GAT-LI: a graph attention network based learning and interpreting method for functional brain network classification

GraphPath: a graph attention model for molecular stratification with interpretability based on the pathway-pathway interaction network

MOTIVATION

Studying the molecular heterogeneity of cancer is essential for achieving personalized therapy. At the same time, understanding the biological processes that drive cancer development can lead to the identification of valuable therapeutic targets. Therefore, achieving accurate and interpretable clinical predictions requires paramount attention to thoroughly characterizing patients at both the molecular and biological pathway levels.

RESULTS

Here, we present GraphPath, a biological knowledge-driven graph neural network with multi-head self-attention mechanism that implements the pathway-pathway interaction network. We train GraphPath to classify the cancer status of patients with prostate cancer based on their multi-omics profiling. Experiment results show that our method outperforms P-NET and other baseline methods. Besides, two external cohorts are used to validate that the model can be generalized to unseen samples with adequate predictive performance. We reduce the dimensionality of latent pathway embeddings and visualize corresponding classes to further demonstrate the optimal performance of the model. Additionally, since GraphPath's predictions are interpretable, we identify target cancer-associated pathways that significantly contribute to the model's predictions. Such a robust and interpretable model has the potential to greatly enhance our understanding of cancer's biological mechanisms and accelerate the development of targeted therapies.

AVAILABILITY AND IMPLEMENTATION

https://github.com/amazingma/GraphPath.

Brain network classification based on dynamic graph attention information bottleneck

BACKGROUND AND OBJECTIVES

Graph neural networks (GNN) have demonstrated remarkable encoding capabilities in the context of brain network classification tasks. They excel at uncovering hidden static connections between brain states. However, brain network signals can be influenced by physiological traits and external variables during clinical detection, resulting in noisy brain graphs. Additionally, many existing algorithms for brain networks primarily focus on static topologies determined by threshold-based criteria, thereby overlooking the real-time variability in brain channel connectivity. These sources of noise and the persistence of static structures inevitably hinder the effective exchange of information during brain network computations.

METHODS

To address these challenges, we propose a novel framework called the dynamic graph attention information bottleneck (DGAIB). This framework is designed to dynamically enhance the input raw brain graph structure from the perspective of information theory and graph theory. First, we employ the Spearman function to construct a raw graph. Then, we use a graph information bottleneck (GIB) to optimize the internal graph connections by selectively masking redundant feature embeddings. Finally, we enhance the feature aggregation of each brain state by utilizing a graph attention network (GAT), which promotes improved information exchange among distinct brain regions within the model. These processed representations serve as input for subsequent classification tasks.

EXPERIMENT AND RESULTS

We systematically evaluated the robustness and generalizability of our proposed framework through a series of experiments. This evaluation included patient-specific experiments using the electroencephalography (EEG)-based CHB-MIT dataset and cross-patient experiments leveraging the functional magnetic resonance imaging (fMRI)-based ABIDE-I dataset from multiple perspectives.

Diagnosis of Autism Spectrum Disorder (ASD) Using Recursive Feature Elimination-Graph Neural Network (RFE-GNN) and Phenotypic Feature Extractor (PFE)

Autism spectrum disorder (ASD) poses as a multifaceted neurodevelopmental condition, significantly impacting children's social, behavioral, and communicative capacities. Despite extensive research, the precise etiological origins of ASD remain elusive, with observable connections to brain activity. In this study, we propose a novel framework for ASD detection, extracting the characteristics of functional magnetic resonance imaging (fMRI) data and phenotypic data, respectively. Specifically, we employ recursive feature elimination (RFE) for feature selection of fMRI data and subsequently apply graph neural networks (GNN) to extract informative features from the chosen data. Moreover, we devise a phenotypic feature extractor (PFE) to extract phenotypic features effectively. We then, synergistically fuse the features and validate them on the ABIDE dataset, achieving 78.7% and 80.6% accuracy, respectively, thereby showcasing competitive performance compared to state-of-the-art methods. The proposed framework provides a promising direction for the development of effective diagnostic tools for ASD.

Do it the transformer way: A comprehensive review of brain and vision transformers for autism spectrum disorder diagnosis and classification

Autism spectrum disorder (ASD) is a condition observed in children who display abnormal patterns of interaction, behavior, and communication with others. Despite extensive research efforts, the underlying causes of this neurodevelopmental disorder and its biomarkers remain unknown. However, advancements in artificial intelligence and machine learning have improved clinicians' ability to diagnose ASD. This review paper investigates various MRI modalities to identify distinct features that characterize individuals with ASD compared to typical control subjects. The review then moves on to explore deep learning models for ASD diagnosis, including convolutional neural networks (CNNs), autoencoders, graph convolutions, attention networks, and other models. CNNs and their variations are particularly effective due to their capacity to learn structured image representations and identify reliable biomarkers for brain disorders. Computer vision transformers often employ CNN architectures with transfer learning techniques like fine-tuning and layer freezing to enhance image classification performance, surpassing traditional machine learning models. This review paper contributes in three main ways. Firstly, it provides a comprehensive overview of a recommended architecture for using vision transformers in the systematic ASD diagnostic process. To this end, the paper investigates various pre-trained vision architectures such as VGG, ResNet, Inception, InceptionResNet, DenseNet, and Swin models that were fine-tuned for ASD diagnosis and classification. Secondly, it discusses the vision transformers of 2020th like BiT, ViT, MobileViT, and ConvNeXt, and applying transfer learning methods in relation to their prospective practicality in ASD classification. Thirdly, it explores brain transformers that are pre-trained on medically rich data and MRI neuroimaging datasets. The paper recommends a systematic architecture for ASD diagnosis using brain transformers. It also reviews recently developed brain transformer-based models, such as METAFormer, Com-BrainTF, Brain Network, ST-Transformer, STCAL, BolT, and BrainFormer, discussing their deep transfer learning architectures and results in ASD detection. Additionally, the paper summarizes and discusses brain-related transformers for various brain disorders, such as MSGTN, STAGIN, and MedTransformer, in relation to their potential usefulness in ASD. The study suggests that developing specialized transformer-based models, following the success of natural language processing (NLP), can offer new directions for image classification problems in ASD brain biomarkers learning and classification. By incorporating the attention mechanism, treating MRI modalities as sequence prediction tasks trained on brain disorder classification problems, and fine-tuned on ASD datasets, brain transformers can show a great promise in ASD diagnosis.

Multi-Slice Generation sMRI and fMRI for Autism Spectrum Disorder Diagnosis Using 3D-CNN and Vision Transformers

Researchers have explored various potential indicators of ASD, including changes in brain structure and activity, genetics, and immune system abnormalities, but no definitive indicator has been found yet. Therefore, this study aims to investigate ASD indicators using two types of magnetic resonance images (MRI), structural (sMRI) and functional (fMRI), and to address the issue of limited data availability. Transfer learning is a valuable technique when working with limited data, as it utilizes knowledge gained from a pre-trained model in a domain with abundant data. This study proposed the use of four vision transformers namely ConvNeXT, MobileNet, Swin, and ViT using sMRI modalities. The study also investigated the use of a 3D-CNN model with sMRI and fMRI modalities. Our experiments involved different methods of generating data and extracting slices from raw 3D sMRI and 4D fMRI scans along the axial, coronal, and sagittal brain planes. To evaluate our methods, we utilized a standard neuroimaging dataset called NYU from the ABIDE repository to classify ASD subjects from typical control subjects. The performance of our models was evaluated against several baselines including studies that implemented VGG and ResNet transfer learning models. Our experimental results validate the effectiveness of the proposed multi-slice generation with the 3D-CNN and transfer learning methods as they achieved state-of-the-art results. In particular, results from 50-middle slices from the fMRI and 3D-CNN showed a profound promise in ASD classifiability as it obtained a maximum accuracy of 0.8710 and F1-score of 0.8261 when using the mean of 4D images across the axial, coronal, and sagittal. Additionally, the use of the whole slices in fMRI except the beginnings and the ends of brain views helped to reduce irrelevant information and showed good performance of 0.8387 accuracy and 0.7727 F1-score. Lastly, the transfer learning with the ConvNeXt model achieved results higher than other transformers when using 50-middle slices sMRI along the axial, coronal, and sagittal planes.

Effects of repetitive transcranial magnetic stimulation and their underlying neural mechanisms evaluated with magnetic resonance imaging-based brain connectivity network analyses

Repetitive transcranial magnetic stimulation (rTMS) is a noninvasive brain modulation and rehabilitation technique used in patients with neuropsychiatric diseases. rTMS can structurally remodel or functionally induce activities of specific cortical regions and has developed to an important therapeutic method in such patients. Magnetic resonance imaging (MRI) provides brain data that can be used as an explanation tool for the neural mechanisms underlying rTMS effects; brain alterations related to different functions or structures may be reflected in changes in the interaction and influence of brain connections within intrinsic specific networks. In this review, we discuss the technical details of rTMS and the biological interpretation of brain networks identified with MRI analyses, comprehensively summarize the neurobiological effects in rTMS-modulated individuals, and elaborate on changes in the brain network in patients with various neuropsychiatric diseases receiving rehabilitation treatment with rTMS. We conclude that brain connectivity network analysis based on MRI can reflect alterations in functional and structural connectivity networks comprising adjacent and separated brain regions related to stimulation sites, thus reflecting the occurrence of intrinsic functional integration and neuroplasticity. Therefore, MRI is a valuable tool for understanding the neural mechanisms of rTMS and practically tailoring treatment plans for patients with neuropsychiatric diseases.

Heterogeneous Graph Convolutional Neural Network via Hodge-Laplacian for Brain Functional Data

This study proposes a novel heterogeneous graph convolutional neural network (HGCNN) to handle complex brain fMRI data at regional and across-region levels. We introduce a generic formulation of spectral filters on heterogeneous graphs by introducing the k-th Hodge-Laplacian (HL) operator. In particular, we propose Laguerre polynomial approximations of HL spectral filters and prove that their spatial localization on graphs is related to the polynomial order. Furthermore, based on the bijection property of boundary operators on simplex graphs, we introduce a generic topological graph pooling (TGPool) method that can be used at any dimensional simplices. This study designs HL-node, HL-edge, and HL-HGCNN neural networks to learn signal representation at a graph node, edge levels, and both, respectively. Our experiments employ fMRI from the Adolescent Brain Cognitive Development (ABCD; n=7693) to predict general intelligence. Our results demonstrate the advantage of the HL-edge network over the HL-node network when functional brain connectivity is considered as features. The HL-HGCNN outperforms the state-of-the-art graph neural networks (GNNs) approaches, such as GAT, BrainGNN, dGCN, BrainNetCNN, and Hypergraph NN. The functional connectivity features learned from the HL-HGCNN are meaningful in interpreting neural circuits related to general intelligence.

Evaluation of Risk of Bias in Neuroimaging-Based Artificial Intelligence Models for Psychiatric Diagnosis: A Systematic Review

IMPORTANCE

Neuroimaging-based artificial intelligence (AI) diagnostic models have proliferated in psychiatry. However, their clinical applicability and reporting quality (ie, feasibility) for clinical practice have not been systematically evaluated.

OBJECTIVE

To systematically assess the risk of bias (ROB) and reporting quality of neuroimaging-based AI models for psychiatric diagnosis.

EVIDENCE REVIEW

PubMed was searched for peer-reviewed, full-length articles published between January 1, 1990, and March 16, 2022. Studies aimed at developing or validating neuroimaging-based AI models for clinical diagnosis of psychiatric disorders were included. Reference lists were further searched for suitable original studies. Data extraction followed the CHARMS (Checklist for Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modeling Studies) and PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines. A closed-loop cross-sequential design was used for quality control. The PROBAST (Prediction Model Risk of Bias Assessment Tool) and modified CLEAR (Checklist for Evaluation of Image-Based Artificial Intelligence Reports) benchmarks were used to systematically evaluate ROB and reporting quality.

FINDINGS

A total of 517 studies presenting 555 AI models were included and evaluated. Of these models, 461 (83.1%; 95% CI, 80.0%-86.2%) were rated as having a high overall ROB based on the PROBAST. The ROB was particular high in the analysis domain, including inadequate sample size (398 of 555 models [71.7%; 95% CI, 68.0%-75.6%]), poor model performance examination (with 100% of models lacking calibration examination), and lack of handling data complexity (550 of 555 models [99.1%; 95% CI, 98.3%-99.9%]). None of the AI models was perceived to be applicable to clinical practices. Overall reporting completeness (ie, number of reported items/number of total items) for the AI models was 61.2% (95% CI, 60.6%-61.8%), and the completeness was poorest for the technical assessment domain with 39.9% (95% CI, 38.8%-41.1%).

CONCLUSIONS AND RELEVANCE

This systematic review found that the clinical applicability and feasibility of neuroimaging-based AI models for psychiatric diagnosis were challenged by a high ROB and poor reporting quality. Particularly in the analysis domain, ROB in AI diagnostic models should be addressed before clinical application.

Technologies to support the diagnosis and/or treatment of neurodevelopmental disorders: A systematic review

In recent years, there has been a great interest in utilizing technology in mental health research. The rapid technological development has encouraged researchers to apply technology as a part of a diagnostic process or treatment of Neurodevelopmental Disorders (NDDs). With the large number of studies being published comes an urgent need to inform clinicians and researchers about the latest advances in this field. Here, we methodically explore and summarize findings from studies published between August 2019 and February 2022. A search strategy led to the identification of 4108 records from PubMed and APA PsycInfo databases. 221 quantitative studies were included, covering a wide range of technologies used for diagnosis and/or treatment of NDDs, with the biggest focus on Autism Spectrum Disorder (ASD). The most popular technologies included machine learning, functional magnetic resonance imaging, electroencephalogram, magnetic resonance imaging, and neurofeedback. The results of the review indicate that technology-based diagnosis and intervention for NDD population is promising. However, given a high risk of bias of many studies, more high-quality research is needed.

An explainable deep learning framework for characterizing and interpreting human brain states

Deep learning approaches have been widely adopted in the medical image analysis field. However, a most of existing deep learning approaches focus on achieving promising performances such as classification, detection, and segmentation, and much less effort is devoted to the explanation of the designed models. Similarly, in the brain imaging field, many deep learning approaches have been designed and applied to characterize and predict human brain states. However, these models lack interpretation. In response, we propose a novel domain knowledge informed self-attention graph pooling-based (SAGPool) graph convolutional neural network to study human brain states. Specifically, the dense individualized and common connectivity-based cortical landmarks system (DICCCOL, structural brain connectivity profiles) and holistic atlases of functional networks and interactions system (HAFNI, functional brain connectivity profiles) are integrated with the SAGPool model to better characterize and interpret the brain states. Extensive experiments are designed and carried out on the large-scale human connectome project (HCP) Q1 and S1200 dataset. Promising brain state classification performances are observed (e.g., an average of 93.7% for seven-task classification and 100% for binary classification). In addition, the importance of the brain regions, which contributes most to the accurate classification, is successfully quantified and visualized. A thorough neuroscientific interpretation suggests that these extracted brain regions and their importance calculated from self-attention graph pooling layer offer substantial explainability.

omicsGAT: Graph Attention Network for Cancer Subtype Analyses

The use of high-throughput omics technologies is becoming increasingly popular in all facets of biomedical science. The mRNA sequencing (RNA-seq) method reports quantitative measures of more than tens of thousands of biological features. It provides a more comprehensive molecular perspective of studied cancer mechanisms compared to traditional approaches. Graph-based learning models have been proposed to learn important hidden representations from gene expression data and network structure to improve cancer outcome prediction, patient stratification, and cell clustering. However, these graph-based methods cannot rank the importance of the different neighbors for a particular sample in the downstream cancer subtype analyses. In this study, we introduce omicsGAT, a graph attention network (GAT) model to integrate graph-based learning with an attention mechanism for RNA-seq data analysis. The multi-head attention mechanism in omicsGAT can more effectively secure information of a particular sample by assigning different attention coefficients to its neighbors. Comprehensive experiments on The Cancer Genome Atlas (TCGA) breast cancer and bladder cancer bulk RNA-seq data and two single-cell RNA-seq datasets validate that (1) the proposed model can effectively integrate neighborhood information of a sample and learn an embedding vector to improve disease phenotype prediction, cancer patient stratification, and cell clustering of the sample and (2) the attention matrix generated from the multi-head attention coefficients provides more useful information compared to the sample correlation-based adjacency matrix. From the results, we can conclude that some neighbors play a more important role than others in cancer subtype analyses of a particular sample based on the attention coefficient.

CNNG: A Convolutional Neural Networks With Gated Recurrent Units for Autism Spectrum Disorder Classification

As a neurodevelopmental disorder, autism spectrum disorder (ASD) severely affects the living conditions of patients and their families. Early diagnosis of ASD can enable the disease to be effectively intervened in the early stage of development. In this paper, we present an ASD classification network defined as CNNG by combining of convolutional neural network (CNN) and gate recurrent unit (GRU). First, CNNG extracts the 3D spatial features of functional magnetic resonance imaging (fMRI) data by using the convolutional layer of the 3D CNN. Second, CNNG extracts the temporal features by using the GRU and finally classifies them by using the Sigmoid function. The performance of CNNG was validated on the international public data—autism brain imaging data exchange (ABIDE) dataset. According to the experiments, CNNG can be highly effective in extracting the spatio-temporal features of fMRI and achieving a classification accuracy of 72.46%.

Identification of Young High-Functioning Autism Individuals Based on Functional Connectome Using Graph Isomorphism Network: A Pilot Study

Accumulated studies have determined the changes in functional connectivity in autism spectrum disorder (ASD) and spurred the application of machine learning for classifying ASD. Graph Neural Network provides a new method for network analysis in brain disorders to identify the underlying network features associated with functional deficits. Here, we proposed an improved model of Graph Isomorphism Network (GIN) that implements the Weisfeiler-Lehman (WL) graph isomorphism test to learn the graph features while taking into account the importance of each node in the classification to improve the interpretability of the algorithm. We applied the proposed method on multisite datasets of resting-state functional connectome from Autism Brain Imaging Data Exchange (ABIDE) after stringent quality control. The proposed method outperformed other commonly used classification methods on five different evaluation metrics. We also identified salient ROIs in visual and frontoparietal control networks, which could provide potential neuroimaging biomarkers for ASD identification.