Accelerating the K-Nearest Neighbors Filtering Algorithm to Optimize the Real-Time Classification of Human Brain Tumor in Hyperspectral Images

The use of hyperspectral imaging (HSI) in the medical field is an emerging approach to help medical doctors in the diagnostics or surgical guidance tasks. However, the processing of HSI data involves high computational requirements due to the large amounts of information captured by the sensor. The main goal of this study is to optimize and parallelize the k-nearest neighbors (KNN) filtering algorithm exploiting the GPU technology to obtain real-time processing during surgical procedures of brain cancer. This parallel version performs a filtering of a classification map (obtained from a supervised classifier), evaluating the classes of the pixels simultaneously. The adopted optimizations and the computational capabilities of the GPU device allow to obtain a speedup up to 66.18x compared to the serial implementation.