Seminar Papers

[Article] Biomarker identification through integrating fmri and epigenetics

Summary: They combined linear regression with CCA in a coupled manner to extract discriminative features for schizophrenia that are co-expressed in the fMRI and DNA methylation data (epigenetic data).

Bai, Yuntong, et al. “Biomarker identification through integrating fmri and epigenetics.” IEEE transactions on Biomedical Engineering 67.4 (2019): 1186-1196.

[Article] Brain structure is linked to the association between family environment and behavioral problems in children in the ABCD study

Summary: In this article, they tried to find a relationship between family conflict, parental monitoring and children’s behavior problems and cognitive scores. They utilized sMRI, CBCL and KSADS to get 20 behavioral problems scores and 10 cognitive scores. They visualized correlation between these criteria and showed structural differences according to these scores. Furthermore, they performed longitudinal association analysis with baseline & 1 year later data. Utilizing CBCL scores into account in my study later seems reasonable.

Gong, et al. Brain structure is linked to the association between family environment and behavioral problems in children in the ABCD study. Nat Commun 12, 3769 (2021)

[Article] Learning patterns of the ageing brain in MRI using deep convolutional networks

Summary: Summary: In this paper, the authors predicted the brain age of T1-weighted MRI images in UK Biobank dataset using 3D CNN architecture. They examined the relationship between predicted ages and UK Biobank variables which are categorized into lifestyle factors (exercise, alcohol, tobacco, …), physiological/medical measurements (bone, cardiac, eye, …), and medical history. They adopted attention gates as attention modules to observe activation of the CNN model, and they also visualized the model using saliency map for comparison.

Dinsdale, Nicola K., et al. “Learning patterns of the ageing brain in MRI using deep convolutional networks.” Neuroimage 224 (2021): 117401.

[Article] Dynamic functional connectivity markers of objective trait mindfulness.

Summary: The goal of study is to identify the differences in dynamic functional connectivity according to the trait mindfulness. They found the high mindfulness trait group showed more strong within-network connectivity in the DMN and the salience network than the low mindfulness trait group.

Lim, J., Teng, J., Patanaik, A., Tandi, J., & Massar, S. A. (2018). Dynamic functional connectivity markers of objective trait mindfulness. NeuroImage, 176, 193-202.

[Article] Integrated 3D motion analysis with functional magnetic resonance neuroimaging to identify neural correlates of lower extremity movement.

Summary: Collected knee biomechanics.using an MRI-compatible motion capture system. The study shows promise for the MRI-compatible system to capture lower-extremity biomechanical data collected concurrently during fMRI, and the present data identified potentially unique neural drivers of aberrant biomechanics.

Anand, M., Diekfuss, J. A., Slutsky-Ganesh, A. B., Grooms, D. R., Bonnette, S., Foss, K. D. B., … & Myer, G. D. (2021). Integrated 3D motion analysis with functional magnetic resonance neuroimaging to identify neural correlates of lower extremity movement. Journal of Neuroscience Methods, 355, 109108.

[Article] Deep learning-based unlearning of dataset bias for MRI harmonisation and confound removal.

Summary: The authors proposed a deep learning based training scheme, inspired by domain adaptation techniques. It is similar to my current study, but this study proposed a consecutive training process since they assumed that they may not know main task label (i.e., prediction target such as age, gender, p-factor) even they can know all available domain label (e.g., site, scanner). It would be good to present this paper because this paper has a similar, but different point of view for dealing with the scanner-effect.

Dinsdale, N. K., Jenkinson, M., & Namburete, A. I. (2021). Deep learning-based unlearning of dataset bias for MRI harmonisation and confound removal. NeuroImage, 228, 117689.

[Article] A map of object space in primate inferotemporal cortex.

Summary: The authors investigated the (category-selective) organization of macaque IT cortex using fMRI, microstimulation, electrophysiology. A total of 51 objects each presented at 24 views belonging to 6 different categories. The fc6 output of AlexNet for object classification was PCA decomposed and used as an object space. Pooling responses across face, body, NML, and stubby networks enabled reasonable reconstruction of general objects demonstrating the importance of these four networks in object representation.

Bao, P., She, L., McGill, M., & Tsao, D. Y. (2020). A map of object space in primate inferotemporal cortex. Nature, 583(7814), 103-108.

[Article] Cross-Hemispheric Complementary Prefrontal Mechanisms during Task Switching under Perceptual Uncertainty.

Summary: This paper examines neural mechanisms for task switching in which task-relevant information involved perceptual uncertainty. As a result, the lateral prefrontal cortex (PFC) in the left hemisphere is associated with behavioral flexibility, and the lateral prefrontal cortex (PFC) in the right hemisphere is associated with the perception of ambiguous stimuli.

Tsumura, Aoki, et al. (2021). “Cross-Hemispheric Complementary Prefrontal Mechanisms during Task Switching under Perceptual Uncertainty.” The Journal of Neuroscience, 41(10), 2197–2213.

[Article] Canonical Correlation Analysis of Imaging Genetics Data Based on Statistical Independence and Structural Sparsity.

Summary: They proposed Independence and Structural sparsity canonical correlation analysis (ISCCA) for imaging genetics study. They combined ICA and CCA to reduce the collinear effects, which also incorporate graph structure of the data into the model to improve the accuracy of feature selection. This method helps identify risk genes and abnormal brain regions in schizophrenia.

Zhang, Yipu, et al. “Canonical Correlation Analysis of Imaging Genetics Data Based on Statistical Independence and Structural Sparsity.” IEEE journal of biomedical and health informatics 24.9 (2020): 2621-2629.

[Article] Deriving disease modules from the compressed transcriptional space embedded in a deep autoencoder.

Summary: Using deep neural networks(DNN) to classify diseases using genetic data is popular these days. In this article, they hypothesized that disease-relevant modules of genes can be discovered within the autoencoder (AE) representations. They compared shallow and deep AE with various node sizes, and showed deep AE works better. Also, they also showed that each different layer captures gradients of biology. By overlapping top 1000 genes for each disease with GWAS, they found a highly significant association for at least one layer in all tested diseases.

Dwivedi, S. K., Tjärnberg, A., Tegnér, J., & Gustafsson, M. (2020). Deriving disease modules from the compressed transcriptional space embedded in a deep autoencoder. Nature communications, 11(1), 1-10.