YOLOv5

This study addresses challenges in Parkinson’s disease (PD) diagnosis from MRI images by proposing an improved YOLOv5 deep learning algorithm that enhances sensitivity to subtle pathological features and optimizes feature extraction and task execution. The model demonstrated high performance on a dataset of 582 MRI images, achieving precision, recall, and mAP scores of 0.961, 0.974, and 0.986, respectively—outperforming other algorithms. These results suggest the model can support more accurate and consistent early PD diagnosis, helping to overcome the limitations of traditional, subjective clinical assessments.