Cortical and Subcortical Contributions to Predicting Intelligence Using 3D ConvNets

Zou, Yukai; Jang, Ikbeom; Reese, Timothy G.; Yao, Jinxia Fiona; Zhu, Wenbin; Rispoli, Joseph V.

doi:10.1007/978-3-030-31901-4_21

Cited by 7 publications

(12 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Those deep features are extracted from convolutions of images with filters (3×3×3, 5×5×5, or other sizes). Several studies 60,66,[107][108][109][110] used a convolutional neural network (CNN), a specific type of image-based deep learning technique, on T1-MRI to predict fluid intelligence in adolescents. They predicted the residual fluid intelligence score of more than 4500 adolescents with an MSE ranging from 92 to 103 (for a range of true residual fluid intelligence score of [-40, 30]), as summarized in Table 6 of the supplementary materials.…”

Section: Structural Mri To Infer Intelligence and Neurocognitionmentioning

confidence: 99%

“…A potential solution is to choose brain regions beforehand and those regions to deep learning models. For example, Zou et al 109 used regions from bilateral transverse temporal gyri (BAs 41, 42), bilateral thalamus, left parahippocampal gyrus (BA 34), left hippocampus, right opercular part of inferior frontal gyrus (BAs 44, 45, 47), left anterior cingulate gyrus (BAs 24, 32, 33), right amygdala, left lingual gyrus (BA 19), left superior parietal lobule (BA 7), right inferior parietal lobule (BAs 39, 40), left angular gyrus (BA 39), left paracentral lobule, and left caudate nucleus (BAs 1-4) in their deep learning model to predict gF score. However, the choice of such regions may be subjective, the accuracy of prediction was not significantly different from inputting the whole image, and treating regions separately may miss the opportunity to consider those regions jointly in the convolutions.…”

Section: Structural Mri To Infer Intelligence and Neurocognitionmentioning

confidence: 99%

See 1 more Smart Citation

Inferring Neurocognition and Intelligence using Brain MRI

Hussain¹,

Grant²,

Ou³

2023

Preprint

View full text Add to dashboard Cite

Brain magnetic resonance imaging (MRI) offers a unique lens to study neuroanatomic support of human neurocognition and intelligence. A core mystery is the MRI explanation of individual differences in neurocognition and intelligence. The past four decades have seen great advancement in studying this century-long mystery, but the sample size and population-level studies limit the explanation at the individual level. The recent rise of big data and artificial intelligence offers novel opportunities. Yet, data sources, harmonization, study design, and interpretation need to be carefully considered. This review aims to summarize past work, discuss rising opportunities and challenges, and facilitate further investigations on machine intelligence inferring human intelligence.

show abstract

Section: Structural Mri To Infer Intelligence and Neurocognitionmentioning

confidence: 99%

Section: Structural Mri To Infer Intelligence and Neurocognitionmentioning

confidence: 99%

Inferring Neurocognition and Intelligence using Brain MRI

Hussain¹,

Grant²,

Ou³

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Second, the post-processing stage only uses the median predicted scores as the final prediction result, or extracts the high-level feature map information for regression, which causes information loss in the pre-processing stage. Finally, the fluid intelligence score regression operation is performed for only one ROI at a time, which ignores the interaction between other brain areas [13].…”

Section: Technical Detailsmentioning

confidence: 99%

“…To a certain extent, the model was in a state of overfitting, and was found to lack generalization ability. In the method proposed by Yukai Zou et al [13], multiple brain regions were selected to predict the residualized fluid intelligence scores using a 3D CNN, but the median predicted score was used as the final prediction result, which caused a substantial amount of information loss in the pre-processing stage. Moreover, the regression operation of the fluid intelligence scores was performed for only one ROI at a time, which ignored the interaction between other brain areas.…”

Section: Plos Onementioning

confidence: 99%

“…The use of traditional deep learning methods for fluid intelligence score regression is characterized by the following weaknesses: (1) the structure of the segmentation model is not well optimized: first, the high-layer and low-layer features of the segmentation framework are not fused [13]. As a result, a substantial amount of spatial information in the image is lost in the lower layer; second, the attention mechanism is not introduced into the segmentation model.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Prediction of fluid intelligence from T1-w MRI images: A precise two-step deep learning framework

Jiang²,

Zhang³

et al. 2022

PLoS ONE

View full text Add to dashboard Cite

The Adolescent Brain Cognitive Development (ABCD) Neurocognitive Prediction Challenge (ABCD-NP-Challenge) is a community-driven competition that challenges competitors to develop algorithms to predict fluid intelligence scores from T1-w MRI images. In this work, a two-step deep learning pipeline is proposed to improve the prediction accuracy of fluid intelligence scores. In terms of the first step, the main contributions of this study include the following: (1) the concepts of the residual network (ResNet) and the squeeze-and-excitation network (SENet) are utilized to improve the original 3D U-Net; (2) in the segmentation process, the pixels in symmetrical brain regions are assigned the same label; (3) to remove redundant background information from the segmented regions of interest (ROIs), a minimum bounding cube (MBC) is used to enclose the ROIs. This new segmentation structure can greatly improve the segmentation performance of the ROIs in the brain as compared with the classical convolutional neural network (CNN), which yields a Dice coefficient of 0.8920. In the second stage, MBCs are used to train neural network regression models for enhanced nonlinearity. The fluid intelligence score prediction results of the proposed method are found to be superior to those of current state-of-the-art approaches, and the proposed method achieves a mean square error (MSE) of 82.56 on a test data set, which reflects a very competitive performance.

show abstract

Predictive models demonstrate age‐dependent association of subcortical volumes and cognitive measures

Weerasekera

Ion-Mărgineanu

Green

et al. 2022

Human Brain Mapping

View full text Add to dashboard Cite

Whether brain matter volume is correlated with cognitive functioning and higher intelligence is controversial. We explored this relationship by analysis of data collected on 193 healthy young and older adults through the "Leipzig Study for Mind-Body-Emotion Interactions" (LEMON) study. Our analysis involved four cognitive measures: fluid intelligence, crystallized intelligence, cognitive flexibility, and working memory. Brain subregion volumes were determined by magnetic resonance imaging.We normalized each subregion volume to the estimated total intracranial volume and conducted training simulations to compare the predictive power of normalized volumes of large regions of the brain (i.e., gray matter, cortical white matter, and cerebrospinal fluid), normalized subcortical volumes, and combined normalized volumes of large brain regions and normalized subcortical volumes. Statistical tests showed significant differences in the performance accuracy and feature importance of the subregion volumes in predicting cognitive skills for young and older adults. Random forest feature selection analysis showed that cortical white matter was the key feature in predicting fluid intelligence in both young and older adults. In young adults, crystallized intelligence was best predicted by caudate nucleus, thalamus, pallidum, and nucleus accumbens volumes, whereas putamen, amygdala, nucleus accumbens, and hippocampus volumes were selected for older adults. Cognitive flexibility was best predicted by the caudate, nucleus accumbens, and hippocampus in young adults and caudate and amygdala in older adults. Finally, working memory was best predicted by the putamen, pallidum, and nucleus accumbens in the younger group, whereas amygdala and hippocampus volumes were predictive in the older group. Thus, machine learning predictive models demonstrated an age-dependent association between subcortical volumes and cognitive measures. These approaches may be useful in predicting the likelihood of age-related cognitive decline and in testing of approaches for targeted improvement of cognitive functioning in older adults.Akila Weerasekera and Adrian Ion-M argineanus contributed equally to this study.

show abstract

Cortical and Subcortical Contributions to Predicting Intelligence Using 3D ConvNets

Cited by 7 publications

References 18 publications

Inferring Neurocognition and Intelligence using Brain MRI

Inferring Neurocognition and Intelligence using Brain MRI

Prediction of fluid intelligence from T1-w MRI images: A precise two-step deep learning framework

Predictive models demonstrate age‐dependent association of subcortical volumes and cognitive measures

Contact Info

Product

Resources

About