Age-Specific Diagnostic Classification of ASD Using Deep Learning Approaches

Autism Spectrum Disorder (ASD) is a highly heterogeneous condition, due to high variance in its etiology, comorbidity, pathogenesis, severity, genetics, and brain functional connectivity (FC). This makes it devoid of any robust universal biomarker. This study aims to analyze the role of age and multivariate patterns in brain FC and their accountability in diagnosing ASD by deep learning algorithms. We utilized functional magnetic resonance imaging data of three age groups (6 to 11, 11 to 18, and 6 to 18 years), available with public databases ABIDE-I and ABIDE-II, to discriminate between ASD and typically developing. The blood-oxygen-level dependent time series were extracted using the Gordon's, Harvard Oxford and Diedrichsen's atlases, over 236 regions of interest, as 236x236 sized FC matrices for each participant, with Pearson correlations. The feature sets, in the form of FC heat maps were computed with respect to each age group and were fed to a convolutional neural network, such as MobileNetV2 and DenseNet201 to build age-specific diagnostic models. The results revealed that DenseNet201 was able to adapt and extract better features from the heat maps, and hence returned better accuracy scores. The age-specific dataset, with participants of ages 6 to 11 years, performed best, followed by 11 to 18 years and 6 to 18 years, with accuracy scores of 72.19%, 71.88%, and 69.74% respectively, when tested using the DenseNet201. Our results suggest that age-specific diagnostic models are able to counter heterogeneity present in ASD, and that enables better discrimination.

A Framework to Diagnose Autism Spectrum Disorder Using Morphological Connectivity of sMRI and XGBoost

In this study, we automated the diagnostic procedure of autism spectrum disorder (ASD) with the help of anatomical alterations found in structural magnetic resonance imaging (sMRI) data of the ASD brain and machine learning tools. Initially, the sMRI data was preprocessed using the FreeSurfer toolbox. Further, the brain regions were segmented into 148 regions of interest using the Destrieux atlas. Features such as volume, thickness, surface area, and mean curvature were extracted for each brain region, and the morphological connectivity was computed using Pearson correlation. These morphological connections were fed to XGBoost for feature reduction and to build the diagnostic model. The results showed an average accuracy of 94.16% for the top 18 features. The frontal and limbic regions contributed more features to the classification model. Our proposed method is thus effective for the classification of ASD and can also be useful for the screening of other similar neurological disorders.

Trust-Aware Routing Mechanism through an Edge Node for IoT-Enabled Sensor Networks

Although IoT technology is advanced, wireless systems are prone to faults and attacks. The replaying information about routing in the case of multi-hop routing has led to the problem of identity deception among nodes. The devastating attacks against the routing protocols as well as harsh network conditions make the situation even worse. Although most of the research in the literature aim at making the IoT system more trustworthy and ensuring faultlessness, it is still a challenging task. Motivated by this, the present proposal introduces a trust-aware routing mechanism (TARM), which uses an edge node with mobility feature that can collect data from faultless nodes. The edge node works based on a trust evaluation method, which segregates the faulty and anomalous nodes from normal nodes. In TARM, a modified gray wolf optimization (GWO) is used for forming the clusters out of the deployed sensor nodes. Once the clusters are formed, each cluster's trust values are calculated, and the edge node starts collecting data only from trustworthy nodes via the respective cluster heads. The artificial bee colony optimization algorithm executes the optimal routing path from the trustworthy nodes to the mobile edge node. The simulations show that the proposed method exhibits around a 58% hike in trustworthiness, ensuring the high security offered by the proposed trust evaluation scheme when validated with other similar approaches. It also shows a detection rate of 96.7% in detecting untrustworthy nodes. Additionally, the accuracy of the proposed method reaches 91.96%, which is recorded to be the highest among the similar latest schemes. The performance of the proposed approach has proved that it has overcome many weaknesses of previous similar techniques with low cost and mitigated complexity.