Hybrid Approach for the Classification and Detection of Brain Tumors
In the analysis of spectral information to identify histological samples affected by tumor tissue, the current proposal integrates HSI linear decomposition algorithms with classification methods such as neural networks.
Abstract
Hyperspectral analysis is a highly complex that supposes the development of algorithms with an enormous computational cost. The practical application of hyperspectral analysis requires the parallelization of said algorithms if it is to provide results in real time, using high-performance and massively parallel processing platforms that allow reducing execution times. In this learning, the potential of hybrid Hyperspectral image classification approaches is demonstrated. In the analysis of spectral information to identify histological samples affected by tumor tissue, the current proposal integrates HSI linear decomposition algorithms with classification methods such as neural networks. The outcomes demonstrate the usefulness of the proposed technique for the classification of histological tissue samples, producing significant accuracy in the analyzed images.