The practical application of machine learning and deep learning continues to expand across various domains, with neuroscience being no exception. In the realm of neuroscience, these computational techniques are increasingly employed for tasks such as brain imaging analysis, neural signal processing, drug discovery, and neurodevelopmental disease classification. This dissertation aims to investigate the efficiency of deep learning models in classifying three neurodevelopmental diseases—schizophrenia, bipolar disorder, and ADHD—and distinguishing them from a control group of healthy individuals using fMRI data. The primary focus is on assessing the generalisation performance of state-of-the-art convolutional neural network (CNN) models. Specifically, we leverage 3D CNN architecture to preserve information within our fMRI data, mitigating the potential information loss associated with traditional 2D convolutional filters. Through this research, we seek to contribute insights into the viability of deep learning methodologies for accurate and robust classification in the context of neurodevelopmental diseases, particularly when applied to fMRI data. The utilisation of advanced CNN architectures reflects our commitment to enhancing model performance and addressing potential limitations in existing classification approaches.
The practical application of machine learning and deep learning continues to expand across various domains, with neuroscience being no exception. In the realm of neuroscience, these computational techniques are increasingly employed for tasks such as brain imaging analysis, neural signal processing, drug discovery, and neurodevelopmental disease classification. This dissertation aims to investigate the efficiency of deep learning models in classifying three neurodevelopmental diseases—schizophrenia, bipolar disorder, and ADHD—and distinguishing them from a control group of healthy individuals using fMRI data. The primary focus is on assessing the generalisation performance of state-of-the-art convolutional neural network (CNN) models. Specifically, we leverage 3D CNN architecture to preserve information within our fMRI data, mitigating the potential information loss associated with traditional 2D convolutional filters. Through this research, we seek to contribute insights into the viability of deep learning methodologies for accurate and robust classification in the context of neurodevelopmental diseases, particularly when applied to fMRI data. The utilisation of advanced CNN architectures reflects our commitment to enhancing model performance and addressing potential limitations in existing classification approaches.
Convolutional Neural Networks for Classification of Neurodevelopmental Disease Using fMRI Data
DARABI, BITA
2023/2024
Abstract
The practical application of machine learning and deep learning continues to expand across various domains, with neuroscience being no exception. In the realm of neuroscience, these computational techniques are increasingly employed for tasks such as brain imaging analysis, neural signal processing, drug discovery, and neurodevelopmental disease classification. This dissertation aims to investigate the efficiency of deep learning models in classifying three neurodevelopmental diseases—schizophrenia, bipolar disorder, and ADHD—and distinguishing them from a control group of healthy individuals using fMRI data. The primary focus is on assessing the generalisation performance of state-of-the-art convolutional neural network (CNN) models. Specifically, we leverage 3D CNN architecture to preserve information within our fMRI data, mitigating the potential information loss associated with traditional 2D convolutional filters. Through this research, we seek to contribute insights into the viability of deep learning methodologies for accurate and robust classification in the context of neurodevelopmental diseases, particularly when applied to fMRI data. The utilisation of advanced CNN architectures reflects our commitment to enhancing model performance and addressing potential limitations in existing classification approaches.File | Dimensione | Formato | |
---|---|---|---|
Darabi_Bita.pdf
accesso aperto
Dimensione
1.92 MB
Formato
Adobe PDF
|
1.92 MB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/64869