Title : Autism spectrum disorder detection using FMRI imaging with convolutional neural networks
Abstract:
Purpose: Autism spectrum disorder (ASD) is a neurological and developmental disorder that affects how people interact with others, communicate, learn, and behave. The number of children diagnosed with autism spectrum disorder (ASD) has risen consistently and dramatically since the 1990s. According to the Centers for Disease Control and Prevention (CDC), 1 in 36 children and 1 in 45 adults are affected by autism in the U.S. in 2023. Currently, diagnosing ASD is difficult because ASD is diagnosed mainly by observing a patient's behavior and developmental history, as there is no automated or pathological solution established for diagnosis. There is a need to develop automated objective ASD diagnosis tools for quickly and accurately diagnosing ASD. In this study, we investigated ASD detection using functional Magnetic Resonance Imaging (fMRI) with Convolutional Neural Networks (CNN).
Methods: We proposed an approach to ASD detection using fMRI with resting-state functional connectivity (RSFC) patterns learned with CNN. RSFC measures the spontaneous low-frequency fluctuations between regions in the brain in fMRI data. The resting-state fMRI data from 895 individuals from the autism brain imaging data exchange (ABIDE) database was used in the study. The imaging data were acquired from 15 acquisition sites across the U.S. and consists of 556 healthy controls and 339 subjects diagnosed with ASD. After data preprocessing and removing the data with quality issues, the dataset consists of 548 controls and 326 ASD subjects.
Results: The proposed approach was tested on the data set with 10-fold validation. The performance of the approach was evaluated with multiple evaluation metrics, such as specificity, sensitivity, precision, accuracy and F1 score. The proposed approach achieved a sensitivity of 0.92±0.05, specificity of 1.0, precision of 1.0, accuracy of 0.93±0.03, and F1 score of 0.96±0.03. The results indicate that the ASD models can detect ASD effectively in the resting-state fMRI data with a high precision.
Conclusion: The experimental results demonstrate the good performance of the ASD detection models learned from the resting-state fMRI data with CNN. The proposed approach would be useful for developing objective ASD diagnosis tools and ASD research.
Audience Take Away Notes:
- Audience could learn a new approach to ASD detection using resting-state fMRI with CNN
- Other researchers could expand the proposed approach to the diagnosis of other brain disease, such as Alzheimer's disease and Parkinson's disease using resting-state fMRI
- Other researchers could also utilize the proposed approach to investigate relationship between the interactions between brain regions and ASD
- Automated ASD detection tools could be developed with the proposed approach, which would be useful for improving the efficiency and accuracy of ASD diagnosis
