Hybrid 3D CNN-LSTM Model for rs-fMRI-Based Parkinson’s Prediction

Parkinson’s disease (PD) is caused by dopamine neuron loss. Dopaminergic neurons in the “substantia nigra” which gradually die. Accurate and timely diagnosis is essential for treating Parkinson’s disease (PD). In this research, we present a unique method using Long Short-Term Memory and a 3D Convolutional Neural Network for Parkinson’s disease diagnosis. Including vocabulary learning increases the model’s ability to recognize neuroimaging characteristics linked to Parkinson’s disease. Three-dimensional structural and functional brain scans from both healthy controls and patients with Parkinson’s disease form our dataset. The 3D CNN and LSTM model outperforms other approaches with achieving an impressive 96% accuracy rate. Lexicon learning improves the model’s ability to identify complex neuroimaging signals associated with the pathophysiology of Parkinson’s disease. This novel approach advances the diagnosis of Parkinson’s disease and paves the way for more accurate early detection instruments. The program can identify and treat Parkinson’s disease patients more quickly, as evidenced by its remarkable 96 percent accuracy and low loss of 0.1. To determine the adaptability and treatment success of this strategy, more research and validation in a variety of populations are necessary.