Current Proceedings on Technology

Yazarlar: İsmail Atacak, Bünyamin Gökpinar

Konular:-

Anahtar Kelimeler:Parkinson’s disease,Classification,Fisher score feature selection algorithm,Multilayer neural network

Özet: Parkinson’s disease (PD) is a degenerative disorder of the central nervous system, which affects the quality of life of the patients and has major social and economical impact. Using clinical examinations and observations on diagnosis of this disease may be insufficient, espicially at its early stages. Thus, nowadays, the research on computer-based solutions to support clinical decision has gained considerable importance. This study proposes a new computer-aided diagnosis system for neurologists to help them detect Parkinson’s disease with high accuracy rates. The proposed system consists of two stages: feature selection and classification stage. In feature selection stage, fisher score feature selection algorithm is utilized to select a set of attributes that better represent the PD data set. The resulting attributes are used as input parameters for multilayer neural network algorithm (MLPNN). The evolution of the system is conducted using k-fold cross-validation, classification accuracy, sensitivity, specificity values and confusion matrix. The high classification accuracy obtained by the fisher score+neural network based system using a 10-fold cross validation method ensures a reliable diagnostic model for detection of PD. Promisingly, the proposed system might serve as a new candidate of powerful tools for diagnosing PD with good performance.

---