Abstract
Over the last few decades, neuroimaging, particularly magnetic resonance imaging (MRI), has played a significant sessional part in studying brain functions and diseases. MRI images, combined with unique ML approaches and developed tools during these years, have opened up new opportunities for diagnosing neurological illnesses. However, due to the apparent symptoms that are similar to each other, brain illnesses are regarded as difficult to precisely detect. This research examines a newly developed algorithm (ParkinsonNet) to classify Parkinson's disorder into two unique classes which are Control (healthy) and Parkinson's (PD), this method is one of the deep learning approaches, Convolutional neural networks (CNN). CNN is one way that may be used to classify a range of brain illnesses such as Parkinson's. We employed a freshly constructed CNN technique from scratch, and we got 97.9% accuracy which is considered outstanding compared with recently published articles using the same dataset