Enhancing the diagnosis of Autism Spectrum Disorder using Phenotypic, Structural, and functional MRI Data

Autism Spectrum Disorder is a neurodevelopmental condition characterized by difficulties with social interaction, verbal and nonverbal communication, interests, hobbies, and stereotyped, constrained behavior. In order to automate the identification of brain disorders marked by social deficiencies and repeated behaviors, machine learning and deep learning approaches have become very significant. In the paper, we have proposed and implemented a machine learning models and convolution neural network (CNN) for classifying subjects with ASD. Data is from Autism Brain Imagining Data Exchange (ABIDE) repository by using phenotypic, s-MRI, and fMRI data. For s-MRI image dataset, the accuracy of the neural network is about 87% whereas for fMRI image dataset the accuracy is 88% which is suitable for real-time usage. We implemented a GUI called Gradio for visualizing the s-MRI and fMRI data analysis. The work also interpreted the different Machine Learning (ML) models for the clinical data of ASD Screening of children (toddlers) which was available in the UCI repository, the different ML techniques used are Decision Tree, Random Forest, and Logistic Regression. The proposed methodology can detect and diagnose ASD early. An automated system helps in faster diagnosis and even minute things are identified and observed. Sometimes, humans can fail in identifying such minute things in the sample while diagnosing. To build such a system, deep learning models such as CNN models are trained on the s-MRI and fMRI images to classify them into ASD and non-ASD. The classification capability of the system developed was measured using the performance metrics such as accuracy, ROC (Receiver Operating Characteristic) curve, and AUC (Area under the Curve). The automated system can detect whether the given image is ASD or normal. The doctors can use this automated system very easily and do the needful after that. The novelty of our work is that we have considered the 3 modalities, for predicting the diseases. As a future work, we can do a fusion to give more accurate results combining 3 modalities results.