Differential near infrared optical density has the advantages of rapid and noninvasive detection when detecting the traumatic subdural hematoma . This technology is a hot research topic in recent years, it has an important application in emergency treatment. To further improve the accuracy of traumatic subdural hematoma degree predication, the multi-channel differential absorbance method is used in this paper. We use 5 detectors that have different distance to the source to obtain absorbance data. In addition, we use partial least square method to establish the calibration model between the optical absorption coefficient of the brain and differential optical density. The model prediction results show that average relative error of absorption coefficient is 11.16%, and average relative error of hematoma depth is less than 1%. The model basically meets the demand of traumatic subdural hematoma degree predication. By introducing multi-channel differential optical density in the detection, it could eliminate the effects of individual differences and improve the accuracy of traumatic subdural hematoma degree predication. This method provides a new idea and an important reference for the brain detection with near infrared spectroscopy.