Figure - available from: Human Brain Mapping
This content is subject to copyright. Terms and conditions apply.
Source publication
Alzheimer's disease (AD) and frontotemporal dementia (FTD) are common causes of dementia with partly overlapping, symptoms and brain signatures. There is a need to establish an accurate diagnosis and to obtain markers for disease tracking. We combined unsupervised and supervised machine learning to discriminate between AD and FTD using brain magnet...
Citations
... . Machine learning techniques and models performance evaluation SVMs, a widely used set of supervised ML algorithms, have emerged as the most frequently employed ML techniques in the reviewed studies (Miltiadous et al., 2021;Garcia-Gutierrez et al., 2022;García-Gutierrez et al., 2022;Pérez-Millan et al., 2023, 2024Garn et al., 2017;Wang et al., 2024;Birba et al., 2022;Maito et al., 2023;Möller et al., 2016;Lage et al., 2021;Raamana et al., 2014;Rostamikia et al., 2024;Ajra et al., 2023;Ma et al., 2020). SVMs are favored for their robustness in handling high-dimensional data and their effectiveness in both binary and multi-class classification tasks. ...
... SVMs are favored for their robustness in handling high-dimensional data and their effectiveness in both binary and multi-class classification tasks. They have been applied to a diverse array of data types, including neuroimaging data such as structural MRI features (Pérez-Millan et al., 2023, 2024Möller et al., 2016) and FDG-PET imaging (García-Gutierrez et al., 2022). Additionally, SVMs have been applied to EEG data (Garn et al., 2017;Wang et al., 2024;Rostamikia et al., 2024) and to cognitive and behavioral assessments (Garcia-Gutierrez et al., 2022;Maito et al., 2023). ...
... . /fnagi. . (Miltiadous et al., 2021;Garcia-Gutierrez et al., 2022;García-Gutierrez et al., 2022;Pérez-Millan et al., 2023, 2024Garn et al., 2017;Wang et al., 2024;Birba et al., 2022;Maito et al., 2023;Möller et al., 2016;Lage et al., 2021;Raamana et al., 2014;Rostamikia et al., 2024;Ajra et al., 2023;Ma et al., 2020) k-Nearest Neighbors (KNN) 7 (Miltiadous et al., 2021;García-Gutierrez et al., 2022;Lage et al., 2021;Rostamikia et al., 2024;Ajra et al., 2023;Lal et al., 2024;Díaz-Álvarez et al., 2022) Naive Bayes 6 (Miltiadous et al., 2021;Garcia-Gutierrez et al., 2022;García-Gutierrez et al., 2022;Rostamikia et al., 2024;Díaz-Álvarez et al., 2022;Wang et al., 2016) Random Forests 6 (Miltiadous et al., 2021;Garcia-Gutierrez et al., 2022;García-Gutierrez et al., 2022;Maito et al., 2023;Rostamikia et al., 2024;Lal et al., 2024) Deep neural networks (DNNs) 4 (Hu et al., 2021;Ma et al., 2020Ma et al., , 2024Nguyen et al., 2023) Decision Trees 3 (Miltiadous et al., 2021;Garcia-Gutierrez et al., 2022;García-Gutierrez et al., 2022) XGBoost 3 (Birba et al., 2022;Lal et al., 2024;Sadeghi et al., 2024) Convolutional neural networks (CNNs) 2 (Ajra et al., 2023;Rogeau et al., 2024) AdaBoost Elastic net regression 1 (Bouts et al., 2018) Extra Trees (ET) 1 (Lal et al., 2024) The reviewed studies are grouped into sections based on the best-performing ML method employed. ...
Introduction
Frontotemporal dementia (FTD) is a neurodegenerative disorder characterized by progressive degeneration of the frontal and temporal lobes, leading to significant changes in personality, behavior, and language abilities. Early and accurate differential diagnosis between FTD, its subtypes, and other dementias, such as Alzheimer's disease (AD), is crucial for appropriate treatment planning and patient care. Machine learning (ML) techniques have shown promise in enhancing diagnostic accuracy by identifying complex patterns in clinical and neuroimaging data that are not easily discernible through conventional analysis.
Methods
This systematic review, following PRISMA guidelines and registered in PROSPERO, aimed to assess the strengths and limitations of current ML models used in differentiating FTD from other neurological disorders. A comprehensive literature search from 2013 to 2024 identified 25 eligible studies involving 6,544 patients with dementia, including 2,984 with FTD, 3,437 with AD, 103 mild cognitive impairment (MCI) and 20 Parkinson's disease dementia or probable dementia with Lewy bodies (PDD/DLBPD).
Results
The review found that Support Vector Machines (SVMs) were the most frequently used ML technique, often applied to neuroimaging and electrophysiological data. Deep learning methods, particularly convolutional neural networks (CNNs), have also been increasingly adopted, demonstrating high accuracy in distinguishing FTD from other dementias. The integration of multimodal data, including neuroimaging, EEG signals, and neuropsychological assessments, has been suggested to enhance diagnostic accuracy.
Discussion
ML techniques showed strong potential for improving FTD diagnosis, but challenges like small sample sizes, class imbalance, and lack of standardization limit generalizability. Future research should prioritize the development of standardized protocols, larger datasets, and explainable AI techniques to facilitate the integration of ML-based tools into real-world clinical practice.
Systematic review registration
https://www.crd.york.ac.uk/PROSPERO/view/CRD42024520902.
... As technology advances, the volume of information produced by neural imaging methods expands the significance of traditional methods and deep neural network rises for extracting precise and pertinent information, as well as for making reliable predictions concerning AD from brainimaging data [12]. In recent years, there has been a notable integration of automated or semi-automated algorithms alongside traditional clinical studies, promising advancements in computer-aided diagnosis [13]. Machine learning techniques, within this context clinical and biomarker data to discern patterns that serve as supportive tools for both differential disease diagnosis and monitoring, particularly for Alzheimer's dementia. ...
... It involves training models on data, allowing them to improve their performance based on experience, and making predictions or identifying patterns within new, unseen data. Machine learning-based classification offers a robust approach for generating predictions regarding an individual's disease status using Magnetic Resonance Images scans, with the current focus predominantly on individual classification of Alzheimer's dementia being achieved primarily through anatomical MRI scans [13]. Various machine learning algorithms such as support vector machine, random forest and feature extraction using CNN techniques have been employed in this study for AD classification, yielding models with good performance [12]. ...
... A study in [15] by Luque et al. used CNN, VGG16, and VGG19 models for six common image analysis metrics, built the comprehensive analysis method focusing on binary classifiers and performance metrics for imbalanced datasets. The study by Millan et al. in [13] utilized a hybrid approach of unsupervised and supervised machine learning on MRI data to accurately classify the Alzheimer's disease, frontotemporal-dementia (FTD), and normal, achieved the accuracies of 60.7% to 71.3% across the classification tasks and data types. Shankar et al. in [1] introduced the model using brain image analysis, incorporating Group Grey Wolf Optimization (GGWO), enhanced with the achieving an accuracy of 96.23%; for Alzheimer's disease detection surpassing existing literature by Karthiga et al. in [16], proposed a model utilizing curvelet-based transform and AdaBoost classifier with support-vector-machine, achieving 95.66% accuracy in early Alzheimer's disease prediction from brain MRI images, outperforming conventional classifiers such as decision tree and random forest. ...
Alzheimer’s dementia (AD) poses a significant global health challenge, characterized by progressive cognitive decline, memory impairment, and behavioral changes. The critical need for early detection to enable timely intervention and personalized care is emphasized by the current lack of effective treatments. This study aims to develop precise diagnostic models for AD by employing machine learning and a customized deep-convolutional neural network (cDCNN) with three convolution layers, utilizing Magnetic Resonance Imaging (MRI) data. Methods involve analyzing two distinct datasets—Alzheimer’s Disease Neuroimaging Initiative (ADNI) and Kaggle—to explore diverse cohorts and imaging features associated with AD pathology. To address class imbalance, the Synthetic Minority Over-sampling Technique (SMOTE) is applied across both datasets. A range of traditional machine learning methods, including support vector machine, k-nearest neighbor, random forest, decision trees, and XGBoost classifier, are evaluated alongside the cDCNN model, which leverages key MRI biomarkers of AD for both datasets. Results show the cDCNN model achieved a specific accuracy of 87% on the ADNI dataset, despite challenges in converting ADNI’s Digital Imaging and Communications in Medicine (DICOM) files to JPEG, impacting image clarity. Conclusions suggest that this research provides critical diagnostic tools for clinicians, offering insights into AD pathology and contributing to the alleviation of AD’s societal impact.
... In the same year Pérez-Millan et al. 77 utilizes machine learning methods with both cross-sectional and longitudinal MRI data to distinguish between AD and FTD. With a cohort of 339 subjects and advanced techniques for feature reduction, the study employs a support vector machine classifier. ...
Introduction
The prevalence of neurodegenerative diseases has significantly increased, necessitating a deeper understanding of their symptoms, diagnostic processes, and prevention strategies. Frontotemporal dementia (FTD) and Alzheimer’s disease (AD) are two prominent neurodegenerative conditions that present diagnostic challenges due to overlapping symptoms. To address these challenges, experts utilize a range of imaging techniques, including magnetic resonance imaging (MRI), diffusion tensor imaging (DTI), functional MRI (fMRI), positron emission tomography (PET), and single-photon emission computed tomography (SPECT). These techniques facilitate a detailed examination of the manifestations of these diseases. Recent research has demonstrated the potential of artificial intelligence (AI) in automating the diagnostic process, generating significant interest in this field.
Materials and Methods
This narrative review aims to compile and analyze articles related to the AI-assisted diagnosis of FTD and AD. We reviewed 31 articles published between 2012 and 2024, with 23 focusing on machine learning techniques and 8 on deep learning techniques. The studies utilized features extracted from both single imaging modalities and multi-modal approaches, and evaluated the performance of various classification models.
Results
Among the machine learning studies, Support Vector Machines (SVM) exhibited the most favorable performance in classifying FTD and AD. In deep learning studies, the ResNet convolutional neural network outperformed other networks.
Conclusion
This review highlights the utility of different imaging modalities as diagnostic aids in distinguishing between FTD and AD. However, it emphasizes the importance of incorporating clinical examinations and patient symptom evaluations to ensure comprehensive and accurate diagnoses.
... A (multi-level) hierarchical classification algorithm of AD versus FTD (and bvFTD versus PPA, and nfvPPA versus svPPA) revealed distinct discriminative areas for each comparison using machine learning and demonstrated an overall accuracy of 75.8% (Kim et al., 2019). A study from Barcelona, which tried to distinguish control, AD, and FTD groups using support vector machines, showed an accuracy of 82% in distinguishing the control and FTD groups, and 63% in distinguishing the AD and FTD groups (the accuracy improves to 75% after adding longitudinal data) (Pérez-Millan et al., 2023a;Pérez-Millan et al., 2023b). White matter hyperintensities and cortical atrophy are associated with a loss of empathy (Ozzoude et al., 2022). ...
... Repeat expansion within C9orf72 is the most common genetic cause of FTD, which especially seems to be associated with gray matter changes (Popuri et al., 2018), a thalamic atrophy (Bonham et al., 2023) and a loss of brain stem white matter (Pérez-Millan et al., 2023a;Pérez-Millan et al., 2023b). Dyslexia susceptibility genes play an important role in frontotemporal dementia as well and are associated with specific local cortical thickness reduction (Paternicó et al., 2016). ...
Background: Dementia can be caused by numerous different diseases that present variable clinical courses and reveal multiple patterns of brain atrophy, making its accurate early diagnosis by conventional examinative means challenging. Although highly accurate and powerful, magnetic resonance imaging (MRI) currently plays only a supportive role in dementia diagnosis, largely due to the enormous volume and diversity of data it generates. AI-based software solutions/algorithms that can perform automated segmentation and volumetry analyses of MRI data are being increasingly used to address this issue. Numerous commercial and non-commercial software solutions for automated brain segmentation and volumetry exist, with FreeSurfer being the most frequently used. Objectives: This Review is an account of the current situation regarding the application of automated brain segmentation and volumetry to dementia diagnosis. Methods: We performed a PubMed search for "FreeSurfer AND Dementia" and obtained 493 results. Based on these search results, we conducted an in-depth source analysis to identify additional publications, software tools, and methods. Studies were analyzed for design, patient collective, and for statistical evaluation (mathematical methods, correlations). Results: In the studies identified, the main diseases and cohorts represented were Alzheimer's disease (n = 276), mild cognitive impairment (n = 157), frontotemporal dementia (n = 34), Parkinson's disease (n = 29), dementia with Lewy bodies (n = 20), and healthy controls (n = 356). The findings and methods of a selection of the studies identified were summarized and discussed. Conclusion: Our evaluation showed that, while a large number of studies and software solutions are available, many diseases are underrepresented in terms of their incidence. There is therefore plenty of scope for targeted research.
... Subcortical gray matter volumes and cortical thickness measures were extracted from MRI images, and dimensionality reduction was then applied to build up a single feature that could subsequently be used for classification according to an SVM. Cross-sectional data allowed researchers to reach a discrimination accuracy of 82.1% between FTD and controls, which increased to 88% when they used longitudinal data, whereas FTD was successfully distinguished from AD in 63.3% and 75% of cases with cross-sectional and longitudinal data, respectively [225]. Relevant results were also obtained by another study, which also used brain MRI and semantic fluency, highlighting regional voxel differences existing between bvFTD and controls by means of the Random Forest (RF) ML model. ...
Frontotemporal dementia (FTD) includes a number of neurodegenerative diseases, often with early onset (before 65 years old), characterized by progressive, irreversible deficits in behavioral, linguistic, and executive functions, which are often difficult to diagnose due to their similar phenotypic characteristics to other dementias and psychiatric disorders. The genetic contribution is of utmost importance, although environmental risk factors also play a role in its pathophysiology. In fact, some metals are known to produce free radicals, which, accumulating in the brain over time, can induce oxidative stress, inflammation, and protein misfolding, all of these being key features of FTD and similar conditions. Therefore, the present review aims to summarize the current evidence about the environmental contribution to FTD―mainly dealing with toxic metal exposure―since the identification of such potential environmental risk factors can lead to its early diagnosis and the promotion of policies and interventions. This would allow us, by reducing exposure to these pollutants, to potentially affect society at large in a positive manner, decreasing the burden of FTD and similar conditions on affected individuals and society overall. Future perspectives, including the application of Artificial Intelligence principles to the field, with related evidence found so far, are also introduced.
... While these are promising, at present, literature mostly involves crosssectional studies in AD. Examples are speech-based artificial intelligence (AI) applications predicting cognitive decline (Fristed et al., 2022), biometric measures (e.g., skin conduction, pupillometry and eye-tracking patterns) reflecting social-emotional and/or linguistic deficits (Mendez et al., 2018;Singleton et al., 2022;El Haj et al., 2024), AI-based imaging algorithms for longitudinal brain mapping (Pérez-Millan et al., 2023), and proteomics technology detecting protein profiles (Katzeff et al., 2022). Crucially, the majority of the large and leading studies on disease progression (of neuroimaging in particular) were predominantly performed in genetic cohorts of bvFTD (Staffaroni et al., 2019b(Staffaroni et al., , 2022. ...
Behavioral variant frontotemporal dementia (bvFTD) is a neurodegenerative disorder characterized by diverse and prominent changes in behavior and personality. One of the greatest challenges in bvFTD is to capture, measure and predict its disease progression, due to clinical, pathological and genetic heterogeneity. Availability of reliable outcome measures is pivotal for future clinical trials and disease monitoring. Detection of change should be objective, clinically meaningful and easily assessed, preferably associated with a biological process. The purpose of this scoping review is to examine the status of longitudinal studies in bvFTD, evaluate current assessment tools and propose potential progression markers. A systematic literature search (in PubMed and Embase.com) was performed. Literature on disease trajectories and longitudinal validity of frequently-used measures was organized in five domains: global functioning, behavior, (social) cognition, neuroimaging and fluid biomarkers. Evaluating current longitudinal data, we propose an adaptive battery, combining a set of sensitive clinical, neuroimaging and fluid markers, adjusted for genetic and sporadic variants, for adequate detection of disease progression in bvFTD. CITATION Fieldhouse JLP, van Paassen DN, van Engelen M-PE, De Boer SCM, Hartog WL, Braak S, Schoonmade LJ, Schouws SNTM, Krudop WA, Oudega ML, Mutsaerts HJMM, Teunissen CE, Vijverberg EGB and Pijnenburg YAL (2024) The pursuit for markers of disease progression in behavioral variant frontotemporal dementia: a scoping review to optimize outcome measures for clinical trials.
... Modeling approaches that account for time are of paramount importance to understanding disease progression and comparing brain status across subjects at different disease stages [13][14][15][16]. Using structural MRI, some authors described CTh loss with time, providing valuable information on the characterization of disease trajectories and validation of prognostic biomarkers [17,18]. ...
Background and objective
Alzheimer’s disease (AD) and frontotemporal dementia (FTD) show different patterns of cortical thickness (CTh) loss compared with healthy controls (HC), even though there is relevant heterogeneity between individuals suffering from each of these diseases. Thus, we developed CTh models to study individual variability in AD, FTD, and HC.
Methods
We used the baseline CTh measures of 379 participants obtained from the structural MRI processed with FreeSurfer. A total of 169 AD patients (63 ± 9 years, 65 men), 88 FTD patients (64 ± 9 years, 43 men), and 122 HC (62 ± 10 years, 47 men) were studied. We fitted region-wise temporal models of CTh using Support Vector Regression. Then, we studied associations of individual deviations from the model with cerebrospinal fluid levels of neurofilament light chain (NfL) and 14–3-3 protein and Mini-Mental State Examination (MMSE). Furthermore, we used real longitudinal data from 144 participants to test model predictivity.
Results
We defined CTh spatiotemporal models for each group with a reliable fit. Individual deviation correlated with MMSE for AD and with NfL for FTD. AD patients with higher deviations from the trend presented higher MMSE values. In FTD, lower NfL levels were associated with higher deviations from the CTh prediction. For AD and HC, we could predict longitudinal visits with the presented model trained with baseline data. For FTD, the longitudinal visits had more variability.
Conclusion
We highlight the value of CTh models for studying AD and FTD longitudinal changes and variability and their relationships with cognitive features and biomarkers.
... A growing body of evidence supports the role of machine learning (ML) techniques using brain MRI (15)(16)(17) to support the clinical diagnosis of these two dementias (18)(19)(20)(21)(22). Many studies have shown that a support vector machine (SVM) with neuroimaging data differentiates AD or FTD patients from healthy controls (22)(23)(24)(25)(26)(27). ...
... A growing body of evidence supports the role of machine learning (ML) techniques using brain MRI (15)(16)(17) to support the clinical diagnosis of these two dementias (18)(19)(20)(21)(22). Many studies have shown that a support vector machine (SVM) with neuroimaging data differentiates AD or FTD patients from healthy controls (22)(23)(24)(25)(26)(27). However, fewer studies exist on the differential diagnosis of these two dementias, even though the clinical symptoms of FTD and AD can display a substantial overlap between them (28-30). ...
... Previous ML algorithms using structural MRI data have reported accuracies between 76 and 97% for AD vs. CTR, 72-88% for FTD versus CTR, 51-90% for AD versus FTD, and 54-70% in discriminating between AD, FTD, and CTR (7,11,18,21,22,(40)(41)(42)(43)(44)(45)(46)(47). These studies used different algorithms, with the SVM being the most common. ...
Background
Neuroimaging and fluid biomarkers are used in clinics to differentiate frontotemporal dementia (FTD) from Alzheimer’s disease (AD) and other neurodegenerative and non-neurodegenerative disorders. We implemented a machine learning (ML) algorithm that provides individual probabilistic scores for these patients based on magnetic resonance imaging (MRI) and cerebrospinal fluid (CSF) data.
Methods
We used a calibrated classifier with a Support Vector Machine with MRI data. We obtained group classifications and individual probabilities associated with group correspondence. We used the individual probabilities to address the clinical problem of confidence in the diagnosis. We investigated whether combining MRI and CSF levels of Neurofilament light (NfL) and 14-3-3 could improve the diagnosis confidence.
Results
215 AD patients (65 ± 10 years, 137 women), 103 FTD patients (64 ± 8 years, 49 women), and 173 healthy controls (CTR) (59 ± 15 years, 106 women) were studied. With MRI data only, we obtained accuracies of 88% in the AD vs. healthy controls (CTR) classification, 87% for FTD vs. CTR, 82% for AD vs. FTD, and 80% when differentiating the three groups. A total of 74% of FTD and 73% of AD participants have a high (≥ 0.8) probability of accurate diagnosis in the FTD vs. AD comparison. Adding CSF-NfL and 14-3-3 levels slightly improved the accuracy and the number of patients in the high diagnosis confidence group.
Conclusion
We propose a ML algorithm that provides individual diagnostic probabilities, and we validate it using MRI and/or CSF data. Our solution holds promise towards clinical applications as support to clinical findings or in settings with limited access to expert diagnoses.
... Using cross-sectional and longitudinal MRI data from 399 subjects, including AD and FTD patients and healthy controls, Pérez-Millan et al. [11] created a classification system. They implemented a feature reduction algorithm using unsupervised techniques and reached 83.3% accuracy in healthy control (HC) versus AD, 82.1% in HC versus FTD, and 63.3% in AD versus FTD classification [10]. Lindau et al. [11] calculated the global field power of six frequency bands and the spectral ratio as the ratio of the sum of fast frequency bands and slow frequency bands. ...
Introduction: Frontotemporal dementia (FTD) encompasses a group of heterogeneous neurodegenerative disorders. Aside from genetic cases, its diagnosis is challenging, particularly in the early stages when symptoms are ambiguous, and structural neuroimaging does not reveal characteristic patterns.
Areas covered: The authors performed a comprehensive literature search through MEDLINE, Scopus, and Web of Science databases to gather evidence to aid the diagnostic process for suspected FTD
patients, particularly in early phases, even in sporadic cases, ranging from established to promising tools. Blood-based biomarkers might help identify very early neuropathological stages and guide further evaluations. Subsequently, neurophysiological measures reflecting functional changes in cortical excitatory/inhibitory circuits, along with functional neuroimaging assessing brain network, connectivity, metabolism, and perfusion alterations, could detect specific changes associated to FTD even decadesbefore symptom onset. As the neuropathological process advances, cognitive-behavioral profiles and atrophy patterns emerge, distinguishing specific FTD subtypes.
Expert opinion: Emerging disease-modifying therapies require early patient enrollment. Therefore, a diagnostic paradigm shift is needed – from relying on typical cognitive and neuroimaging profiles of advanced cases to widely applicable biomarkers, primarily fluid biomarkers, and, subsequently, neurophysiological and functional neuroimaging biomarkers where appropriate. Additionally, exploring subjective complaints and behavioral changes detected by home-based technologies might be crucial for early diagnosis.
