The recognition of substantia nigra in brain stem ultrasound images based on Principal Component Analysis

摘要

This paper assays the recognition of substantia nigra of brain stem ultrasound images based on Principal Component Analysis. As input we have a collection of sonographical slices which were preprocessed and optimized and we must detect a ROI substantia nigra. Furthermore demonstrates a principle of PCA and practical implementation with results and contains a comparison of results from different software. A main goal is a classification of these images and recognition results. This processing is important for detection of Parkinson's disease, reflected well recognition of ROI substantia nigra. We got an output as selected principal components and we assessed a threshold for classification. Core implementation were realized in C# optimized application and computed in another existing software. We used cropped images contains ROI and we optimized PCA algorithm to more effective computing.