1. Agostinho D, Correia R, Catarina Duarte I, Sousa D, Abreu R, Pina Rodrigues A, Castelo-Branco M, Simoes M. Parametric fMRI analysis of videos of variable arousal levels reveals different dorsal vs ventral activation preferences between autism and controls. Annual International Conference of the IEEE Engineering in Medicine and Biology Society IEEE Engineering in Medicine and Biology Society Annual International Conference. 2021; 2021: 6412-5.

Atypical sensory processing is now considered a ubiquitous feature of autism spectrum disorder (ASD) and is responsible for the atypical sensory-based behaviours seen in these individuals. Specifically, emotional arousal is a critical ASD target since it comprises emotion regulation and sensory processing, two core aspects of autism. So, in this project, we used task-based fMRI and a well-catalogued dataset of videos with variable arousal levels to characterize the sensory processing of emotional arousal content in ASD and typically developed controls. Our analysis revealed a difference in the secondary attention network where ASD individuals showed a clear yet lateralized preference to the dorsal attention network, whereas the neurotypical individuals preferred the ventral attention network.

Lien vers le texte intégral (Open Access ou abonnement)

2. Arabian H, Wagner-Hartl V, Geoffrey Chase J, Moller K. Facial Emotion Recognition Focused on Descriptive Region Segmentation. Annual International Conference of the IEEE Engineering in Medicine and Biology Society IEEE Engineering in Medicine and Biology Society Annual International Conference. 2021; 2021: 3415-8.

Facial emotion recognition (FER) is useful in many different applications and could offer significant benefit as part of feedback systems to train children with Autism Spectrum Disorder (ASD) who struggle to recognize facial expressions and emotions. This project explores the potential of real time FER based on the use of local regions of interest combined with a machine learning approach. Histogram of Oriented Gradients (HOG) was implemented for feature extraction, along with 3 different classifiers, 2 based on k-Nearest Neighbor and 1 using Support Vector Machine (SVM) classification. Model performance was compared using accuracy of randomly selected validation sets after training on random training sets of the Oulu-CASIA database. Image classes were distributed evenly, and accuracies of up to 98.44% were observed with small variation depending on data distributions. The region selection methodology provided a compromise between accuracy and number of extracted features, and validated the hypothesis a focus on smaller informative regions performs just as well as the entire image.

Lien vers le texte intégral (Open Access ou abonnement)

3. Correia R, Agostinho D, Duarte IC, Sousa D, Rodrigues AP, Castelo-Branco M, Simoes M. Assessing Arousal Through Multimodal Biosignals: A Preliminary Approach. Annual International Conference of the IEEE Engineering in Medicine and Biology Society IEEE Engineering in Medicine and Biology Society Annual International Conference. 2021; 2021: 1508-11.

The increase in Autism Spectrum Disorder (ASD) prevalence estimates over the last decades has driven a quest to develop new forms of rehabilitation that can be accessible to a larger part of this population. These rehabilitation approaches often take the form of computer games that are blind to the user’s emotional state, which compromises their efficacy. In this study, a set of physiological signals were acquired in simultaneous with functional Magnetic Resonance Imaging (fMRI) with the future prospect of combining both kinds of data to create models capable of assessing the true emotional state of their users based on physiological response as a measure of autonomic nervous system, having as ground truth the activity of targeted brain regions. This paper describes an initial approach, focusing on the information contained on the physiological signals alone. A total of 35 features were extracted from biosignals’ segments and subsequently used for automatic classification of arousal state (High Arousal vs. Low Arousal). The suboptimal results, although some extracted features present statistically significant differences, underline the challenging nature of our proposal and the added obstacles of recording physiological signals in the magnetic resonance environment. Further exploration of the measured signals is needed to gather a bigger number of discriminative features that can improve classification outcomes.

Lien vers le texte intégral (Open Access ou abonnement)

4. Du Y, Hao H, Xing Y, Niu J, Calhoun VD. A Transdiagnostic Biotype Detection Method for Schizophrenia and Autism Spectrum Disorder Based on Graph Kernel. Annual International Conference of the IEEE Engineering in Medicine and Biology Society IEEE Engineering in Medicine and Biology Society Annual International Conference. 2021; 2021: 3241-4.

Psychiatric diagnoses based on clinical manifestations are prone to be inaccurate. Schizophrenia (SZ) and autism spectrum disorder (ASD) were historically considered as the same disorder, and they still have many overlaps of clinical symptoms in the current standard. Therefore, there is an urgent need to explore the potential biotypes for them using neuroimaging measures such as brain functional connectivity (FC). However, previous studies have not effectively leveraged FC in detecting biotypes. Considering that graph theory helps reveal the topological information in FC, in this paper, we propose a graph kernel-based clustering method to explore transdiagnostic biotypes using FC estimated from functional magnetic resonance imaging (fMRI) data. In our method, frequent subnetworks are identified from the whole-brain FCs of all subjects, and then the graph kernel similarity is computed to measure the relationship between subjects for clustering. Based on fMRI data of 137 SZ and 150 ASD subjects, we obtained meaningful biotypes using our method, which shows significant differences between the identified biotypes in FC. In brief, our graph kernel-based clustering method is promising for transdiagnostic biotype detection.

Lien vers le texte intégral (Open Access ou abonnement)

5. Ghosh S, Guha T. Towards Autism Screening through Emotion-guided Eye Gaze Response. Annual International Conference of the IEEE Engineering in Medicine and Biology Society IEEE Engineering in Medicine and Biology Society Annual International Conference. 2021; 2021: 820-3.

Individuals with Autism Spectrum Disorder (ASD) are known to have significantly limited social interaction abilities, which are often manifested in different non-verbal cues of communication such as facial expression, atypical eye gaze response. While prior works leveraged the role of pupil response for screening ASD, limited works have been carried out to find the influence of emotion stimuli on pupil response for ASD screening. We, in this paper, design, develop, and evaluate a light-weight LSTM (Long-short Term Memory) model that captures pupil responses (pupil diameter, fixation duration, and fixation location) based on the social interaction with a virtual agent and detects ASD sessions based on short interactions. Our findings demonstrate that all the pupil responses vary significantly in the ASD sessions in response to the different emotion (angry, happy, neutral) stimuli applied. These findings reinforce the ASD screening with an average accuracy of 77%, while the accuracy improves further (>80%) with respect to angry and happy emotion stimuli.

Lien vers le texte intégral (Open Access ou abonnement)

6. Matta J, Dobrino D, Howard S, Yeboah D, Kopel J, El-Manzalawy Y, Obafemi-Ajayi T. A PheWAS Model of Autism Spectrum Disorder. Annual International Conference of the IEEE Engineering in Medicine and Biology Society IEEE Engineering in Medicine and Biology Society Annual International Conference. 2021; 2021: 2110-4.

Children with Autism Spectrum Disorder (ASD) exhibit a wide diversity in type, number, and severity of social deficits as well as communicative and cognitive difficulties. It is a challenge to categorize the phenotypes of a particular ASD patient with their unique genetic variants. There is a need for a better understanding of the connections between genotype information and the phenotypes to sort out the heterogeneity of ASD. In this study, single nucleotide polymorphism (SNP) and phenotype data obtained from a simplex ASD sample are combined using a PheWAS-inspired approach to construct a phenotype-phenotype network. The network is clustered, yielding groups of etiologically related phenotypes. These clusters are analyzed to identify relevant genes associated with each set of phenotypes. The results identified multiple discriminant SNPs associated with varied phenotype clusters such as ASD aberrant behavior (self-injury, compulsiveness and hyperactivity), as well as IQ and language skills. Overall, these SNPs were linked to 22 significant genes. An extensive literature search revealed that eight of these are known to have strong evidence of association with ASD. The others have been linked to related disorders such as mental conditions, cognition, and social functioning.Clinical relevance- This study further informs on connections between certain groups of ASD phenotypes and their unique genetic variants. Such insight regarding the heterogeneity of ASD would support clinicians to advance more tailored interventions and improve outcomes for ASD patients.

Lien vers le texte intégral (Open Access ou abonnement)

7. Mayor-Torres JM, Ravanelli M, Medina-DeVilliers SE, Lerner MD, Riccardi G. Interpretable SincNet-based Deep Learning for Emotion Recognition from EEG brain activity. Annual International Conference of the IEEE Engineering in Medicine and Biology Society IEEE Engineering in Medicine and Biology Society Annual International Conference. 2021; 2021: 412-5.

Machine learning methods, such as deep learning, show promising results in the medical domain. However, the lack of interpretability of these algorithms may hinder their applicability to medical decision support systems. This paper studies an interpretable deep learning technique, called SincNet. SincNet is a convolutional neural network that efficiently learns customized band-pass filters through trainable sinc-functions. In this study, we use SincNet to analyze the neural activity of individuals with Autism Spectrum Disorder (ASD), who experience characteristic differences in neural oscillatory activity. In particular, we propose a novel SincNet-based neural network for detecting emotions in ASD patients using EEG signals. The learned filters can be easily inspected to detect which part of the EEG spectrum is used for predicting emotions. We found that our system automatically learns the high-α (9-13 Hz) and β (13-30 Hz) band suppression often present in individuals with ASD. This result is consistent with recent neuroscience studies on emotion recognition, which found an association between these band suppressions and the behavioral deficits observed in individuals with ASD. The improved interpretability of SincNet is achieved without sacrificing performance in emotion recognition.

Lien vers le texte intégral (Open Access ou abonnement)

8. Qian K, Koike T, Tamada K, Takumi T, Schuller BW, Yamamoto Y. Sensing the Sounds of Silence: A Pilot Study on the Detection of Model Mice of Autism Spectrum Disorder from Ultrasonic Vocalisations. Annual International Conference of the IEEE Engineering in Medicine and Biology Society IEEE Engineering in Medicine and Biology Society Annual International Conference. 2021; 2021: 68-71.

Studying the animal models of human neuropsychiatric disorders can facilitate the understanding of mechanisms of symptoms both physiologically and genetically. Previous studies have shown that ultrasonic vocalisations (USVs) of mice might be efficient markers to distinguish the wild type group and the model of autism spectrum disorder (mASD). Nevertheless, in-depth analysis of these ‘silence’ sounds by leveraging the power of advanced computer audition technologies (e. g., deep learning) is limited. To this end, we propose a pilot study on using a large-scale pre-trained audio neural network to extract high-level representations from the USVs of mice for the task on detection of mASD. Experiments have shown a best result reaching an unweighted average recall of 79.2 % for the binary classification task in a rigorous subject-independent scenario. To the best of our knowledge, this is the first time to analyse the sounds that cannot be heard by human beings for the detection of mASD mice. The novel findings can be significant to motivate future works with according means on studying animal models of human patients.

Lien vers le texte intégral (Open Access ou abonnement)

9. Ren T, Zhang J, Yu Y, Pedersen LH, Wang H, Li F, Henriksen TB, Li J. Association of labour epidural analgesia with neurodevelopmental disorders in offspring: a Danish population-based cohort study. British journal of anaesthesia. 2022; 128(3): 513-21.

BACKGROUND: Whether labour epidural analgesia impacts risk of neurodevelopmental disorders in offspring is unsettled, raising public and scientific concerns. We explored the association between maternal labour epidural analgesia and autism spectrum disorder, and specific developmental disorder, attention-deficit hyperactivity disorder, intellectual disability, and epilepsy in offspring. METHODS: This nationwide population-based cohort study included 624 952 live-born singletons delivered by women who intended to deliver vaginally (i.e. vaginal and intrapartum Caesarean deliveries) in Denmark from 2005 to 2016. A total of 80 862 siblings discordant for exposure to labour epidural analgesia were analysed in a sibling-matched analysis. Both full-cohort and sibling-matched analyses were performed to estimate hazard ratios (HRs) of offspring risk of autism spectrum disorder, specific developmental disorder, attention-deficit hyperactivity disorder, intellectual disability, and epilepsy, according to exposure to labour epidural analgesia, adjusted for maternal socio-economic, pregnancy, and perinatal covariates. RESULTS: In the full cohort, maternal labour epidural analgesia was associated with autism spectrum disorder in offspring (HR 1.11; 95% confidence interval [CI]: 1.04-1.18); however, in the sibling-matched analysis, no association with autism spectrum disorder was found (HR 1.03; 95% CI: 0.84-1.27). The association between labour epidural analgesia and specific developmental disorder (HR 1.12; 95% CI: 1.03-1.22) in the full cohort also disappeared in the sibling-matched analysis (HR 1.01; 95% CI: 0.78-1.31). No association between maternal labour epidural analgesia and the remaining neurodevelopmental disorders was found overall (attention-deficit hyperactivity disorder, HR 0.98; 95% CI: 0.92-1.03; intellectual disability, HR 0.98; 95% CI: 0.85-1.14; epilepsy, HR 0.89; 95% CI: 0.79-1.00) or in the sibling-matched analyses. CONCLUSIONS: Our findings did not support an association between maternal attention-deficit hyperactivity disorder and autism spectrum disorder, specific developmental disorder, attention-deficit hyperactivity disorder, intellectual disability, or epilepsy.

Lien vers le texte intégral (Open Access ou abonnement)

10. Richardson R, Baralle D, Bennett C, Briggs T, Bijlsma EK, Clayton-Smith J, Constantinou P, Foulds N, Jarvis J, Jewell R, Johnson DS, McEntagart M, Parker MJ, Radley JA, Robertson L, Ruivenkamp C, Rutten JW, Tellez J, Turnpenny PD, Wilson V, Wright M, Balasubramanian M. Further delineation of phenotypic spectrum of SCN2A-related disorder. American journal of medical genetics Part A. 2022; 188(3): 867-77.

SCN2A-related disorders include intellectual disability, autism spectrum disorder, seizures, episodic ataxia, and schizophrenia. In this study, the phenotype-genotype association in SCN2A-related disorders was further delineated by collecting detailed clinical and molecular characteristics. Using previously proposed genotype-phenotype hypotheses based on variant function and position, the potential of phenotype prediction from the variants found was examined. Patients were identified through the Deciphering Developmental Disorders study and gene matching strategies. Phenotypic information and variant interpretation evidence were collated. Seventeen previously unreported patients and five patients who had been previously reported (but with minimal phenotypic and segregation data) were included (10 males, 12 females; median age 10.5 years). All patients had developmental delays and the majority had intellectual disabilities. Seizures were reported in 15 of 22 (68.2%), four of 22 (18.2%) had autism spectrum disorder and no patients were reported with episodic ataxia. The majority of variants were de novo. One family had presumed gonadal mosaicism. The correlation of the use of sodium channel-blocking antiepileptic drugs with phenotype or genotype was variable. These data suggest that variant type and position alone can provide some predictive information about the phenotype in a proportion of cases, but more precise assessment of variant function is needed for meaningful phenotype prediction.

Lien vers le texte intégral (Open Access ou abonnement)

11. Tanaka H, Nakamura S. Virtual Agent Design for Social Skills Training Considering Autistic Traits. Annual International Conference of the IEEE Engineering in Medicine and Biology Society IEEE Engineering in Medicine and Biology Society Annual International Conference. 2021; 2021: 4953-6.

Social skills training by human coaches is a well-established method to obtain appropriate social interaction skills and strengthen social self-efficacy. Our previous works automated social skills training by developing a virtual agent that teaches social skills through interaction. This study attempts to investigate the effect of virtual agent design on automated social skills training. We prepared images and videos of a virtual agent, and a total of 912 crowdsourced workers rated the virtual agents by answering questions. We investigated the acceptability, likeability, and other impressions of the virtual agents and their relationship to the individuals’ characteristics to design personalized virtual agents. As a result, a female anime-type virtual agent was rated as the most likable. We also confirmed that participants’ gender, age, and autistic traits are related to the ratings. We believe our findings are important in designing a personalized virtual trainer.Clinical relevance- This study examines the effect of virtual agent design on social skills training. Our findings are important in designing a personalized virtual trainer.

Lien vers le texte intégral (Open Access ou abonnement)

12. Widatalla N, Khandoker A, Yoshida C, Nakanishi K, Fukase M, Suzuki A, Kasahara Y, Saito M, Kimura Y. Effect of Valproic Acid on Maternal – Fetal Heart Rates and Coupling in Mice on Embryonic day 15.5 (E15.5). Annual International Conference of the IEEE Engineering in Medicine and Biology Society IEEE Engineering in Medicine and Biology Society Annual International Conference. 2021; 2021: 5504-7.

Prenatal uptake of valproic acid (VPA) was associated with increased risk of fetal cardiac anomalies and autism spectrum disorder (ASD), but uptake of VPA is considered the only effective treatment for epilepsy and other neurological disorders. Up until now, little is known about the effect of VPA on maternal – fetal heart rate (HR) coupling patterns; therefore, this study aims at studying such patterns in mice on embryonic day 15.5 (E15.5). At E12.5, 8 mothers were injected with VPA (VPA group) and another 8 mothers were injected with saline (control group). At E15.5, electrocardiogram (ECG) records of 15 minutes were collected from the 16 mothers and 25 fetuses. A maximum of 5-minutes and a minimum of 1-minute were selected from the ECG data for analysis. Mean RR intervals and coupling ratios and their occurrence percentages were calculated per 1minute. 1-minute analysis was done for periods with no arrhythmia and clear R peaks. The total number of 1-minute segments that were analyzed was 56 for the saline group and 54 for the VPA group. The correlation analysis between the 1:3 and 2:6 coupling ratios and RR intervals revealed that the ratios were significantly correlated in the saline group, whereas no significant correlations were observed in the VPA group. The results further revealed that fetal RR intervals are strongly correlated with maternal RR intervals in the saline group, but the same correlation is different in the VPA group. The presented results imply that maintaining certain coupling patterns are important for proper fetal cardiac development and maternal uptake of VPA may affect maternal-fetal HRs interactions.

Lien vers le texte intégral (Open Access ou abonnement)

13. Zhang Y, Peng B, Xue Z, Bao J, Li BK, Liu Y, Liu Y, Sheng M, Pang C, Dai Y. Self-Paced Learning and Privileged Information based Cascaded Multi-column Classification algorithm for ASD diagnosis. Annual International Conference of the IEEE Engineering in Medicine and Biology Society IEEE Engineering in Medicine and Biology Society Annual International Conference. 2021; 2021: 3281-4.

Autism spectrum disorder (ASD) is one of the most serious mental disorder in children. Machine learning based computer aided diagnosis (CAD) on resting-state functional magnetic resonance imaging (rs-fMRI) for ASD has attracted widespread attention. In recent years, learning using privileged information (LUPI), a supervised transfer learning method, has been generally used on multi-modality cases, which can transfer knowledge from source domain to target domain in order to improve the prediction capability on the target domain. However, multi-modality data is difficult to collect in clinical cases. LUPI method without introducing additional imaging modality images is worth further study. Random vector function link network plus (RVFL+) is a LUPI diagnosis algorithm, which has been proven to be effective for classification tasks. In this work, we proposed a self-paced learning based cascaded multi-column RVFL+ algorithm (SPL-cmcRVFL+) for ASD diagnosis. Initial classification model is trained using RVFL on the single-modal data (e.g. rs-fMRI). The output of the initial layer is then sent as privileged information (PI) to train the next layer of classification model. During this process, samples are selected using self-paced learning (SPL), which can adaptively select simple to difficult samples according to the loss value. The procedure is repeated until all samples are included. Experimental results show that our proposed method can accurately identify ASD and normal control, and outperforms other methods by a relatively higher classification accuracy.

Lien vers le texte intégral (Open Access ou abonnement)