Recently, the hypoxia imaging has been recognized as the advanced technique to detect cancers because of a strong relationship with the biological characterization of cancer. In previous studies, hypoxia imaging systems for endoscopic diagnosis have been developed. However, these imaging technologies using the visible light can observe only blood vessels in gastric mucous membrane. Therefore, they could not detect scirrhous gastric cancer which accounts for 10% of all gastric cancers and spreads rapidly into submucous membrane. To overcome this problem, we developed a measuring system of blood oxygen saturation in submucous membrane by using near-infrared (NIR) spectroscopy. NIR, which has high permeability for bio-tissues and high absorbency for hemoglobin, can image and observe blood vessels in submucous membrane. NIR system with LED lights and a CCD camera module was developed to image blood vessels. We measured blood oxygen saturation using the optical density ratio (ODR) of two wavelengths, based on Lambert-Beer law. To image blood vessel clearly and measure blood oxygen saturation accurately, we searched two optimum wavelengths by using a multilayer human gastric-like phantom which has same optical properties as human gastric one. By using Monte Carlo simulation of light propagation, we derived the relationship between the ODR and blood oxygen saturation and elucidated the influence of blood vessel depth on measuring blood oxygen saturation. The oxygen saturation measuring methodology was validated with experiments using our NIR system. Finally, it was confirmed that our system can detect oxygen saturation in various depth blood vessels accurately.