António J. Bastos-Leite

Standardized evaluation of algorithms for computer-aided diagnosis of dementia based on structural MRI: The CADDementia challenge

Algorithms for computer-aided diagnosis of dementia based on structural MRI have demonstrated high performance in the literature, but are difficult to compare as different data sets and methodology were used for evaluation. In addition, it is unclear how the algorithms would perform on previously unseen data, and thus, how they would perform in clinical practice when there is no real opportunity to adapt the algorithm to the data at hand. To address these comparability, generalizability and clinical applicability issues, we organized a grand challenge that aimed to objectively compare algorithms based on a clinically representative multi-center data set. Using clinical practice as the starting point, the goal was to reproduce the clinical diagnosis. Therefore, we evaluated algorithms for multi-class classification of three diagnostic groups: patients with probable Alzheimers disease, patients with mild cognitive impairment and healthy controls. The diagnosis based on clinical criteria was used as reference standard, as it was the best available reference despite its known limitations. For evaluation, a previously unseen test set was used consisting of 354 T1-weighted MRI scans with the diagnoses blinded. Fifteen research teams participated with a total of 29 algorithms. The algorithms were trained on a small training set (n=30) and optionally on data from other sources (e.g., the Alzheimers Disease Neuroimaging Initiative, the Australian Imaging Biomarkers and Lifestyle flagship study of aging). The best performing algorithm yielded an accuracy of 63.0% and an area under the receiver-operating-characteristic curve (AUC) of 78.8%. In general, the best performances were achieved using feature extraction based on voxel-based morphometry or a combination of features that included volume, cortical thickness, shape and intensity. The challenge is open for new submissions via the web-based framework: http://caddementia.grand-challenge.org.

The Contribution of Medial Temporal Lobe Atrophy and Vascular Pathology to Cognitive Impairment in Vascular Dementia

Background and Purpose— Besides cerebrovascular disease, medial temporal lobe atrophy (MTA), a neuroimaging finding suggestive of degenerative pathology, has been shown in vascular dementia (VaD). However, it is unknown to what extent MTA contributes to the pattern of cognitive impairment observed in VaD. Therefore, our purpose was to investigate the relative contribution of cerebrovascular disease and MTA to cognitive impairment in patients fulfilling diagnostic criteria for VaD. Methods— We examined 590 patients (374 men; mean age, 73 years; standard deviation, 8) with probable VaD according to the National Institute of Neurological Disorders and Stroke–Association Internationale pour la Recherche et l’Enseignement en Neurosciences criteria at inclusion into a multicenter clinical trial. Cerebrovascular disease and the degree of MTA were evaluated by using MRI. Cognitive testing included the Mini-Mental State Examination, and the vascular dementia assessment scale. Results— On the basis of the operational definitions for the neuroimaging part of the National Institute of Neurological Disorders and Stroke–Association Internationale pour la Recherche et l’Enseignement en Neurosciences criteria, 485 (82.2%) patients had small vessel VaD and 153 (25.9%) had large vessel VaD. More than half (59.8%) of the patients had considerable MTA. Multiple linear regression analyses revealed that after correction for sex, age, education, and duration of dementia, neuropsychological tests showed that patients with higher grades of MTA or large vessel VaD had significantly worse general cognitive and executive functioning, whereas associations with small vessel disease were restricted to worse executive functioning. Conclusions— Both MTA and large vessel disease contribute to global cognitive impairment in VaD. Small vessel disease contributes to executive dysfunction.

Reliability and sensitivity of visual scales versus volumetry for evaluating white matter hyperintensity progression.

Background: Investigating associations between the change of white matter hyperintensities (WMH) and clinical symptoms over time is crucial for establishing a causal relationship. However, the most suitable method for measuring WMH progression has not been established yet. We compared the reliability and sensitivity of cross-sectional and longitudinal visual scales with volumetry for measuring WMH progression. Methods: Twenty MRI scan pairs (interval 2 years) were included from the Amsterdam center of the LADIS study. Semi-automated volumetry of WMH was performed twice by one rater. Three cross-sectional scales (Fazekas Scale, Age-Related White Matter Changes Scale, Scheltens Scale) and two progression scales (Rotterdam Progression Scale, Schmidt Progression Scale) were scored by 4 and repeated by 2 raters. Results: Mean WMH volume (24.6 ± 27.9 ml at baseline) increased by 4.6 ± 5.1 ml [median volume change (range) = 2.7 (–0.6 to 15.7) ml]. Measuring volumetric change in WMH was reliable (intraobserver:intraclass coefficient = 0.88). All visual scales showed significant change of WMH over time, although the sensitivity was highest for both of the progression scales. Proportional volumetric change of WMH correlated best with the Rotterdam Progression Scale (Spearman’s r = 0.80, p < 0.001) and the Schmidt Progression Scale (Spearman’s r = 0.64, p < 0.01). Although all scales were reliable for assessment of WMH cross-sectionally, WMH progression assessment using visual scales was less reliable, except for the Rotterdam Progression scale which had moderate to good reliability [weighted Cohen’s ĸ = 0.63 (intraobserver), 0.59 (interobserver)]. Conclusion: To determine change in WMH, dedicated progression scales are more sensitive and/or reliable and correlate better with volumetric volume change than cross-sectional scales.

Cerebral Blood Flow by Using Pulsed Arterial Spin-Labeling in Elderly Subjects with White Matter Hyperintensities

BACKGROUND AND PURPOSE: On MR imaging, white matter hyperintensities (WMH) on T2-weighted images are generally considered as a surrogate marker of ischemic small vessel disease in elderly subjects. Pulsed arterial spin-labeling (PASL) is a noninvasive MR perfusion-weighted technique. We hypothesized that elderly subjects with diffuse confluent WMH should have lower cerebral blood flow (CBF) measurements than subjects with punctiform or beginning confluent WMH. MATERIALS AND METHODS: MR images of 21 subjects (13 women; mean age, 76 years; SD, 5), stratified for the degree of WMH, from a single center within the multinational Leukoaraiosis and Disability (LADIS) study, were investigated. CBF images were obtained by means of quantitative imaging of perfusion by using a single-subtraction second version, with thin-section TI periodic saturation PASL. Values of cortical gray matter, subcortical (including white matter and deep gray matter), and global CBF were calculated. CBF measurements of subjects with diffuse confluent WMH (n = 7) were compared with those of subjects with punctiform or beginning confluent WMH (n = 14). RESULTS: Subjects with diffuse confluent WMH were found to have approximately 20% lower mean global CBF (43.5 mL/100 mL/min; SD, 6.3) than subjects with punctiform or beginning confluent WMH (57.9 mL/100 mL/min; SD, 8.6; P < .01), as well as approximately 20% lower mean subcortical (P < .01) and cortical gray matter CBF (P < .05). CONCLUSION: PASL revealed a significant reduction of CBF measurements in elderly subjects with diffuse confluent WMH.

Dysconnectivity Within the Default Mode in First-Episode Schizophrenia: A Stochastic Dynamic Causal Modeling Study With Functional Magnetic Resonance Imaging

We report the first stochastic dynamic causal modeling (sDCM) study of effective connectivity within the default mode network (DMN) in schizophrenia. Thirty-three patients (9 women, mean age = 25.0 years, SD = 5) with a first episode of psychosis and diagnosis of schizophrenia—according to the Diagnostic and Statistic Manual of Mental Disorders, 4th edition, revised criteria—were studied. Fifteen healthy control subjects (4 women, mean age = 24.6 years, SD = 4) were included for comparison. All subjects underwent resting state functional magnetic resonance imaging (fMRI) interspersed with 2 periods of continuous picture viewing. The anterior frontal (AF), posterior cingulate (PC), and the left and right parietal nodes of the DMN were localized in an unbiased fashion using data from 16 independent healthy volunteers (using an identical fMRI protocol). We used sDCM to estimate directed connections between and within nodes of the DMN, which were subsequently compared with t tests at the between subject level. The excitatory effect of the PC node on the AF node and the inhibitory self-connection of the AF node were significantly weaker in patients (mean values = 0.013 and −0.048 Hz, SD = 0.09 and 0.05, respectively) relative to healthy subjects (mean values = 0.084 and −0.088 Hz, SD = 0.15 and 0.77, respectively; P < .05). In summary, sDCM revealed reduced effective connectivity to the AF node of the DMN—reflecting a reduced postsynaptic efficacy of prefrontal afferents—in patients with first-episode schizophrenia.

Intracranial Arterial Stenosis

Intracranial arterial stenosis (IAS) is usually attributable to atherosclerosis and corresponds to the most common cause of stroke worldwide. It is very prevalent among African, Asian, and Hispanic populations. Advancing age, systolic hypertension, diabetes mellitus, high levels of low-density lipoprotein cholesterol, and metabolic syndrome are some of its major risk factors. IAS may be associated with transient or definite neurological symptoms or can be clinically asymptomatic. Transcranial Doppler and magnetic resonance angiography are the most frequently used ancillary examinations for screening and follow-up. Computed tomography angiography can either serve as a screening tool for the detection of IAS or increasingly as a confirmatory test approaching the diagnostic accuracy of catheter digital subtraction angiography, which is still considered the gold (confirmation) standard. The risk of stroke in patients with asymptomatic atherosclerotic IAS is low (up to 6% over a mean follow-up period of approximately 2 years), but the annual risk of stroke recurrence in the presence of a symptomatic stenosis may exceed 20% when the degree of luminal narrowing is 70% or more, recently after an ischemic event, and in women. It is a matter of controversy whether there is a specific type of treatment other than medical management (including aggressive control of vascular risk factors and antiplatelet therapy) that may alter the high risk of stroke recurrence among patients with symptomatic IAS. Endovascular treatment has been thought to be helpful in patients who fail to respond to medical treatment alone, but recent data contradict such expectation.

More than one century of schizophrenia: an evolving perspective.

Abstract Schizophrenia is a very complex psychiatric disorder of unknown etiology, and there is controversy as to whether its name is even appropriate to describe the associated variety of clinical presentations and symptoms. Currently, the diagnosis is essentially based on clinical criteria. These enable a clinical profile to be recognized as encompassing positive symptoms, negative symptoms, disorganization of thinking and behavior, cognitive impairment, mood abnormalities, motor abnormalities, chronic clinical course, and incomplete remissions. The concept has evolved during the past century, and schizophrenia is currently questioned as a single disease entity. Established diagnostic criteria do not mirror the heterogeneity of the disorder. A strategy to deal with clinical heterogeneity in schizophrenia is, perhaps, the adoption of a classification system based on dimensions and stages. An additional strategy to deal with etiological and pathophysiological heterogeneity is to try to identify biomarkers, namely, on the basis of intermediate phenotypes. Despite extensive biological research, the biomarkers for schizophrenia are still lacking.

Genetic determinants of white matter hyperintensities and amyloid angiopathy in familial Alzheimer's disease.

Familial Alzheimers disease (FAD) treatment trials raise interest in the variable occurrence of cerebral amyloid angiopathy (CAA); an emerging important factor in amyloid-modifying therapy. Previous pathological studies reported particularly severe CAA with postcodon 200 PSEN1 mutations and amyloid beta coding domain APP mutations. As CAA may manifest as white matter hyperintensities (WMH) on magnetic resonance imaging, particularly posteriorly, we investigated WMH in 52 symptomatic FAD patients for associations with mutation position. WMH were visually rated in 39 PSEN1 (18 precodon 200); 13 APP mutation carriers and 25 healthy controls. Ten PSEN1 mutation carriers (5 precodon 200) had postmortem examination. Increased WMH were observed in the PSEN1 postcodon 200 group and in the single APP patient with an amyloid beta coding domain (p.Ala692Gly, Flemish) mutation. WMH burden on MRI correlated with severity of CAA and cotton wool plaques in several areas. The precodon 200 group had younger ages at onset, decreased axonal density and/or integrity, and a greater T-lymphocytic response in occipital deep white matter. Mutation site contributes to the phenotypic and pathological heterogeneity witnessed in FAD.

Cerebral microbleeds in familial Alzheimer’s disease

ARTICLE Sir, We read the recent review on microbleeds in Alzheimer’s disease by Cordonnier and van der Flier (2011) with interest. We agree that determining the relevant contributions of cerebrovascular disease (related to conventional risk factors including hypertension) and cerebral amyloid angiopathy to microbleeds identified on T2*-weighted imaging is an important issue, which may lead to new insights into the pathology and pathogenesis of Alzheimer’s disease. One of the difficulties in attempting to delineate the relative contributions of conventional cerebrovascular and cerebral amyloid angiopathy pathology in sporadic Alzheimer’s disease is that multiple pathologies are common at more advanced ages. An alternative strategy is to study patients with autosomal dominantly inherited familial Alzheimer’s disease: such individuals can be diagnosed definitively during life and, having younger age at symptom onset, are much more likely to have ‘pure’ Alzheimer’s disease without co-existing cerebrovascular disease. To our knowledge, no studies have attempted to assess the frequency of microbleeds in familial Alzheimer’s disease. To investigate this question, we identified all individuals with genetically confirmed familial Alzheimer’s disease who had undergone T2*-weighted MRI as part of a longitudinal study of familial Alzheimer’s disease at the Dementia Research Centre, UCL. A total of 12 patients with familial Alzheimer’s disease (mean age: 49.4 years; range: 35.8–63.4 years; mean disease duration: 4.0 years, range: 1.1–6.7 years; 75% male) were eligible for the study. Eleven patients had a mutation in the Presenilin 1 (PSEN1) gene (three intron 4 mutations and one each of L166R, Y115H, E120K, L250S, P264L, R269H, R278I, E280G), and one patient carried the V717I amyloid precursor protein mutation. Twelve controls (mean age: 59.8 …

Early-onset Alzheimer disease: The contribution of neuroimaging for the diagnosis

The diagnosis of Alzheimer disease (AD) at an early age of onset may be a challenging task. The diagnosis of such a type of dementia may be even more difficult when concomitant depressive symptoms occur. We report the case of a 51-year-old woman who was admitted at a Psychiatric Day Hospital presenting with depressive symptoms, visuospatial deficits, apraxia, and minor memory loss. The patient underwent long-term antidepressant therapy, but despite the improvement of depressive symptoms, there was progressive cognitive deterioration. Otherwise, the prior clinical history was unremarkable, and there was no family history of dementia. The clinical examination revealed cognitive deficits in several domains. The patient scored 12 in the Mini-Mental State Examination. Routine laboratory tests were normal. Magnetic resonance (MR) imaging showed global brain volume loss more pronounced than would be expected for someone of the patients age, especially in the precuneus-a pattern of posterior cortical atrophy consistent with the diagnosis of early-onset AD. Images obtained with 99mTc-HMPAO single-photon emission computed tomography (SPECT) also revealed marked brain hypoperfusion involving the left parietal lobe, far beyond the regions of brain volume loss. This clinical case report illustrates the relative contribution of both MR imaging and SPECT for the diagnosis of dementia in a patient with concomitant depressive symptoms. Apart from contributing to the diagnosis of dementia beyond the traditional exclusionary approach, neuroimaging is increasingly being used to classify its particular subtypes. The role of neuroimaging role in AD is also supported by a recent proposal of revised diagnostic criteria, which take into account abnormal biomarkers of disease.