Accurate measurement of retinal blood oxygen saturation (SO₂) provides valuable early insights into the pathophysiology of ocular diseases such as diabetic retinopathy, glaucoma, and retinal vein occlusion. Visible-light optical coherence tomography (OCT) can directly measure SO₂ through spectral fitting, but its application is limited due to the irritability of visible light and its influence on the physiological state of the retina. Near-infrared band I (NIR-I) causes less ocular stimulation. Even though the hemoglobin absorption effect is weaker in this region, its scattering property is also related to SO₂. According to this principle, we propose a novel optical coherence tomography (OCTA) guided NIR-I technique for retinal blood oxygen saturation measurement. This method is used to calculate SO₂ through calibrating the optical density ratio (ODR) of oxygen-sensitive wavelength (855 nm) to isosbestic wavelength (805 nm). By utilizing the three-dimensional (3D) blood flow maps generated by the OCTA, this technique can automatically identify retinal vessels and surrounding tissue regions, thereby minimizing the errors caused by manual selection. Consequently, the classification accuracy of arteries and veins increases from 82.1% to 96.7%. The calibrated average retinal blood oxygen saturation is 94%±21% for arteries and 56%±13% for veins, which aligns with normal physiological range. The representative result of artery-vein classification is presented in the following figure. This method greatly improves the accuracy and efficiency of measurement, and provides a reliable tool for early diagnosis, disease assessment, and treatment monitoring of ophthalmic diseases, which has broad application prospects.