To be diagnostically useful, structural MRI must reliably distinguish Alzheimer's disease (AD) from normal aging in individual scans. Recent advances in statistical learning theory have led to the ...application of support vector machines to MRI for detection of a variety of disease states. The aims of this study were to assess how successfully support vector machines assigned individual diagnoses and to determine whether data-sets combined from multiple scanners and different centres could be used to obtain effective classification of scans. We used linear support vector machines to classify the grey matter segment of T1-weighted MR scans from pathologically proven AD patients and cognitively normal elderly individuals obtained from two centres with different scanning equipment. Because the clinical diagnosis of mild AD is difficult we also tested the ability of support vector machines to differentiate control scans from patients without post-mortem confirmation. Finally we sought to use these methods to differentiate scans between patients suffering from AD from those with frontotemporal lobar degeneration. Up to 96% of pathologically verified AD patients were correctly classified using whole brain images. Data from different centres were successfully combined achieving comparable results from the separate analyses. Importantly, data from one centre could be used to train a support vector machine to accurately differentiate AD and normal ageing scans obtained from another centre with different subjects and different scanner equipment. Patients with mild, clinically probable AD and age/sex matched controls were correctly separated in 89% of cases which is compatible with published diagnosis rates in the best clinical centres. This method correctly assigned 89% of patients with post-mortem confirmed diagnosis of either AD or frontotemporal lobar degeneration to their respective group. Our study leads to three conclusions: Firstly, support vector machines successfully separate patients with AD from healthy aging subjects. Secondly, they perform well in the differential diagnosis of two different forms of dementia. Thirdly, the method is robust and can be generalized across different centres. This suggests an important role for computer based diagnostic image analysis for clinical practice.
Normal ageing is associated with characteristic changes in brain microstructure. Although in vivo neuroimaging captures spatial and temporal patterns of age-related changes of anatomy at the ...macroscopic scale, our knowledge of the underlying (patho)physiological processes at cellular and molecular levels is still limited. The aim of this study is to explore brain tissue properties in normal ageing using quantitative magnetic resonance imaging (MRI) alongside conventional morphological assessment. Using a whole-brain approach in a cohort of 26 adults, aged 18–85years, we performed voxel-based morphometric (VBM) analysis and voxel-based quantification (VBQ) of diffusion tensor, magnetization transfer (MT), R1, and R2* relaxation parameters. We found age-related reductions in cortical and subcortical grey matter volume paralleled by changes in fractional anisotropy (FA), mean diffusivity (MD), MT and R2*. The latter were regionally specific depending on their differential sensitivity to microscopic tissue properties. VBQ of white matter revealed distinct anatomical patterns of age-related change in microstructure. Widespread and profound reduction in MT contrasted with local FA decreases paralleled by MD increases. R1 reductions and R2* increases were observed to a smaller extent in overlapping occipito-parietal white matter regions. We interpret our findings, based on current biophysical models, as a fingerprint of age-dependent brain atrophy and underlying microstructural changes in myelin, iron deposits and water. The VBQ approach we present allows for systematic unbiased exploration of the interaction between imaging parameters and extends current methods for detection of neurodegenerative processes in the brain. The demonstrated parameter-specific distribution patterns offer insights into age-related brain structure changes in vivo and provide essential baseline data for studying disease against a background of healthy ageing.
►High-resolution FLASH-based parameter mapping is suitable for clinical purposes. ►Patterns of age-dependent parameter changes reflect specificity to tissue properties. ►Combining VBM and VBQ offers complementary information about brain architecture.
Treatment of neurodegenerative diseases is likely to be most beneficial in the very early, possibly preclinical stages of degeneration. We explored the usefulness of fully automatic structural MRI ...classification methods for detecting subtle degenerative change. The availability of a definitive genetic test for Huntington disease (HD) provides an excellent metric for judging the performance of such methods in gene mutation carriers who are free of symptoms.
Using the gray matter segment of MRI scans, this study explored the usefulness of a multivariate support vector machine to automatically identify presymptomatic HD gene mutation carriers (PSCs) in the absence of any a priori information. A multicenter data set of 96 PSCs and 95 age- and sex-matched controls was studied. The PSC group was subclassified into three groups based on time from predicted clinical onset, an estimate that is a function of DNA mutation size and age.
Subjects with at least a 33% chance of developing unequivocal signs of HD in 5 years were correctly assigned to the PSC group 69% of the time. Accuracy improved to 83% when regions affected by the disease were selected a priori for analysis. Performance was at chance when the probability of developing symptoms in 5 years was less than 10%.
Presymptomatic Huntington disease gene mutation carriers close to estimated diagnostic onset were successfully separated from controls on the basis of single anatomic scans, without additional a priori information. Prior information is required to allow separation when degenerative changes are either subtle or variable.
There has been recent interest in the application of machine learning techniques to neuroimaging-based diagnosis. These methods promise fully automated, standard PC-based clinical decisions, unbiased ...by variable radiological expertise. We recently used support vector machines (SVMs) to separate sporadic Alzheimer's disease from normal ageing and from fronto-temporal lobar degeneration (FTLD). In this study, we compare the results to those obtained by radiologists. A binary diagnostic classification was made by six radiologists with different levels of experience on the same scans and information that had been previously analysed with SVM. SVMs correctly classified 95% (sensitivity/specificity: 95/95) of sporadic Alzheimer's disease and controls into their respective groups. Radiologists correctly classified 65–95% (median 89%; sensitivity/specificity: 88/90) of scans. SVM correctly classified another set of sporadic Alzheimer's disease in 93% (sensitivity/specificity: 100/86) of cases, whereas radiologists ranged between 80% and 90% (median 83%; sensitivity/specificity: 80/85). SVMs were better at separating patients with sporadic Alzheimer's disease from those with FTLD (SVM 89%; sensitivity/specificity: 83/95; compared to radiological range from 63% to 83%; median 71%; sensitivity/specificity: 64/76). Radiologists were always accurate when they reported a high degree of diagnostic confidence. The results show that well-trained neuroradiologists classify typical Alzheimer's disease-associated scans comparable to SVMs. However, SVMs require no expert knowledge and trained SVMs can readily be exchanged between centres for use in diagnostic classification. These results are encouraging and indicate a role for computerized diagnostic methods in clinical practice.
Movement-related brain activation patterns after subcortical stroke are characterized by relative overactivations in cortical motor areas compared with controls. In patients able to perform a motor ...task, overactivations are greater in those with more motor impairment. We hypothesized that recruitment of motor regions would shift from primary to secondary motor networks in response to impaired functional integrity of the corticospinal system (CSS). We measured the magnitude of brain activation using functional MRI during a motor task in eight chronic subcortical stroke patients. CSS functional integrity was assessed using transcranial magnetic stimulation to obtain stimulus/response curves for the affected first dorsal interosseus muscle, with a shallower gradient representing increasing disruption of CSS functional integrity. A negative correlation between the gradient of stimulus/response curve and magnitude of task-related brain activation was found in several motor-related regions, including ipsilesional posterior primary motor cortex Brodmann area (BA) 4p, contralesional anterior primary motor cortex (BA 4a), bilateral premotor cortex, supplementary motor area, intraparietal sulcus, dorsolateral prefrontal cortex and contralesional superior cingulate sulcus. There were no significant positive correlations in any brain region. These results suggest that impaired functional integrity of the CSS is associated with recruitment of secondary motor networks in both hemispheres in an attempt to generate motor output to spinal cord motoneurons. Secondary motor regions are less efficient at generating motor output so this reorganization can only be considered partially successful in reducing motor impairment after stroke.
Machine learning and pattern recognition methods have been used to diagnose Alzheimer's disease (AD) and mild cognitive impairment (MCI) from individual MRI scans. Another application of such methods ...is to predict clinical scores from individual scans. Using relevance vector regression (RVR), we predicted individuals' performances on established tests from their MRI T1 weighted image in two independent data sets. From Mayo Clinic, 73 probable AD patients and 91 cognitively normal (CN) controls completed the Mini-Mental State Examination (MMSE), Dementia Rating Scale (DRS), and Auditory Verbal Learning Test (AVLT) within 3months of their scan. Baseline MRI's from the Alzheimer's disease Neuroimaging Initiative (ADNI) comprised the other data set; 113 AD, 351 MCI, and 122 CN subjects completed the MMSE and Alzheimer's Disease Assessment Scale—Cognitive subtest (ADAS-cog) and 39 AD, 92 MCI, and 32 CN ADNI subjects completed MMSE, ADAS-cog, and AVLT. Predicted and actual clinical scores were highly correlated for the MMSE, DRS, and ADAS-cog tests (P<0.0001). Training with one data set and testing with another demonstrated stability between data sets. DRS, MMSE, and ADAS-Cog correlated better than AVLT with whole brain grey matter changes associated with AD. This result underscores their utility for screening and tracking disease. RVR offers a novel way to measure interactions between structural changes and neuropsychological tests beyond that of univariate methods. In clinical practice, we envision using RVR to aid in diagnosis and predict clinical outcome.
It is controversial whether the dorsolateral prefrontal cortex is involved in the maintenance of items in working memory or in the selection of responses. We used event-related functional magnetic ...resonance imaging to study the performance of a spatial working memory task by humans. We distinguished the maintenance of spatial items from the selection of an item from memory to guide a response. Selection, but not maintenance, was associated with activation of prefrontal area 46 of the dorsal lateral prefrontal cortex. In contrast, maintenance was associated with activation of prefrontal area 8 and the intraparietal cortex. The results support a role for the dorsal prefrontal cortex in the selection of representations. This accounts for the fact that this area is activated both when subjects select between items on working memory tasks and when they freely select between movements on tasks of willed action.
Recovery of motor function after stroke may occur over weeks or months and is often attributed to neuronal reorganization. Functional imaging studies investigating patients who have made a good ...recovery after stroke have suggested that recruitment of other motor‐related networks underlies this recovery. However, patients with less complete recovery have rarely been studied, or else the degree of recovery has not been taken into account. We set out to investigate the relationship between the degree of recovery after stroke and the pattern of recruitment of brain regions during a motor task as measured using functional MRI. We recruited 20 patients who were at least 3 months after their first ever stroke, and 26 right‐handed age‐matched control subjects. None of our patients had infarcts involving the hand region of the primary motor cortex. All subjects were scanned whilst performing an isometric, dynamic visually paced handgrip task. The degree of functional recovery of each patient was assessed using a battery of outcome measures. Single‐patient versus control group analysis revealed that patients with poor recovery were more likely to recruit a number of motor‐related brain regions over and above those seen in the control group during the motor task, whereas patients with more complete recovery were more likely to have ‘normal’ task‐related brain activation. Across the whole patient group and across stroke subtypes, we were able to demonstrate a negative correlation between outcome and the degree of task‐related activation in regions such as the supplementary motor area, cingulate motor areas, premotor cortex, posterior parietal cortex, and cerebellum. This negative correlation was also seen in parts of both contralateral and ipsilateral primary motor cortex. These results further our understanding of the recovery process by demonstrating for the first time a clear relationship between task‐related activation of the motor system and outcome after stroke.
Event‐related functional MRI (fMRI) was used to investigate the neural correlates of memory encoding as a function of age. While fMRI data were obtained, 14 younger (mean age 21 years) and 14 older ...subjects (mean age 68 years) made animacy decisions about words. Recognition memory for these words was tested at two delays such that older subjects’ performance at the short delay was comparable to that of the young subjects at the long delay. This allowed age‐associated changes in the neural correlates of encoding to be dissociated from the correlates of differential recognition performance. Activity in left inferior prefrontal cortex and the left hippocampal formation was greater for subsequently recognized words in both age groups, consistent with the findings of previous studies in young adults. In the prefrontal cortex, these ‘subsequent memory effects’ were, however, left‐lateralized in the younger group but bilateral in the older subjects. In addition, for the younger group only, greater activity for remembered words was observed in anterior inferior temporal cortex, as were reversed effects (‘subsequent forgetting’ effects) in anterior prefrontal regions. The data indicate that older subjects engage much of the same neural circuitry as younger subjects when encoding new memories. However, the findings also point to age‐related differences in both prefrontal and temporal activity during successful episodic encoding.
Recovery of motor function after stroke may occur over weeks or months and is often attributed to cerebral reorganization. We have investigated the longitudinal relationship between recovery after ...stroke and task‐related brain activation during a motor task as measured using functional MRI (fMRI). Eight first‐ever stroke patients presenting with hemiparesis resulting from cerebral infarction sparing the primary motor cortex, and four control subjects were recruited. Subjects were scanned on a number of occasions whilst performing an isometric dynamic visually paced hand grip task. Recovery in the patient group was assessed using a battery of outcome measures at each time point. Task‐related brain activations decreased over sessions as a function of recovery in a number of primary and non‐primary motor regions in all patients, but no session effects were seen in the controls. Furthermore, consistent decreases across sessions correlating with recovery were seen across the whole patient group independent of rate of recovery or initial severity, in primary motor cortex, premotor and prefrontal cortex, supplementary motor areas, cingulate sulcus, temporal lobe, striate cortex, cerebellum, thalamus and basal ganglia. Although recovery‐related increases were seen in different brain regions in four patients, there were no consistent effects across the group. These results further our understanding of the recovery process by demonstrating for the first time a clear temporal relationship between recovery and task‐related activation of the motor system after stroke.