AVRA: Automatic visual ratings of atrophy from MRI images using recurrent convolutional neural networks

Cerebrovascular co-pathology and cholinergic white matter pathways along the Lewy body continuum

Dementia with Lewy bodies often presents with cholinergic degeneration and varying degrees of cerebrovascular disease. There is a lack of radiological methods for evaluating cholinergic degeneration in dementia with Lewy bodies. We investigated the potential of the Cholinergic Pathway Hyperintensities Scale (CHIPS) in identifying cerebrovascular disease-related disruptions in cholinergic white matter pathways, offering a practical and accessible method for assessing cholinergic integrity in neurodegenerative diseases. We assessed the associations of CHIPS with regional brain atrophy, Alzheimer's disease co-pathology and clinical phenotype. Additionally, we compared its diagnostic performance to that of other manual and automated evaluation methods. We included 82 individuals (41 patients in the Lewy body continuum with either probable dementia with Lewy bodies or mild cognitive impairment with Lewy bodies, and 41 healthy controls) from the Sant Pau Initiative on Neurodegeneration cohort. We used CHIPS to assess cholinergic white matter signal abnormalities (WMSA) on MRI, while tractography mean diffusivity provided a complementary measure of cholinergic WMSA. For global WMSA evaluation, we used the Fazekas scale and FreeSurfer. CHIPS successfully identified cerebrovascular disease-related disruptions in cholinergic white matter pathways, as evidenced by its association with tractography and global WMSA markers (P < 0.005 for all associations). Lewy body patients showed a significantly higher degree of WMSA in the external capsule cholinergic pathway despite no significant differences in global WMSA compared to controls. CHIPS score in the posterior external capsule and the mean diffusivity in the external capsule and cingulum exceeded the threshold for an optimal biomarker (sensitivity and specificity values above 80%) in discriminating Lewy body patients from controls. Furthermore, higher CHIPS scores, Fazekas scale and tractography mean diffusivity were associated with more pronounced frontal atrophy in Lewy body patients but not in controls. No associations were found for the four WMSA and integrity methods with the core clinical features, clinical or cognitive measures, or CSF biomarkers. In conclusion, cholinergic WMSA were more pronounced in Lewy body patients compared to healthy controls, independently of global WMSA. Our findings indicate that cerebrovascular disease-related disruptions in cholinergic white matter may be linked to frontal atrophy in Lewy body patients. Clinically, we demonstrate the potential of CHIPS to assess cholinergic WMSA using widely available MRI sequences. Our data suggest cerebrovascular disease co-pathology could drive the cholinergic degeneration in Lewy body patients, opening opportunities for therapeutic interventions targeting vascular health from mild cognitive impairment with Lewy bodies through manifest dementia with Lewy bodies.

Anatomical Determinants of STN Coordinate Shift in Idiopathic Parkinson's Disease DBS Surgery

OBJECTIVE

This study examines how anatomical variations influence the targeting coordinates of the subthalamic nucleus (STN) in patients with Idiopathic Parkinson's Disease (IPD) undergoing Deep Brain Stimulation (DBS), with the goal of enhancing targeting accuracy.

METHODS

A retrospective analysis was performed on 202 STNs from patients who received bilateral STN-DBS surgery. Pre- and postoperative imaging data were used to determine accurate STN coordinates, while brain volume measurements, ventricle size, Evans Index, and AC -PC length were analyzed. Atrophy grading scales were also applied. Correlation and regression analyses assessed the relationship between the STN target location and all anatomical parameters on the x, y, and z axes.

RESULTS

Age showed a significant positive correlation with lateral STN coordinate shift on the x-axis, with each additional year leading to a 0.046 mm shift. An increase in peripheral gray matter volume and a decrease in white matter volume were significantly associated with the lateral displacement of the STN. Total ventricle volume demonstrated a positive correlation with STN shift on both the x-axis (0.0227 mm per cm3 increase) and z-axis (0.0087 mm per cm3 increase). Significant correlations were also found for the Evans Index with lateral shift on the x-axis and for AC-PC length with vertical shifts.

CONCLUSION

Anatomical factors, such as brain volume, ventricle size, Evans Index, AC-PC length, and atrophy scores, significantly influence STN localization in PD patients undergoing DBS. Accounting for these shifts during surgical planning may improve electrode placement accuracy and enhance therapeutic outcomes, underscoring the importance of personalized targeting strategies.

Cerebrovascular and Alzheimer's disease biomarkers in dementia with Lewy bodies and other dementias

Co-pathologies are common in dementia with Lewy bodies and other dementia disorders. We investigated cerebrovascular and Alzheimer's disease co-pathologies in patients with dementia with Lewy bodies in comparison with patients with mild cognitive impairment, Alzheimer's disease, mixed dementia, vascular dementia or Parkinson's disease with dementia and cognitively unimpaired participants. We assessed the association of biomarkers of cerebrovascular and Alzheimer's disease co-pathologies with medial temporal atrophy and global cognitive performance. Additionally, we evaluated whether the findings were specific to dementia with Lewy bodies. We gathered a multi-cohort dataset of 4549 participants (dementia with Lewy bodies = 331, cognitively unimpaired = 1505, mild cognitive impairment = 1489, Alzheimer's disease = 708, mixed dementia = 268, vascular dementia = 148, Parkinson's disease with dementia = 120) from the MemClin Study, Karolinska Imaging in Dementia Study, Gothenburg H70 Birth Cohort Studies and the European DLB Consortium. Cerebrovascular co-pathology was assessed with visual ratings of white matter hyperintensities using the Fazekas scale through structural imaging. Alzheimer's disease biomarkers of β-amyloid and phosphorylated tau were assessed in the cerebrospinal fluid for a subsample (N = 2191). Medial temporal atrophy was assessed with visual ratings and global cognition with the mini-mental state examination. Differences and associations were assessed through regression models, including interaction terms. In dementia with Lewy bodies, 43% had a high white matter hyperintensity load, which was significantly higher than that in cognitively unimpaired (14%), mild cognitive impairment (26%) and Alzheimer's disease (27%), but lower than that in vascular dementia (62%). In dementia with Lewy bodies, white matter hyperintensities were associated with medial temporal atrophy, and the interaction term showed that this association was stronger than that in cognitively unimpaired and mixed dementia. However, the association between white matter hyperintensities and medial temporal atrophy was non-significant when β-amyloid was included in the model. Instead, β-amyloid predicted medial temporal atrophy in dementia with Lewy bodies, in contrast to the findings in mild cognitive impairment where medial temporal atrophy scores were independent of β-amyloid. Dementia with Lewy bodies had the lowest performance on global cognition, but this was not associated with white matter hyperintensities. In Alzheimer's disease, global cognitive performance was lower in patients with more white matter hyperintensities. We conclude that white matter hyperintensities are common in dementia with Lewy bodies and are associated with more atrophy in medial temporal lobes, but this association depended on β-amyloid-related pathology in our cohort. The associations between biomarkers were overall stronger in dementia with Lewy bodies than in some of the other diagnostic groups.

Evaluation of a multimodal diagnostic algorithm for prediction of cognitive impairment in elderly patients with dizziness

BACKGROUND

The current diagnostic workup for chronic dizziness in elderly patients often neglects neuropsychological assessment, thus missing a relevant proportion of patients, who perceive dizziness as a subjective chief complaint of a concomitant cognitive impairment. This study aimed to establish risk prediction models for cognitive impairment in chronic dizzy patients based on data sources routinely collected in a dizziness center.

METHODS

One hundred patients (age: 74.7 ± 7.1 years, 41.0% women) with chronic dizziness were prospectively characterized by (1) neuro-otological testing, (2) quantitative gait assessment, (3) graduation of focal brain atrophy and white matter lesion load, and (4) cognitive screening (MoCA). A linear regression model was trained to predict patients' total MoCA score based on 16 clinical features derived from demographics, vestibular testing, gait analysis, and imaging scales. Additionally, we trained a binary logistic regression model on the same data sources to identify those patients with a cognitive impairment (i.e., MoCA < 25).

RESULTS

The linear regression model explained almost half of the variance of patients' total MoCA score (R2 = 0.49; mean absolute error: 1.7). The most important risk-predictors of cognitive impairment were age (β = - 0.75), pathological Romberg's sign (β = - 1.05), normal caloric test results (β = - 0.8), slower timed-up-and-go test (β = - 0.67), frontal (β = - 0.6) and temporal (β = - 0.54) brain atrophy. The binary classification yielded an area under the curve of 0.84 (95% CI 0.70-0.98) in distinguishing between cognitively normal and impaired patients.

CONCLUSIONS

The need for cognitive testing in patients with chronic dizziness can be efficiently approximated by available data sources from routine diagnostic workup in a dizziness center.

The role of visual rating and automated brain volumetry in early detection and differential diagnosis of Alzheimer's disease

BACKGROUND

Medial temporal lobe atrophy (MTA) is a diagnostic marker for mild cognitive impairment (MCI) and Alzheimer's disease (AD), but the accuracy of quantitative MTA (QMTA) in diagnosing early AD is unclear. This study aimed to investigate the accuracy of QMTA and its related components (inferior lateral ventricle [ILV] and hippocampus) with MTA in the early diagnosis of MCI and AD.

METHODS

This study included four groups: normal (NC), MCI stable (MCIs), MCI converted to AD (MCIs), and mild AD (M-AD) groups. Magnetic resonance image analysis software was used to quantify the hippocampus, ILV, and QMTA. MTA was rated by two experienced neurologists. Receiver operating characteristic area under the curve (AUC) analysis was performed to compare their capability in differentiating AD from NC and MCI, and optimal thresholds were determined using the Youden index.

RESULTS

QMTA distinguished M-AD from NC and MCI with higher diagnostic accuracy than MTA, hippocampus, and ILV (AUCNC = 0.976, AUCMCI = 0.836, AUCMCIs = 0.894, AUCMCIc = 0.730). The diagnostic accuracy of QMTA was superior to that of MTA, the hippocampus, and ILV in differentiating MCI from AD. The diagnostic accuracy of QMTA was found to remain the best across age, sex, and pathological subgroups analyzed. The sensitivity (92.45%) and specificity (90.64%) were higher in this study when a cutoff value of 0.635 was chosen for QMTA.

CONCLUSIONS

QMTA may be a better choice than the MTA scale or the associated quantitative components alone in identifying AD patients and MCI individuals with higher progression risk.

Regional MRI volumetry using NeuroQuant versus visual rating scales in patients with cognitive impairment and dementia

BACKGROUND AND PURPOSE

The aims were to compare the novel regional brain volumetric measures derived by the automatic software NeuroQuant (NQ) with clinically used visual rating scales of medial temporal lobe atrophy (MTA), global cortical atrophy-frontal (GCA-f), and posterior atrophy (PA) brain regions, assessing their diagnostic validity, and to explore if combining automatic and visual methods would increase diagnostic prediction accuracy.

METHODS

Brain magnetic resonance imaging (MRI) examinations from 86 patients with subjective and mild cognitive impairment (i.e., non-dementia, n = 41) and dementia (n = 45) from the Memory Clinic at Oslo University Hospital were assessed using NQ volumetry and with visual rating scales. Correlations, receiver operating characteristic analyses calculating area under the curves (AUCs) for diagnostic accuracy, and logistic regression analyses were performed.

RESULTS

The correlations between NQ volumetrics and visual ratings of corresponding regions were generally high between NQ hippocampi/temporal volumes and MTA (r = -0.72/-0.65) and between NQ frontal volume and GCA-f (r = -0.62) but lower between NQ parietal/occipital volumes and PA (r = -0.49/-0.37). AUCs of each region, separating non-dementia from dementia, were generally comparable between the two methods, except that NQ hippocampi volume did substantially better than visual MTA (AUC = 0.80 vs. 0.69). Combining both MRI methods increased only the explained variance of the diagnostic prediction substantially regarding the posterior brain region.

CONCLUSIONS

The findings of this study encourage the use of regional automatic volumetry in locations lacking neuroradiologists with experience in the rating of atrophy typical of neurodegenerative diseases, and in primary care settings.

A comprehensive review of machine learning algorithms and their application in geriatric medicine: present and future

The increasing access to health data worldwide is driving a resurgence in machine learning research, including data-hungry deep learning algorithms. More computationally efficient algorithms now offer unique opportunities to enhance diagnosis, risk stratification, and individualised approaches to patient management. Such opportunities are particularly relevant for the management of older patients, a group that is characterised by complex multimorbidity patterns and significant interindividual variability in homeostatic capacity, organ function, and response to treatment. Clinical tools that utilise machine learning algorithms to determine the optimal choice of treatment are slowly gaining the necessary approval from governing bodies and being implemented into healthcare, with significant implications for virtually all medical disciplines during the next phase of digital medicine. Beyond obtaining regulatory approval, a crucial element in implementing these tools is the trust and support of the people that use them. In this context, an increased understanding by clinicians of artificial intelligence and machine learning algorithms provides an appreciation of the possible benefits, risks, and uncertainties, and improves the chances for successful adoption. This review provides a broad taxonomy of machine learning algorithms, followed by a more detailed description of each algorithm class, their purpose and capabilities, and examples of their applications, particularly in geriatric medicine. Additional focus is given on the clinical implications and challenges involved in relying on devices with reduced interpretability and the progress made in counteracting the latter via the development of explainable machine learning.

Differential response to donepezil in MRI subtypes of mild cognitive impairment

BACKGROUND

Donepezil is an approved therapy for the treatment of Alzheimer's disease (AD). Results across clinical trials have been inconsistent, which may be explained by design-methodological issues, the pathophysiological heterogeneity of AD, and diversity of included study participants. We investigated whether response to donepezil differs in mild cognitive impaired (MCI) individuals demonstrating different magnetic resonance imaging (MRI) subtypes.

METHODS

From the Hippocampus Study double-blind, randomized clinical trial, we included 173 MCI individuals (donepezil = 83; placebo = 90) with structural MRI data, at baseline and at clinical follow-up assessments (6-12-month). Efficacy outcomes were the annualized percentage change (APC) in hippocampal, ventricular, and total grey matter volumes, as well as in the AD cortical thickness signature. Participants were classified into MRI subtypes as typical AD, limbic-predominant, hippocampal-sparing, or minimal atrophy at baseline. We primarily applied a subtyping approach based on continuous scale of two subtyping dimensions. We also used the conventional categorical subtyping approach for comparison.

RESULTS

Donepezil-treated MCI individuals showed slower atrophy rates compared to the placebo group, but only if they belonged to the minimal atrophy or hippocampal-sparing subtypes. Importantly, only the continuous subtyping approach, but not the conventional categorical approach, captured this differential response.

CONCLUSIONS

Our data suggest that individuals with MCI, with hippocampal-sparing or minimal atrophy subtype, may have improved benefit from donepezil, as compared with MCI individuals with typical or limbic-predominant patterns of atrophy. The newly proposed continuous subtyping approach may have advantages compared to the conventional categorical approach. Future research is warranted to demonstrate the potential of subtype stratification for disease prognosis and response to treatment.

TRIAL REGISTRATION

ClinicalTrial.gov NCT00403520. Submission Date: November 21, 2006.

Behavioral variant frontotemporal dementia in patients with primary psychiatric disorder: A magnetic resonance imaging study

BACKGROUND

The clinical diagnosis of behavioral variant frontotemporal dementia (bvFTD) in patients with a history of primary psychiatric disorder (PPD) is challenging. PPD shows the typical cognitive impairments observed in patients with bvFTD. Therefore, the correct identification of bvFTD onset in patients with a lifetime history of PPD is pivotal for an optimal management.

METHODS

Twenty-nine patients with PPD were included in this study. After clinical and neuropsychological evaluations, 16 patients with PPD were clinically classified as bvFTD (PPD-bvFTD+), while in 13 cases clinical symptoms were associated with the typical course of the psychiatric disorder itself (PPD-bvFTD-). Voxel- and surface-based investigations were used to characterize gray matter changes. Volumetric and cortical thickness measures were used to predict the clinical diagnosis at a single-subject level using a support vector machine (SVM) classification framework. Finally, we compared classification performances of magnetic resonance imaging (MRI) data with automatic visual rating scale of frontal and temporal atrophy.

RESULTS

PPD-bvFTD+ showed a gray matter decrease in thalamus, hippocampus, temporal pole, lingual, occipital, and superior frontal gyri compared to PPD-bvFTD- (p < .05, family-wise error-corrected). SVM classifier showed a discrimination accuracy of 86.2% in differentiating PPD patients with bvFTD from those without bvFTD.

CONCLUSIONS

Our study highlights the utility of machine learning applied to structural MRI data to support the clinician in the diagnosis of bvFTD in patients with a history of PPD. Gray matter atrophy in temporal, frontal, and occipital brain regions may represent a useful hallmark for a correct identification of dementia in PPD at a single-subject level.

Multimodal fusion diagnosis of depression and anxiety based on CNN-LSTM model

BACKGROUND

In recent years, more and more people suffer from depression and anxiety. These symptoms are hard to be spotted and can be very dangerous. Currently, the Self-Reported Anxiety Scale (SAS) and Self-Reported Depression Scale (SDS) are commonly used for initial screening for depression and anxiety disorders. However, the information contained in these two scales is limited, while the symptoms of subjects are various and complex, which results in the inconsistency between the questionnaire evaluation results and the clinician's diagnosis results. To fully mine the scale data, we propose a method to extract the features from the facial expression and movements, which are generated from the video recorded simultaneously when subjects fill in the scale. Then we collect the facial expression, movements and scale information to establish a multimodal framework for improving the accuracy and robustness of the diagnosis of depression and anxiety.

METHODS

We collect the scale results of the subjects and the videos when filling in the scales. Given the two scales, SAS and SDS, we construct a model with two branches, where each branch processes the multimodal data of SAS and SDS, respectively. In the branch, we first build a convolutional neural network (CNN) to extracts the facial expression features in each frame of images. Secondly, we establish a long short-term memory (LSTM) network to further embedding the facial expression feature and build the connections between various frames, so that the movement feature in the video can be generated. Thirdly, we transform the scale scores into one-hot format, and feed them into the corresponding branch of the network to further mining the information of the multimodal data. Finally, we fuse the embeddings of these two branches to generate inference results of depression and anxiety.

RESULTS AND CONCLUSIONS

Based on the score results of SAS and SDS, our multimodal model further mines the video information, and can reach the accuracy of 0.946 in diagnosing depression and anxiety. This study demonstrates the feasibility of using our CNN-LSTM-based multimodal model for initial screening and diagnosis of depression and anxiety disorders with high diagnostic performance.

Small Vessel Disease, a Marker of Brain Health: What the Radiologist Needs to Know

Small vessel disease, a disorder of cerebral microvessels, is an expanding epidemic and a common cause of stroke and dementia. Despite being almost ubiquitous in brain imaging, the clinicoradiologic association of small vessel disease is weak, and the underlying pathogenesis is poorly understood. The STandards for ReportIng Vascular changes on nEuroimaging (STRIVE) criteria have standardized the nomenclature. These include white matter hyperintensities of presumed vascular origin, recent small subcortical infarcts, lacunes of presumed vascular origin, prominent perivascular spaces, cerebral microbleeds, superficial siderosis, cortical microinfarcts, and brain atrophy. Recently, the rigid categories among cognitive impairment, vascular dementia, stroke, and small vessel disease have become outdated, with a greater emphasis on brain health. Conventional and advanced small vessel disease imaging markers allow a comprehensive assessment of global brain heath. In this review, we discuss the pathophysiology of small vessel disease neuroimaging nomenclature by means of the STRIVE criteria, clinical implications, the role of advanced imaging, and future directions.

Inter-modality assessment of medial temporal lobe atrophy in a non-demented population: application of a visual rating scale template across radiologists with varying clinical experience

OBJECTIVES

To assess inter-modality agreement and accuracy for medial temporal lobe atrophy (MTA) ratings across radiologists with varying clinical experience in a non-demented population.

METHODS

Four raters (two junior radiologists and two senior neuroradiologists) rated MTA on CT and MRI scans using Scheltens' MTA scale. Ratings were compared to a consensus rating by two experienced neuroradiologists for estimation of true positive and negative rates (TPR and TNR) and over- and underestimation of MTA. Inter-modality agreement expressed as Cohen's κ (dichotomized data), Cohen's κ_w, and two-way mixed, single measures, consistency ICC (ordinal data) were determined. Adequate agreement was defined as κ/κ_w ≥ 0.80 and ICC ≥ 0.80 (significance level at 95% CI ≥ 0.65).

RESULTS

Forty-nine subjects (median age 72 years, 27% abnormal MTA) with cognitive impairment were included. Only junior radiologists achieved adequate agreement expressed as Cohen's κ. All raters achieved adequate agreement expressed as Cohen's κ_w and ICC. True positive rates varied from 69 to 100% and TNR varied from 85 to 100%. No under- or overestimation of MTA was observed. Ratings did not differ between radiologists.

CONCLUSION

We conclude that radiologists with varying experience achieve adequate inter-modality agreement and similar accuracy when Scheltens' MTA scale is used to rate MTA on a non-demented population. However, TPR varied between radiologists which could be attributed to rating style differences.

KEY POINTS

• Radiologists with varying experience achieve adequate inter-modality agreement with similar accuracy when Scheltens' MTA scale is used to rate MTA on a non-demented population. • Differences in rating styles might affect accuracy, this was most evident for senior neuroradiologists, and only junior radiologists achieved adequate agreement on dichotomized (abnormal/normal) ratings. • The use of an MTA scale template might compensate for varying clinical experience which could make it applicable for clinical use.

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-β_1-42 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.

Structural brain imaging as a diagnostic tool in dementia, why and how?

The demands for more people to be investigated due to cognitive failure and suspected dementia are increasing as increasing numbers of us get older and the incidence of dementia increases. An important part of a dementia study includes the structural imaging of the brain. Two imaging techniques, Computed Tomography (CT) and Magnetic Resonance Imaging (MRI), are used in this context. They differ in many ways and one question is which of the methods should be used in the first instance. Considering the large number of investigations to be expected in the future it is vital that they be cost-effective. Structural imaging aims partly to find secondary causes of cognitive failure and partly to provide support in the differential diagnostic reasoning. The methods differ; CT is significantly cheaper but exposes the patient to radiation, MRI is expensive but does not use X-rays. MRI provides better imaging of cerebrovascular lesions than CT as well as better imaging of structures near the skull base. The difference in diagnostic accuracy is small and it is doubtful whether that difference justifies the large difference in cost.

Magnetic resonance imaging manifestations of cerebral small vessel disease: automated quantification and clinical application

The common cerebral small vessel disease (CSVD) neuroimaging features visible on conventional structural magnetic resonance imaging include recent small subcortical infarcts, lacunes, white matter hyperintensities, perivascular spaces, microbleeds, and brain atrophy. The CSVD neuroimaging features have shared and distinct clinical consequences, and the automatic quantification methods for these features are increasingly used in research and clinical settings. This review article explores the recent progress in CSVD neuroimaging feature quantification and provides an overview of the clinical consequences of these CSVD features as well as the possibilities of using these features as endpoints in clinical trials. The added value of CSVD neuroimaging quantification is also discussed for researches focused on the mechanism of CSVD and the prognosis in subjects with CSVD.

Comparison of subtyping methods for neuroimaging studies in Alzheimer's disease: a call for harmonization

Biological subtypes in Alzheimer's disease, originally identified on neuropathological data, have been translated to in vivo biomarkers such as structural magnetic resonance imaging and positron emission tomography, to disentangle the heterogeneity within Alzheimer's disease. Although there is methodological variability across studies, comparable characteristics of subtypes are reported at the group level. In this study, we investigated whether group-level similarities translate to individual-level agreement across subtyping methods, in a head-to-head context. We compared five previously published subtyping methods. Firstly, we validated the subtyping methods in 89 amyloid-beta positive Alzheimer's disease dementia patients (reference group: 70 amyloid-beta negative healthy individuals) using structural magnetic resonance imaging. Secondly, we extended and applied the subtyping methods to 53 amyloid-beta positive prodromal Alzheimer's disease and 30 amyloid-beta positive Alzheimer's disease dementia patients (reference group: 200 amyloid-beta negative healthy individuals) using structural magnetic resonance imaging and tau positron emission tomography. Subtyping methods were implemented as outlined in each original study. Group-level and individual-level comparisons across methods were performed. Each individual subtyping method was replicated, and the proof-of-concept was established. At the group level, all methods captured subtypes with similar patterns of demographic and clinical characteristics, and with similar cortical thinning and tau positron emission tomography uptake patterns. However, at the individual level, large disagreements were found in subtype assignments. Although characteristics of subtypes are comparable at the group level, there is a large disagreement at the individual level across subtyping methods. Therefore, there is an urgent need for consensus and harmonization across subtyping methods. We call for the establishment of an open benchmarking framework to overcome this problem.

Fetal brain age estimation and anomaly detection using attention-based deep ensembles with uncertainty

MRI-based brain age prediction has been widely used to characterize normal brain development, and deviations from the typical developmental trajectory are indications of brain abnormalities. Age prediction of the fetal brain remains unexplored, although it can be of broad interest to prenatal examination given the limited diagnostic tools available for assessment of the fetal brain. In this work, we built an attention-based deep residual network based on routine clinical T2-weighted MR images of 659 fetal brains, which achieved an overall mean absolute error of 0.767 weeks and R² of 0.920 in fetal brain age prediction. The predictive uncertainty and estimation confidence were simultaneously quantified from the network as markers for detecting fetal brain anomalies using an ensemble method. The novel markers overcame the limitations in conventional brain age estimation and demonstrated promising diagnostic power in differentiating several types of fetal abnormalities, including small head circumference, malformations and ventriculomegaly with the area under the curve of 0.90, 0.90 and 0.67, respectively. In addition, attention maps were derived from the network, which revealed regional features that contributed to fetal age estimation at each gestational stage. The proposed attention-based deep ensembles demonstrated superior performance in fetal brain age estimation and fetal anomaly detection, which has the potential to be translated to prenatal diagnosis in clinical practice.

Analysis of MRI Data in Diagnostic Neuroradiology

Magnetic resonance imaging (MRI) is a noninvasive imaging tool for neuroradiological diagnosis. Numerous concepts of automated MRI analysis and the use of machine learning have been proposed to assist diagnosis and prognosis. While these academic innovations have proven effective in principle within controlled environments, their application to clinical practice has faced unmet requirements, such as the ability to perform reliably across a heterogeneous population, to work robustly in the presence of comorbidities, and to be invariant to scanner hardware and image quality. The lack of realistic confidence bounds and the inability to handle missing data have also reduced the application of most of these methods outside of academic studies. Mastering the complex challenges in the diagnostic process may help researchers discover novel biological constructs in multimodal data and improve stratification for clinical trials, paving the way for precision medicine. This review presents the state of the art of computerized brain MRI analysis for diagnostic purposes. We critically evaluate the current clinical usefulness of the methods and highlight challenges and future perspectives of the field.

Medial temporal atrophy in preclinical dementia: Visual and automated assessment during six year follow-up

•

Visual MTA ratings can detect longitudinal changes in preclinical dementia patients.

•

Medial temporal atrophy rate is greater in individuals with an AD biomarker profile.

•

Visual MTA ratings provide a robust alternative to automated measures.

Medial temporal lobe (MTL) atrophy is an important morphological marker of many dementias and is closely related to cognitive decline. In this study we aimed to characterize longitudinal progression of MTL atrophy in 93 individuals with subjective cognitive decline and mild cognitive impairment followed up over six years, and to assess if clinical rating scales are able to detect these changes. All MRI images were visually rated according to Scheltens’ scale of medial temporal atrophy (MTA) by two neuroradiologists and AVRA, a software for automated MTA ratings. The images were also segmented using FreeSurfer’s longitudinal pipeline in order to compare the MTA ratings to volumes of the hippocampi and inferior lateral ventricles. We found that MTL atrophy rates increased with CSF biomarker abnormality, used to define preclinical stages of Alzheimer’s Disease. Both AVRA’s and the radiologists’ MTA ratings showed similar longitudinal trends as the subcortical volumes, suggesting that visual rating scales provide a valid alternative to automatic segmentations. Our results further showed that it took more than 8 years on average for individuals with mild cognitive impairment, and an Alzheimer’s disease biomarker profile, to increase the MTA score by one. This suggests that discrete MTA ratings are too coarse for tracking individual MTL atrophy in short time spans. While the MTA scores from each radiologist showed strong correlations to subcortical volumes, the inter-rater agreement was low. We conclude that the main limitation of quantifying MTL atrophy with visual ratings in clinics is the subjectiveness of the assessment.

The reliability of a deep learning model in clinical out-of-distribution MRI data: A multicohort study

Deep learning (DL) methods have in recent years yielded impressive results in medical imaging, with the potential to function as clinical aid to radiologists. However, DL models in medical imaging are often trained on public research cohorts with images acquired with a single scanner or with strict protocol harmonization, which is not representative of a clinical setting. The aim of this study was to investigate how well a DL model performs in unseen clinical datasets-collected with different scanners, protocols and disease populations-and whether more heterogeneous training data improves generalization. In total, 3117 MRI scans of brains from multiple dementia research cohorts and memory clinics, that had been visually rated by a neuroradiologist according to Scheltens' scale of medial temporal atrophy (MTA), were included in this study. By training multiple versions of a convolutional neural network on different subsets of this data to predict MTA ratings, we assessed the impact of including images from a wider distribution during training had on performance in external memory clinic data. Our results showed that our model generalized well to datasets acquired with similar protocols as the training data, but substantially worse in clinical cohorts with visibly different tissue contrasts in the images. This implies that future DL studies investigating performance in out-of-distribution (OOD) MRI data need to assess multiple external cohorts for reliable results. Further, by including data from a wider range of scanners and protocols the performance improved in OOD data, which suggests that more heterogeneous training data makes the model generalize better. To conclude, this is the most comprehensive study to date investigating the domain shift in deep learning on MRI data, and we advocate rigorous evaluation of DL models on clinical data prior to being certified for deployment.