a high percentage of APOE ε3/ε4 or ε4/ε4 carriers; medial-temporal hypometabolism, characterized by older age, the lowest AD-CSF pathology, the most severe hippocampal atrophy, and a benign course. Within the whole cohort, the severity of temporo-parietal hypometabolism, correlated with AD-CSF pathology and marked the rate of progression of cognitive decline. DISCUSSION: FDG-PET can distinguish clinically comparable aMCI at single-subject level with different risk of progression to AD dementia or stability. The obtained results can be useful for the optimization of pharmacological trials and automated-classification models. HIGHLIGHTS: Algorithm based on FDG-PET hypometabolism demonstrates distinct subtypes across aMCI; Three different subtypes show heterogeneous biological profiles and risk of progression; The cortical hypometabolism is associated with AD pathology and cognitive decline; MTL hypometabolism is associated with the lowest conversion rate and CSF-AD pathology.

Italian, European, and international neuroinformatics efforts: An overview.

Neuroinformatics is a research field that focusses on software tools capable of identifying, analysing, modelling, organising and sharing multiscale neuroscience data. Neuroinformatics has exploded in the last two decades with the emergence of the Big Data phenomenon, characterised by the so-called 3Vs (volume, velocity and variety), which provided neuroscientists with an improved ability to acquire and process data faster and more cheaply thanks to technical improvements in clinical, genomic and radiological technologies. This situation has led to a 'data deluge', as neuroscientists can routinely collect more study data in a few days than they could in a year just a decade ago. To address this phenomenon, several neuroimaging-focussed neuroinformatics platforms have emerged, funded by national or transnational agencies, with the following goals: (i) development of tools for archiving and organising analytical data (XNAT, REDCap and LabKey); (ii) development of data-driven models evolving from reductionist approaches to multidimensional models (RIN, IVN, HBD, EuroPOND, E-DADS and GAAIN BRAIN); and (iii) development of e-infrastructures to provide sufficient computational power and storage resources (neuGRID, HBP-EBRAINS, LONI and CONP). Although the scenario is still fragmented, there are technological and economical attempts at both national and international levels to introduce high standards for open and Findable, Accessible, Interoperable and Reusable (FAIR) neuroscience worldwide.

Localizing in-domain adaptation of transformer-based biomedical language models.

In the era of digital healthcare, the huge volumes of textual information generated every day in hospitals constitute an essential but underused asset that could be exploited with task-specific, fine-tuned biomedical language representation models, improving patient care and management. For such specialized domains, previous research has shown that fine-tuning models stemming from broad-coverage checkpoints can largely benefit additional training rounds over large-scale in-domain resources. However, these resources are often unreachable for less-resourced languages like Italian, preventing local medical institutions to employ in-domain adaptation. In order to reduce this gap, our work investigates two accessible approaches to derive biomedical language models in languages other than English, taking Italian as a concrete use-case: one based on neural machine translation of English resources, favoring quantity over quality; the other based on a high-grade, narrow-scoped corpus natively written in Italian, thus preferring quality over quantity. Our study shows that data quantity is a harder constraint than data quality for biomedical adaptation, but the concatenation of high-quality data can improve model performance even when dealing with relatively size-limited corpora. The models published from our investigations have the potential to unlock important research opportunities for Italian hospitals and academia. Finally, the set of lessons learned from the study constitutes valuable insights towards a solution to build biomedical language models that are generalizable to other less-resourced languages and different domain settings.

Advancing Italian biomedical information extraction with transformers-based models: Methodological insights and multicenter practical application.

The introduction of computerized medical records in hospitals has reduced burdensome activities like manual writing and information fetching. However, the data contained in medical records are still far underutilized, primarily because extracting data from unstructured textual medical records takes time and effort. Information Extraction, a subfield of Natural Language Processing, can help clinical practitioners overcome this limitation by using automated text-mining pipelines. In this work, we created the first Italian neuropsychiatric Named Entity Recognition dataset, PsyNIT, and used it to develop a Transformers-based model. Moreover, we collected and leveraged three external independent datasets to implement an effective multicenter model, with overall F1-score 84.77 %, Precision 83.16 %, Recall 86.44 %. The lessons learned are: (i) the crucial role of a consistent annotation process and (ii) a fine-tuning strategy that combines classical methods with a "low-resource" approach. This allowed us to establish methodological guidelines that pave the way for Natural Language Processing studies in less-resourced languages.

The impact of automated hippocampal volumetry on diagnostic confidence in patients with suspected Alzheimer's disease: A European Alzheimer's Disease Consortium study.

INTRODUCTION: Hippocampal volume is a core biomarker of Alzheimer's disease (AD). However, its contribution over the standard diagnostic workup is unclear. METHODS: Three hundred fifty-six patients, under clinical evaluation for cognitive impairment, with suspected AD and Mini-Mental State Examination ≥20, were recruited across 17 European memory clinics. After the traditional diagnostic workup, diagnostic confidence of AD pathology (DCAD) was estimated by the physicians in charge. The latter were provided with the results of automated hippocampal volumetry in standardized format and DCAD was reassessed. RESULTS: An increment of one interquartile range in hippocampal volume was associated with a mean change of DCAD of -8.0% (95% credible interval: [-11.5, -5.0]). Automated hippocampal volumetry showed a statistically significant impact on DCAD beyond the contributions of neuropsychology, 18F-fluorodeoxyglucose positron emission tomography/single-photon emission computed tomography, and cerebrospinal fluid markers (-8.5, CrI: [-11.5, -5.6]; -14.1, CrI: [-19.3, -8.8]; -10.6, CrI: [-14.6, -6.1], respectively). DISCUSSION: There is a measurable effect of hippocampal volume on DCAD even when used on top of the traditional diagnostic workup.

Integrating longitudinal information in hippocampal volume measurements for the early detection of Alzheimer's disease.

BACKGROUND: Structural MRI measures for monitoring Alzheimer's Disease (AD) progression are becoming instrumental in the clinical practice, and more so in the context of longitudinal studies. This investigation addresses the impact of four image analysis approaches on the longitudinal performance of the hippocampal volume. METHODS: We present a hippocampal segmentation algorithm and validate it on a gold-standard manual tracing database. We segmented 460 subjects from ADNI, each subject having been scanned twice at baseline, 12-month and 24month follow-up scan (1.5T, T1 MRI). We used the bilateral hippocampal volume v and its variation, measured as the annualized volume change Λ=δv/year(mm(3)/y). Four processing approaches with different complexity are compared to maximize the longitudinal information, and they are tested for cohort discrimination ability. Reference cohorts are Controls vs. Alzheimer's Disease (CTRL/AD) and CTRL vs. Mild Cognitive Impairment who subsequently progressed to AD dementia (CTRL/MCI-co). We discuss the conditions on v and the added value of Λ in discriminating subjects. RESULTS: The age-corrected bilateral annualized atrophy rate (%/year) were: -1.6 (0.6) for CTRL, -2.2 (1.0) for MCI-nc, -3.2 (1.2) for MCI-co and -4.0 (1.5) for AD. Combined (v, Λ) discrimination ability gave an Area under the ROC curve (auc)=0.93 for CTRL vs AD and auc=0.88 for CTRL vs MCI-co. CONCLUSIONS: Longitudinal volume measurements can provide meaningful clinical insight and added value with respect to the baseline provided the analysis procedure embeds the longitudinal information.

Operationalizing protocol differences for EADC-ADNI manual hippocampal segmentation.

BACKGROUND: Hippocampal volumetry on magnetic resonance imaging is recognized as an Alzheimer's disease (AD) biomarker, and manual segmentation is the gold standard for measurement. However, a standard procedure is lacking. We operationalize and quantitate landmark differences to help a Delphi panel converge on a set of landmarks. METHODS: One hundred percent of anatomic landmark variability across 12 different protocols for manual segmentation was reduced into four segmentation units (the minimum hippocampus, the alveus/fimbria, the tail, and the subiculum), which were segmented on magnetic resonance images by expert raters to estimate reliability and AD-related atrophy. RESULTS: Intra- and interrater reliability were more than 0.96 and 0.92, respectively, except for the alveus/fimbria, which were 0.86 and 0.77, respectively. Of all AD-related atrophy, the minimum hippocampus contributed to 67%; tail, 24%; alveus/fimbria, 4%; and the subiculum, 5%. CONCLUSIONS: Anatomic landmark variability in available protocols can be reduced to four discrete and measurable segmentation units. Their quantitative assessment will help a Delphi panel to define a set of landmarks for a harmonized protocol.

Training labels for hippocampal segmentation based on the EADC-ADNI harmonized hippocampal protocol.

BACKGROUND: The European Alzheimer's Disease Consortium and Alzheimer's Disease Neuroimaging Initiative (ADNI) Harmonized Protocol (HarP) is a Delphi definition of manual hippocampal segmentation from magnetic resonance imaging (MRI) that can be used as the standard of truth to train new tracers, and to validate automated segmentation algorithms. Training requires large and representative data sets of segmented hippocampi. This work aims to produce a set of HarP labels for the proper training and certification of tracers and algorithms. METHODS: Sixty-eight 1.5 T and 67 3 T volumetric structural ADNI scans from different subjects, balanced by age, medial temporal atrophy, and scanner manufacturer, were segmented by five qualified HarP tracers whose absolute interrater intraclass correlation coefficients were 0.953 and 0.975 (left and right). Labels were validated as HarP compliant through centralized quality check and correction. RESULTS: Hippocampal volumes (mm(3)) were as follows: controls: left = 3060 (standard deviation [SD], 502), right = 3120 (SD, 897); mild cognitive impairment (MCI): left = 2596 (SD, 447), right = 2686 (SD, 473); and Alzheimer's disease (AD): left = 2301 (SD, 492), right = 2445 (SD, 525). Volumes significantly correlated with atrophy severity at Scheltens' scale (Spearman's ρ = <-0.468, P = <.0005). Cerebrospinal fluid spaces (mm(3)) were as follows: controls: left = 23 (32), right = 25 (25); MCI: left = 15 (13), right = 22 (16); and AD: left = 11 (13), right = 20 (25). Five subjects (3.7%) presented with unusual anatomy. CONCLUSIONS: This work provides reference hippocampal labels for the training and certification of automated segmentation algorithms. The publicly released labels will allow the widespread implementation of the standard segmentation protocol.

Delphi definition of the EADC-ADNI Harmonized Protocol for hippocampal segmentation on magnetic resonance.

BACKGROUND: This study aimed to have international experts converge on a harmonized definition of whole hippocampus boundaries and segmentation procedures, to define standard operating procedures for magnetic resonance (MR)-based manual hippocampal segmentation. METHODS: The panel received a questionnaire regarding whole hippocampus boundaries and segmentation procedures. Quantitative information was supplied to allow evidence-based answers. A recursive and anonymous Delphi procedure was used to achieve convergence. Significance of agreement among panelists was assessed by exact probability on Fisher's and binomial tests. RESULTS: Agreement was significant on the inclusion of alveus/fimbria (P = .021), whole hippocampal tail (P = .013), medial border of the body according to visible morphology (P = .0006), and on this combined set of features (P = .001). This definition captures 100% of hippocampal tissue, 100% of Alzheimer's disease-related atrophy, and demonstrated good reliability on preliminary intrarater (0.98) and inter-rater (0.94) estimates. DISCUSSION: Consensus was achieved among international experts with respect to hippocampal segmentation using MR resulting in a harmonized segmentation protocol.

Resting state cortical electroencephalographic rhythms are related to gray matter volume in subjects with mild cognitive impairment and Alzheimer's disease.

Cortical gray matter volume and resting state cortical electroencephalographic rhythms are typically abnormal in subjects with amnesic mild cognitive impairment (MCI) and Alzheimer's disease (AD). Here we tested the hypothesis that in amnesic MCI and AD subjects, abnormalities of EEG rhythms are a functional reflection of cortical atrophy across the disease. Eyes-closed resting state EEG data were recorded in 57 healthy elderly (Nold), 102 amnesic MCI, and 108 AD patients. Cortical gray matter volume was indexed by magnetic resonance imaging recorded in the MCI and AD subjects according to Alzheimer's disease neuroimaging initiative project (http://www.adni-info.org/). EEG rhythms of interest were delta (2-4 Hz), theta (4-8 Hz), alpha1 (8-10.5 Hz), alpha2 (10.5-13 Hz), beta1 (13-20 Hz), beta2 (20-30 Hz), and gamma (30-40 Hz). These rhythms were indexed by LORETA. Compared with the Nold, the MCI showed a decrease in amplitude of alpha 1 sources. With respect to the Nold and MCI, the AD showed an amplitude increase of delta sources, along with a strong amplitude reduction of alpha 1 sources. In the MCI and AD subjects as a whole group, the lower the cortical gray matter volume, the higher the delta sources, the lower the alpha 1 sources. The better the score to cognitive tests the higher the gray matter volume, the lower the pathological delta sources, and the higher the alpha sources. These results suggest that in amnesic MCI and AD subjects, abnormalities of resting state cortical EEG rhythms are not epiphenomena but are strictly related to neurodegeneration (atrophy of cortical gray matter) and cognition.

Differential diagnosis of neurodegenerative dementias with the explainable MRI based machine learning algorithm MUQUBIA.

Biomarker-based differential diagnosis of the most common forms of dementia is becoming increasingly important. Machine learning (ML) may be able to address this challenge. The aim of this study was to develop and interpret a ML algorithm capable of differentiating Alzheimer's dementia, frontotemporal dementia, dementia with Lewy bodies and cognitively normal control subjects based on sociodemographic, clinical, and magnetic resonance imaging (MRI) variables. 506 subjects from 5 databases were included. MRI images were processed with FreeSurfer, LPA, and TRACULA to obtain brain volumes and thicknesses, white matter lesions and diffusion metrics. MRI metrics were used in conjunction with clinical and demographic data to perform differential diagnosis based on a Support Vector Machine model called MUQUBIA (Multimodal Quantification of Brain whIte matter biomArkers). Age, gender, Clinical Dementia Rating (CDR) Dementia Staging Instrument, and 19 imaging features formed the best set of discriminative features. The predictive model performed with an overall Area Under the Curve of 98%, high overall precision (88%), recall (88%), and F1 scores (88%) in the test group, and good Label Ranking Average Precision score (0.95) in a subset of neuropathologically assessed patients. The results of MUQUBIA were explained by the SHapley Additive exPlanations (SHAP) method. The MUQUBIA algorithm successfully classified various dementias with good performance using cost-effective clinical and MRI information, and with independent validation, has the potential to assist physicians in their clinical diagnosis.

Normative values of the topological metrics of the structural connectome: A multi-site reproducibility study across the Italian Neuroscience network.

PURPOSE: The use of topological metrics to derive quantitative descriptors from structural connectomes is receiving increasing attention but deserves specific studies to investigate their reproducibility and variability in the clinical context. This work exploits the harmonization of diffusion-weighted acquisition for neuroimaging data performed by the Italian Neuroscience and Neurorehabilitation Network initiative to obtain normative values of topological metrics and to investigate their reproducibility and variability across centers. METHODS: Different topological metrics, at global and local level, were calculated on multishell diffusion-weighted data acquired at high-field (e.g. 3 T) Magnetic Resonance Imaging scanners in 13 different centers, following the harmonization of the acquisition protocol, on young and healthy adults. A "traveling brains" dataset acquired on a subgroup of subjects at 3 different centers was also analyzed as reference data. All data were processed following a common processing pipeline that includes data pre-processing, tractography, generation of structural connectomes and calculation of graph-based metrics. The results were evaluated both with statistical analysis of variability and consistency among sites with the traveling brains range. In addition, inter-site reproducibility was assessed in terms of intra-class correlation variability. RESULTS: The results show an inter-center and inter-subject variability of <10%, except for "clustering coefficient" (variability of 30%). Statistical analysis identifies significant differences among sites, as expected given the wide range of scanners' hardware. CONCLUSIONS: The results show low variability of connectivity topological metrics across sites running a harmonised protocol.

Virtual brain simulations reveal network-specific parameters in neurodegenerative dementias.

Introduction: Neural circuit alterations lay at the core of brain physiopathology, and yet are hard to unveil in living subjects. The Virtual Brain (TVB) modeling, by exploiting structural and functional magnetic resonance imaging (MRI), yields mesoscopic parameters of connectivity and synaptic transmission. Methods: We used TVB to simulate brain networks, which are key for human brain function, in Alzheimer's disease (AD) and frontotemporal dementia (FTD) patients, whose connectivity and synaptic parameters remain largely unknown; we then compared them to healthy controls, to reveal novel in vivo pathological hallmarks. Results: The pattern of simulated parameter differed between AD and FTD, shedding light on disease-specific alterations in brain networks. Individual subjects displayed subtle differences in network parameter patterns that significantly correlated with their individual neuropsychological, clinical, and pharmacological profiles. Discussion: These TVB simulations, by informing about a new personalized set of networks parameters, open new perspectives for understanding dementias mechanisms and design personalized therapeutic approaches.

Natural language processing in clinical neuroscience and psychiatry: A review.

Natural language processing (NLP) is rapidly becoming an important topic in the medical community. The ability to automatically analyze any type of medical document could be the key factor to fully exploit the data it contains. Cutting-edge artificial intelligence (AI) architectures, particularly machine learning and deep learning, have begun to be applied to this topic and have yielded promising results. We conducted a literature search for 1,024 papers that used NLP technology in neuroscience and psychiatry from 2010 to early 2022. After a selection process, 115 papers were evaluated. Each publication was classified into one of three categories: information extraction, classification, and data inference. Automated understanding of clinical reports in electronic health records has the potential to improve healthcare delivery. Overall, the performance of NLP applications is high, with an average F1-score and AUC above 85%. We also derived a composite measure in the form of Z-scores to better compare the performance of NLP models and their different classes as a whole. No statistical differences were found in the unbiased comparison. Strong asymmetry between English and non-English models, difficulty in obtaining high-quality annotated data, and train biases causing low generalizability are the main limitations. This review suggests that NLP could be an effective tool to help clinicians gain insights from medical reports, clinical research forms, and more, making NLP an effective tool to improve the quality of healthcare services.

MRI data quality assessment for the RIN - Neuroimaging Network using the ACR phantoms.

PURPOSE: Generating big-data is becoming imperative with the advent of machine learning. RIN-Neuroimaging Network addresses this need by developing harmonized protocols for multisite studies to identify quantitative MRI (qMRI) biomarkers for neurological diseases. In this context, image quality control (QC) is essential. Here, we present methods and results of how the RIN performs intra- and inter-site reproducibility of geometrical and image contrast parameters, demonstrating the relevance of such QC practice. METHODS: American College of Radiology (ACR) large and small phantoms were selected. Eighteen sites were equipped with a 3T scanner that differed by vendor, hardware/software versions, and receiver coils. The standard ACR protocol was optimized (in-plane voxel, post-processing filters, receiver bandwidth) and repeated monthly. Uniformity, ghosting, geometric accuracy, ellipse's ratio, slice thickness, and high-contrast detectability tests were performed using an automatic QC script. RESULTS: Measures were mostly within the ACR tolerance ranges for both T1- and T2-weighted acquisitions, for all scanners, regardless of vendor, coil, and signal transmission chain type. All measurements showed good reproducibility over time. Uniformity and slice thickness failed at some sites. Scanners that upgraded the signal transmission chain showed a decrease in geometric distortion along the slice encoding direction. Inter-vendor differences were observed in uniformity and geometric measurements along the slice encoding direction (i.e. ellipse's ratio). CONCLUSIONS: Use of the ACR phantoms highlighted issues that triggered interventions to correct performance at some sites and to improve the longitudinal stability of the scanners. This is relevant for establishing precision levels for future multisite studies of qMRI biomarkers.

Quantitative MRI Harmonization to Maximize Clinical Impact: The RIN-Neuroimaging Network.

Neuroimaging studies often lack reproducibility, one of the cardinal features of the scientific method. Multisite collaboration initiatives increase sample size and limit methodological flexibility, therefore providing the foundation for increased statistical power and generalizable results. However, multisite collaborative initiatives are inherently limited by hardware, software, and pulse and sequence design heterogeneities of both clinical and preclinical MRI scanners and the lack of benchmark for acquisition protocols, data analysis, and data sharing. We present the overarching vision that yielded to the constitution of RIN-Neuroimaging Network, a national consortium dedicated to identifying disease and subject-specific in-vivo neuroimaging biomarkers of diverse neurological and neuropsychiatric conditions. This ambitious goal needs efforts toward increasing the diagnostic and prognostic power of advanced MRI data. To this aim, 23 Italian Scientific Institutes of Hospitalization and Care (IRCCS), with technological and clinical specialization in the neurological and neuroimaging field, have gathered together. Each IRCCS is equipped with high- or ultra-high field MRI scanners (i.e., ≥3T) for clinical or preclinical research or has established expertise in MRI data analysis and infrastructure. The actions of this Network were defined across several work packages (WP). A clinical work package (WP1) defined the guidelines for a minimum standard clinical qualitative MRI assessment for the main neurological diseases. Two neuroimaging technical work packages (WP2 and WP3, for clinical and preclinical scanners) established Standard Operative Procedures for quality controls on phantoms as well as advanced harmonized quantitative MRI protocols for studying the brain of healthy human participants and wild type mice. Under FAIR principles, a web-based e-infrastructure to store and share data across sites was also implemented (WP4). Finally, the RIN translated all these efforts into a large-scale multimodal data collection in patients and animal models with dementia (i.e., case study). The RIN-Neuroimaging Network can maximize the impact of public investments in research and clinical practice acquiring data across institutes and pathologies with high-quality and highly-consistent acquisition protocols, optimizing the analysis pipeline and data sharing procedures.

Assessing the reproducibility of the SienaX and Siena brain atrophy measures using the ADNI back-to-back MP-RAGE MRI scans.

SienaX and Siena are widely used and fully automated algorithms for measuring whole brain volume and volume change in cross-sectional and longitudinal MRI studies and are particularly useful in studies of brain atrophy. The reproducibility of the algorithms was assessed using the 3D T1 weighted MP-RAGE scans from the Alzheimer's Disease Neuroimaging Initiative (ADNI) study. The back-to-back (BTB) MP-RAGE scans in the ADNI data set makes it a valuable benchmark against which to assess the performance of algorithms of measuring atrophy in the human brain with MRI scans. A total of 671 subjects were included for SienaX and 385 subjects for Siena. The annual percentage brain volume change (PBVC) rates were -0.65±0.82%/year for the healthy controls, -1.15±1.21%/year for mild cognitively impairment (MCI) and -1.84±1.33%/year for AD, in line with previous findings. The median of the absolute value of the reproducibility of SienaX's normalized brain volume (NBV) was 0.96% while the 90th percentile was 5.11%. The reproducibility of Siena's PBVC had a median of 0.35% and a 90th percentile of 1.37%. While the median reproducibility for SienaX's NBV was in line with the values previously reported in the literature, the median reproducibility of Siena's PBVC was about twice that reported. Also, the 90th percentiles for both SienaX and Siena were about twice the size that would be expected for a Gaussian distribution. Because of the natural variation of the disease among patients over a year, a perfectly reproducible whole brain atrophy algorithm would reduce the estimated group size needed to detect a specified treatment effect by only 30% to 40% as compared to Siena's.

Norms for Automatic Estimation of Hippocampal Atrophy and a Step Forward for Applicability to the Italian Population.

INTRODUCTION: Hippocampal volume is one of the main biomarkers of Alzheimer's Dementia (AD). Over the years, advanced tools that performed automatic segmentation of Magnetic Resonance Imaging (MRI) T13D scans have been developed, such as FreeSurfer (FS) and ACM-Adaboost (AA). Hippocampal volume is considered abnormal when it is below the 5th percentile of the normative population. The aim of this study was to set norms, established from the Alzheimer's Disease Neuroimaging Initiative (ADNI) population, for hippocampal volume measured with FS v.6.0 and AA tools in the neuGRID platform (www.neugrid2.eu) and demonstrate their applicability for the Italian population. METHODS: Norms were set from a large group of 545 healthy controls belonging to ADNI. For each pipeline, subjects with segmentation errors were discarded, resulting in 532 valid segmentations for FS and 421 for AA (age range 56-90 years). The comparability of ADNI and the Italian Brain Normative Archive (IBNA), representative of the Italian general population, was assessed testing clinical variables, neuropsychological scores and normalized hippocampal volumes. Finally, percentiles were validated using the Italian Alzheimer's disease Repository Without Borders (ARWiBo) as external independent data set to evaluate FS and AA generalizability. RESULTS: Hippocampal percentiles were checked with the chi-square goodness of fit test. P-values were not significant, showing that FS and AA algorithm distributions fitted the data well. Clinical, neuropsychological and volumetric features were similar in ADNI and IBNA (p > 0.01). Hippocampal volumes measured with both FS and AA were associated with age (p < 0.001). The 5th percentile thresholds, indicating left/right hippocampal atrophy were respectively: (i) below 3,223/3,456 mm3 at 56 years and 2,506/2,415 mm3 at 90 years for FS; (ii) below 4,583/4,873 mm3 at 56 years and 3,831/3,870 mm3 at 90 years for AA. The average volumes computed on 100 cognitively intact healthy controls (CN) selected from ARWiBo were close to the 50th percentiles, while those for 100 AD patients were close to the abnormal percentiles. DISCUSSION: Norms generated from ADNI through the automatic FS and AA segmentation tools may be used as normative references for Italian patients with suspected AD.

Inter-Cohort Validation of SuStaIn Model for Alzheimer's Disease.

Alzheimer's disease (AD) is a neurodegenerative disorder which spans several years from preclinical manifestations to dementia. In recent years, interest in the application of machine learning (ML) algorithms to personalized medicine has grown considerably, and a major challenge that such models face is the transferability from the research settings to clinical practice. The objective of this work was to demonstrate the transferability of the Subtype and Stage Inference (SuStaIn) model from well-characterized research data set, employed as training set, to independent less-structured and heterogeneous test sets representative of the clinical setting. The training set was composed of MRI data of 1043 subjects from the Alzheimer's disease Neuroimaging Initiative (ADNI), and the test set was composed of data from 767 subjects from OASIS, Pharma-Cog, and ViTA clinical datasets. Both sets included subjects covering the entire spectrum of AD, and for both sets volumes of relevant brain regions were derived from T1-3D MRI scans processed with Freesurfer v5.3 cross-sectional stream. In order to assess the predictive value of the model, subpopulations of subjects with stable mild cognitive impairment (MCI) and MCIs that progressed to AD dementia (pMCI) were identified in both sets. SuStaIn identified three disease subtypes, of which the most prevalent corresponded to the typical atrophy pattern of AD. The other SuStaIn subtypes exhibited similarities with the previously defined hippocampal sparing and limbic predominant atrophy patterns of AD. Subject subtyping proved to be consistent in time for all cohorts and the staging provided by the model was correlated with cognitive performance. Classification of subjects on the basis of a combination of SuStaIn subtype and stage, mini mental state examination and amyloid-ß1-42 cerebrospinal fluid concentration was proven to predict conversion from MCI to AD dementia on par with other novel statistical algorithms, with ROC curves that were not statistically different for the training and test sets and with area under curve respectively equal to 0.77 and 0.76. This study proves the transferability of a SuStaIn model for AD from research data to less-structured clinical cohorts, and indicates transferability to the clinical setting.