RESUMO
Convolutional neural networks (CNNs) have been increasingly used in the computer-aided diagnosis of Alzheimer's Disease (AD). This study takes the advantage of the 2D-slice CNN fast computation and ensemble approaches to develop a Monte Carlo Ensemble Neural Network (MCENN) by introducing Monte Carlo sampling and an ensemble neural network in the integration with ResNet50. Our goals are to improve the 2D-slice CNN performance and to design the MCENN model insensitive to image resolution. Unlike traditional ensemble approaches with multiple base learners, our MCENN model incorporates one neural network learner and generates a large number of possible classification decisions via Monte Carlo sampling of feature importance within the combined slices. This can overcome the main weakness of the lack of 3D brain anatomical information in 2D-slice CNNs and develop a neural network to learn the 3D relevance of the features across multiple slices. Brain images from Alzheimer's Disease Neuroimaging Initiative (ADNI, 7199 scans), the Open Access Series of Imaging Studies-3 (OASIS-3, 1992 scans), and a clinical sample (239 scans) are used to evaluate the performance of the MCENN model for the classification of cognitively normal (CN), patients with mild cognitive impairment (MCI) and AD. Our MCENN with a small number of slices and minimal image processing (rigid transformation, intensity normalization, skull stripping) achieves the AD classification accuracy of 90%, better than existing 2D-slice CNNs (accuracy: 63%â¼84%) and 3D CNNs (accuracy: 74%â¼88%). Furthermore, the MCENN is robust to be trained in the ADNI dataset and applied to the OASIS-3 dataset and the clinical sample. Our experiments show that the AD classification accuracy of the MCENN model is comparable when using high- and low-resolution brain images, suggesting the insensitivity of the MCENN to image resolution. Hence, the MCENN does not require high-resolution 3D brain structural images and comprehensive image processing, which supports its potential use in a clinical setting.
Assuntos
Doença de Alzheimer , Disfunção Cognitiva , Humanos , Doença de Alzheimer/diagnóstico por imagem , Imageamento por Ressonância Magnética/métodos , Redes Neurais de Computação , Neuroimagem/métodos , Diagnóstico por Computador , Disfunção Cognitiva/diagnóstico por imagemRESUMO
Motion-related artifacts are one of the major challenges associated with pediatric neuroimaging. Recent studies have shown a relationship between visual quality ratings of T1 images and cortical reconstruction measures. Automated algorithms offer more precision in quantifying movement-related artifacts compared to visual inspection. Thus, the goal of this study was to test three different automated quality assessment algorithms for structural MRI scans. The three algorithms included a Fourier-, integral-, and a gradient-based approach which were run on raw T1 -weighted imaging data collected from four different scanners. The four cohorts included a total of 6,662 MRI scans from two waves of the Generation R Study, the NIH NHGRI Study, and the GUSTO Study. Using receiver operating characteristics with visually inspected quality ratings of the T1 images, the area under the curve (AUC) for the gradient algorithm, which performed better than either the integral or Fourier approaches, was 0.95, 0.88, and 0.82 for the Generation R, NHGRI, and GUSTO studies, respectively. For scans of poor initial quality, repeating the scan often resulted in a better quality second image. Finally, we found that even minor differences in automated quality measurements were associated with FreeSurfer derived measures of cortical thickness and surface area, even in scans that were rated as good quality. Our findings suggest that the inclusion of automated quality assessment measures can augment visual inspection and may find use as a covariate in analyses or to identify thresholds to exclude poor quality data.
Assuntos
Artefatos , Imageamento por Ressonância Magnética , Reconhecimento Automatizado de Padrão , Garantia da Qualidade dos Cuidados de Saúde/métodos , Algoritmos , Área Sob a Curva , Encéfalo/anatomia & histologia , Encéfalo/diagnóstico por imagem , Encéfalo/crescimento & desenvolvimento , Criança , Pré-Escolar , Estudos de Coortes , Feminino , Humanos , Imageamento por Ressonância Magnética/métodos , Masculino , Movimento (Física) , Tamanho do Órgão , Reconhecimento Automatizado de Padrão/métodos , Curva ROCRESUMO
Purpose: The purpose of this study was to improve standardized language assessments among bilingual toddlers by investigating and removing the effects of bias due to unfamiliarity with cultural norms or a distributed language system. Method: The Expressive and Receptive Bayley-III language scales were adapted for use in a multilingual country (Singapore). Differential item functioning (DIF) was applied to data from 459 two-year-olds without atypical language development. This involved investigating if the probability of success on each item varied according to language exposure while holding latent language ability, gender, and socioeconomic status constant. Associations with language, behavioral, and emotional problems were also examined. Results: Five of 16 items showed DIF, 1 of which may be attributed to cultural bias and another to a distributed language system. The remaining 3 items favored toddlers with higher bilingual exposure. Removal of DIF items reduced associations between language scales and emotional and language problems, but improved the validity of the expressive scale from poor to good. Conclusions: Our findings indicate the importance of considering cultural and distributed language bias in standardized language assessments. We discuss possible mechanisms influencing performance on items favoring bilingual exposure, including the potential role of inhibitory processing.
Assuntos
Cultura , Transtornos do Desenvolvimento da Linguagem/diagnóstico , Testes de Linguagem , Multilinguismo , Pré-Escolar , Análise Fatorial , Feminino , Humanos , Transtornos do Desenvolvimento da Linguagem/psicologia , Estudos Longitudinais , Masculino , Análise de Regressão , Reprodutibilidade dos Testes , Singapura , Fatores SocioeconômicosRESUMO
This study included 168 and 85 mother-infant dyads from Asian and United States of America cohorts to examine whether a genomic profile risk score for major depressive disorder (GPRSMDD) moderates the association between antenatal maternal depressive symptoms (or socio-economic status, SES) and fetal neurodevelopment, and to identify candidate biological processes underlying such association. Both cohorts showed a significant interaction between antenatal maternal depressive symptoms and infant GPRSMDD on the right amygdala volume. The Asian cohort also showed such interaction on the right hippocampal volume and shape, thickness of the orbitofrontal and ventromedial prefrontal cortex. Likewise, a significant interaction between SES and infant GPRSMDD was on the right amygdala and hippocampal volumes and shapes. After controlling for each other, the interaction effect of antenatal maternal depressive symptoms and GPRSMDD was mainly shown on the right amygdala, while the interaction effect of SES and GPRSMDD was mainly shown on the right hippocampus. Bioinformatic analyses suggested neurotransmitter/neurotrophic signaling, SNAp REceptor complex, and glutamate receptor activity as common biological processes underlying the influence of antenatal maternal depressive symptoms on fetal cortico-limbic development. These findings suggest gene-environment interdependence in the fetal development of brain regions implicated in cognitive-emotional function. Candidate biological mechanisms involve a range of brain region-specific signaling pathways that converge on common processes of synaptic development.
Assuntos
Mapeamento Encefálico , Encéfalo/crescimento & desenvolvimento , Encéfalo/patologia , Transtorno Depressivo Maior/patologia , Relações Materno-Fetais , Classe Social , Povo Asiático , Encéfalo/diagnóstico por imagem , Estudos de Coortes , Biologia Computacional , Transtorno Depressivo Maior/diagnóstico por imagem , Transtorno Depressivo Maior/genética , Transtorno Depressivo Maior/psicologia , Feminino , Desenvolvimento Fetal/genética , Redes Reguladoras de Genes/fisiologia , Genótipo , Idade Gestacional , Humanos , Processamento de Imagem Assistida por Computador , Recém-Nascido , Imageamento por Ressonância Magnética , Masculino , Polimorfismo de Nucleotídeo Único/genética , Gravidez , Efeitos Tardios da Exposição Pré-NatalRESUMO
BACKGROUND: Breastfeeding has been shown to enhance global measures of intelligence in children. However, few studies have examined associations between breastfeeding and specific cognitive task performance in the first 2 y of life, particularly in an Asian population. OBJECTIVE: We assessed associations between early infant feeding and detailed measures of cognitive development in the first 2 y of life in healthy Asian children born at term. DESIGN: In a prospective cohort study, neurocognitive testing was performed in 408 healthy children (aged 6, 18, and 24 mo) from uncomplicated pregnancies (i.e., birth weight >2500 and <4000 g, gestational age ≥37 wk, and 5-min Apgar score ≥9). Tests included memory (deferred imitation, relational binding, habituation) and attention tasks (visual expectation, auditory oddball) as well as the Bayley Scales of Infant and Toddler Development, Third Edition (BSID-III). Children were stratified into 3 groups (low, intermediate, and high) on the basis of breastfeeding duration and exclusivity. RESULTS: After potential confounding variables were controlled for, significant associations and dose-response relations were observed for 4 of the 15 tests. Higher breastfeeding exposure was associated with better memory at 6 mo, demonstrated by greater preferential looking toward correctly matched items during early portions of a relational memory task (i.e., relational binding task: P-trend = 0.015 and 0.050 for the first two 1000-ms time bins, respectively). No effects of breastfeeding were observed at 18 mo. At 24 mo, breastfed children were more likely to display sequential memory during a deferred imitation memory task (P-trend = 0.048), and toddlers with more exposure to breastfeeding scored higher in receptive language [+0.93 (0.23, 1.63) and +1.08 (0.10, 2.07) for intermediate- and high-breastfeeding groups, respectively, compared with the low-breastfeeding group], as well as expressive language [+0.58 (-0.06, 1.23) and +1.22 (0.32, 2.12) for intermediate- and high-breastfeeding groups, respectively] assessed via the BSID-III. CONCLUSIONS: Our findings suggest small but significant benefits of breastfeeding for some aspects of memory and language development in the first 2 y of life, with significant improvements in only 4 of 15 indicators. Whether the implicated processes confer developmental advantages is unknown and represents an important area for future research. This trial was registered at www.clinicaltrials.gov as NCT01174875.