Development of High-Performance Pathological Diagnosis Software Using a Hyperspectral Camera

摘要

Pathological diagnoses are done by skillful pathologists who trained themselves with more than two- years training. Even though their classification is accurate and useful most of the time, there exists a few controversial samples. In addition, the classification is not always true for the malignancy of tissues. In order to solve these problems, we are developing a high-performance pathological diagnosis software ANSWER by using a hyperspectral camera. Hyperspectral cameras are originally employed for remote sensing and agriculture. The spectral resolution is so high that a hyperspectral camera can tell the photosynthesis degree of plants, freshness of meats and so on. Our research revealed that this camera can distinguish the four stages of canceration of colon tissues with 95.1% accuracy rate. This paper shows the analysis method of hyperspectral data and the result. Additionally, we introduce the product design of ANSWER.