This is a lightly processed version of data from the ABIDE2 study.

Autism Brain Imaging Data Exchange II (ABIDE II). An extension of the ABIDE dataset, with greater phenotypic characterization, particularly in regard to measures of core ASD and associated symptoms.

restricted

This dataset was curated by downloading the Harvard oxford time-series for UCLA and UM sites from the Preprocessed Connectomes Project. Since the original ABIDE data was released under the CC-BY-NC-SA license. The same applies to this dataset as well. Figshare did not offer this option. http://fcon_1000.projects.nitrc.org/indi/abide/index.html

The CSV file contains the cortical thickness (CT) and fractal dimension (FD) estimated from the brain MR T₁-weighted images contained in the following online repositories:

• Autism Brain Imaging Data Exchange (ABIDE)
• 1000 Functional Connectomes Project (FCP) – ICBM dataset
• Consortium for Reliability and Reproducibility (CoRR) - NKI 2 - Nathan Kline Institute (Milham)

Each CSV file contains the following columns:

• Subject: id of each subject
• SITE: imaging site label (ABIDEI or ABIDEII followed by the institution name that collected the images; ICBM; NKI2)
• Age: each subject's age, expressed in years
• Sex: 0=male; 1=female.
• cortex_CT (FD): CT (or FD) of the cerebral cortical gray matter (GM).
• lh_cortex_CT (FD), rh_cortex_CT (FD): CT (or FD) of the left (lh) and right (rh) cerebral cortical GM.
• lh_frontal_cortex_CT (FD), rh_frontal_cortex_CT (FD): CT (or FD) of the left (lh) and right (rh) cerebral cortical GM of the frontal lobe.
• lh_temporal_cortex_CT (FD), rh_temporal_cortex_CT (FD): CT (or FD) of the left (lh) and right (rh) cerebral cortical GM of the temporal lobe.
• lh_parietal_cortex_CT (FD), rh_parietal_cortex_CT (FD): CT (or FD) of the left (lh) and right (rh) cerebral cortical GM of the parietal lobe.
• lh_occipital_cortex_CT (FD), rh_occipital_cortex_CT (FD): CT (or FD) of the left (lh) and right (rh) cerebral cortical GM of the occipital lobe.

Resting-state functional magnetic resonance imaging (rs-fMRI) data are 4-dimensional volumes (3-space + 1-time) that have been posited to reflect the underlying mechanisms of information exchange between brain regions, thus making it an attractive modality to develop diagnostic biomarkers of brain dysfunction. The enormous success of deep learning in computer vision has sparked recent interest in applying deep learning in neuroimaging. But the dimensionality of rs-fMRI data is too high (~20 M), making it difficult to meaningfully process the data in its raw form for deep learning experiments. It is currently not clear how the data should be engineered to optimally extract the time information, and whether combining different representations of time could provide better results. In this paper, we explored various transformations that retain the full spatial resolution by summarizing the temporal dimension of the rs-fMRI data, therefore making it possible to train a full three-dimensional convolutional neural network (3D-CNN) even on a moderately sized [~2,000 from Autism Brain Imaging Data Exchange (ABIDE)-I and II] data set. These transformations summarize the activity in each voxel of the rs-fMRI or that of the voxel and its neighbors to a single number. For each brain volume, we calculated regional homogeneity, the amplitude of low-frequency fluctuations, the fractional amplitude of low-frequency fluctuations, degree centrality, eigenvector centrality, local functional connectivity density, entropy, voxel-mirrored homotopic connectivity, and auto-correlation lag. We trained the 3D-CNN on a publically available autism dataset to classify the rs-fMRI images as being from individuals with autism spectrum disorder (ASD) or from healthy controls (CON) at an individual level. We attained results competitive on this task for a combined ABIDE-I and II datasets of ~66%. When all summary measures were combined the result was still only as good as that of the best single measure which was regional homogeneity (ReHo). In addition, we also applied the support vector machine (SVM) algorithm on the same dataset and achieved comparable results, suggesting that 3D-CNNs could not learn additional information from these temporal transformations that were more useful to differentiate ASD from CON.

Partitioning of the ABIDE I, ABIDE II, and ADHD200 datasets into training, validation and testing sets.

In the work "Profiling brain morphology for autism spectrum disorder with two cross-culture large-scale consortia" by Xue-Ru Fan et al., we explored neurodevelopmental heterogeneity in Autism Spectrum Disorder (ASD) through normative modeling of cross-cultural cohorts. Leveraging large-scale datasets from Autism Brain Imaging Data Exchange (ABIDE) and China Autism Brain Imaging Consortium (CABIC), the model identifies two ASD subgroups with distinct brain morphological abnormalities: subgroup "L" is characterized by generally smaller brain region volumes and higher rates of abnormality, while subgroup "H" exhibits larger volumes with less pronounced deviations in specific areas. Key areas, such as the isthmus cingulate and transverse temporal gyrus, were identified as critical for subgroup differentiation and ASD trait correlations. In subgroup H, the regional volume of the isthmus cingulate cortex showed a direct correlation with individuals' autistic mannerisms, potentially corresponding to its slower post-peak volumetric declines during development. These findings offer insights into the biological mechanisms underlying ASD and support the advancement of subgroup-driven precision clinical practices. This dataset includes the key data they calculated and used to conducted all the statistic analysis within the paper.

Autism spectrum disorder (ASD) is a heterogeneous disease that is characterized by abnormalities in social communication and interaction as well as repetitive behaviors and restricted interests. Structural brain imaging has identified significant cortical folding alterations in ASD; however, relatively less known is whether the core symptoms are related to neuroanatomical differences. In this study, we aimed to explore core-symptom-anchored gyrification alterations and their developmental trajectories in ASD. We measured the cortical vertex-wise gyrification index (GI) in 321 patients with ASD (aged 7–39 years) and 350 typically developing (TD) subjects (aged 6–33 years) across 8 sites from the Autism Brain Imaging Data Exchange I (ABIDE I) repository and a longitudinal sample (14 ASD and 7 TD, aged 9–14 years in baseline and 12–18 years in follow-up) from ABIDE II. Compared with TD, the general ASD patients exhibited a mixed pattern of both hypo- and hyper- and different developmental trajectories of gyrification. By parsing the ASD patients into three subgroups based on the subscores of the Autism Diagnostic Interview—Revised (ADI-R) scale, we identified core-symptom-specific alterations in the reciprocal social interaction (RSI), communication abnormalities (CA), and restricted, repetitive, and stereotyped patterns of behavior (RRSB) subgroups. We also showed atypical gyrification patterns and developmental trajectories in the subgroups. Furthermore, we conducted a meta-analysis to locate the core-symptom-anchored brain regions (circuits). In summary, the current study shows that ASD is associated with abnormal cortical folding patterns. Core-symptom-based classification can find more subtle changes in gyrification. These results suggest that cortical folding pattern encodes changes in symptom dimensions, which promotes the understanding of neuroanatomical basis, and clinical utility in ASD.

SCORING SEGMENTATIONS

Qualitative segmentation scores by two raters (rater1; rater2).

Scale: excellent (4); good (3); doubtful (2) and failed (1).

DATABASES

IXI database, Imperial College of London (https://brain-development.org/ixi-dataset/)

Autism Brain Imaging Data Exchange (ABIDE) database (http://fcon_1000.projects.nitrc.org)

SchizConnect database (http://schizconnect.org)

SEGMENTATION METHODS

MR-TIM (Taberna et al., 2021), green rows

WTS (Liu et al., 2017), red rows

TABLES

IXI_young

20 MRI from the IXI database, participants 20–35 years old;

MR scanners: Philips Intera 3.0T (HH); Philips Gyroscan Intera 1.5T (G)

IXI_older

20 MRI from the IXI database, participants 60–75 years old;

MR scanners: Philips Intera 3.0T (HH); Philips Gyroscan Intera 1.5T (G)

ABIDE

10 MRI from the ABIDE database, participants 18-25 years old;

MR scanner: Philips Achieva 3.0T

SchizConnect

10 MRI from the SchizConnect database, participants 19-66 years old;

MR scanner: Siemens Trio Tim 3.0T

REFERENCES

Liu, Q., Farahibozorg, S., Porcaro, C., Wenderoth, N., & Mantini, D. (2017). Detecting large-scale networks in the human brain using high-density electroencephalography. Hum Brain Mapp, 38(9), 4631-4643. doi:10.1002/hbm.23688

Taberna, G. A., Samogin, J., & Mantini, D. (2021). Automated Head Tissue Modelling Based on Structural Magnetic Resonance Images for Electroencephalographic Source Reconstruction. Neuroinformatics. doi:10.1007/s12021-020-09504-5

OpenBHB: a Multi-Site Brain MRI Dataset for Age Prediction and Debiasing The Open Big Healthy Brains (OpenBHB) dataset is a large (N>5000) multi-site 3D brain MRI dataset gathering 10 public datasets (IXI, ABIDE 1, ABIDE 2, CoRR, GSP, Localizer, MPI-Leipzig, NAR, NPC, RBP) of T1 images acquired across 93 different centers, spread worldwide (North America, Europe and China). Only healthy controls have been included in OpenBHB with age ranging from 6 to 88 years old, balanced between males and… See the full description on the dataset page: https://huggingface.co/datasets/benoit-dufumier/openBHB.

This is a set of 27x27 matrices constructed from the fMRI data for 2 participants from the ABIDE data set [Craddock et al., 2013] and 10 from the HCP Young Adult data set [Van Essen et al., 2013, Glasser et al., 2013]. A subset of 118 matrices (saved in a folder titled 'Random') was produced by sampling uniformly from the interval [0,1].

Each matrix represents the affinity in the immediate neighbourhood of a randomly-selected vertex on the midthickness surface of the brain, and was generated with the VB toolbox as outlined in [Bajada et al., 2020]. Note, however, that the referenced work takes a surface-based approach, but we used a version of the toolbox that is newer and employs a hybrid method whereby the randomly-chosen vertex is first mapped to the corresponding voxel, a 27-voxel neighbourhood (which includes the original voxel) is identified, and the affinity in said neighbourhood calculated. The element a_ij of the matrix constructed for the neighbourhood indicates the degree of correlation between voxels i and j on the basis of their fMRI data. The particular version of the VB toolbox utilised for this work is available at https://github.com/VBIndex/py_vb_toolbox/tree/Local-gradients-paper (but please note that it would have to be modified to print out the affinity matrices for a set of randomly-selected vertices).

The matrices are made available as numpy array files and are separated into folders according to the origin of the fMRI data used to construct them (the prefix ABIDE_ added to folder names indicates that the matrices contained inside were generated with the fMRI data of subjects from the ABIDE data set, while HCP is the Human Connectome Project counterpart. Each folder pertains to a different data subject, but note that the numbering scheme employed (DS1, DS2, etc.) has no correspondence to the original ABIDE or HCP participant numbers.

Results from ASD vs. control classification analysis performed for all five representations (Aregion through AFC_combo) within each of the two largest sites in the ABIDE dataset: Site #20 (N = 106) and Site #5 (N = 98).

(:unav)...........................................

Autism Spectrum Disorder (ASD) is a neurodevelopmental condition whose current psychiatric diagnostic process is subjective and behavior-based. In contrast, functional magnetic resonance imaging (fMRI) can objectively measure brain activity and is useful for identifying brain disorders. However, the ASD diagnostic models employed to date have not reached satisfactory levels of accuracy. This study proposes the use of MAACNN, a method that utilizes multi-view convolutional neural networks (CNNs) in conjunction with attention mechanisms for identifying ASD in multi-scale fMRI. The proposed algorithm effectively combines unsupervised and supervised learning. In the initial stage, we employ stacked denoising autoencoders, an unsupervised learning method for feature extraction, which provides different nodes to adapt to multi-scale data. In the subsequent stage, we perform supervised learning by employing multi-view CNNs for classification and obtain the final results. Finally, multi-scale data fusion is achieved by using the attention fusion mechanism. The ABIDE dataset is used to evaluate the model we proposed., and the experimental results show that MAACNN achieves superior performance with 75.12% accuracy and 0.79 AUC on ABIDE-I, and 72.88% accuracy and 0.76 AUC on ABIDE-II. The proposed method significantly contributes to the clinical diagnosis of ASD.

Autism Spectrum Disorder (ASD) is a neurodevelopmental condition whose current psychiatric diagnostic process is subjective and behavior-based. In contrast, functional magnetic resonance imaging (fMRI) can objectively measure brain activity and is useful for identifying brain disorders. However, the ASD diagnostic models employed to date have not reached satisfactory levels of accuracy. This study proposes the use of MAACNN, a method that utilizes multi-view convolutional neural networks (CNNs) in conjunction with attention mechanisms for identifying ASD in multi-scale fMRI. The proposed algorithm effectively combines unsupervised and supervised learning. In the initial stage, we employ stacked denoising autoencoders, an unsupervised learning method for feature extraction, which provides different nodes to adapt to multi-scale data. In the subsequent stage, we perform supervised learning by employing multi-view CNNs for classification and obtain the final results. Finally, multi-scale data fusion is achieved by using the attention fusion mechanism. The ABIDE dataset is used to evaluate the model we proposed., and the experimental results show that MAACNN achieves superior performance with 75.12% accuracy and 0.79 AUC on ABIDE-I, and 72.88% accuracy and 0.76 AUC on ABIDE-II. The proposed method significantly contributes to the clinical diagnosis of ASD.

Gender breakdown and distribution of age and FIQ score for each dataset (training, validation, testing, testing 2 sets).

Autism Spectrum Disorder (ASD) is a neurodevelopmental condition whose current psychiatric diagnostic process is subjective and behavior-based. In contrast, functional magnetic resonance imaging (fMRI) can objectively measure brain activity and is useful for identifying brain disorders. However, the ASD diagnostic models employed to date have not reached satisfactory levels of accuracy. This study proposes the use of MAACNN, a method that utilizes multi-view convolutional neural networks (CNNs) in conjunction with attention mechanisms for identifying ASD in multi-scale fMRI. The proposed algorithm effectively combines unsupervised and supervised learning. In the initial stage, we employ stacked denoising autoencoders, an unsupervised learning method for feature extraction, which provides different nodes to adapt to multi-scale data. In the subsequent stage, we perform supervised learning by employing multi-view CNNs for classification and obtain the final results. Finally, multi-scale data fusion is achieved by using the attention fusion mechanism. The ABIDE dataset is used to evaluate the model we proposed., and the experimental results show that MAACNN achieves superior performance with 75.12% accuracy and 0.79 AUC on ABIDE-I, and 72.88% accuracy and 0.76 AUC on ABIDE-II. The proposed method significantly contributes to the clinical diagnosis of ASD.

Datasets are Training, Validation, Testing (No Comorbidity), and Testing Set 2 (With Comorbidities).

Autism Spectrum Disorder (ASD) is a neurodevelopmental condition whose current psychiatric diagnostic process is subjective and behavior-based. In contrast, functional magnetic resonance imaging (fMRI) can objectively measure brain activity and is useful for identifying brain disorders. However, the ASD diagnostic models employed to date have not reached satisfactory levels of accuracy. This study proposes the use of MAACNN, a method that utilizes multi-view convolutional neural networks (CNNs) in conjunction with attention mechanisms for identifying ASD in multi-scale fMRI. The proposed algorithm effectively combines unsupervised and supervised learning. In the initial stage, we employ stacked denoising autoencoders, an unsupervised learning method for feature extraction, which provides different nodes to adapt to multi-scale data. In the subsequent stage, we perform supervised learning by employing multi-view CNNs for classification and obtain the final results. Finally, multi-scale data fusion is achieved by using the attention fusion mechanism. The ABIDE dataset is used to evaluate the model we proposed., and the experimental results show that MAACNN achieves superior performance with 75.12% accuracy and 0.79 AUC on ABIDE-I, and 72.88% accuracy and 0.76 AUC on ABIDE-II. The proposed method significantly contributes to the clinical diagnosis of ASD.