Bone marrow is both the main hematopoietic and important immune organ. Bone marrow lesions (BMLs) may cause a series of severe complications and even myeloma. The traditional diagnosis of BMLs rely on mostly bone marrow biopsy/ puncture, and sometimes MRI, X-ray, and etc., which are either invasive and dangerous, or ionizing and costly. A diagnosis technology with advantages in noninvasive, safe, real-time continuous detection, and low cost is requested. Here we reported our preliminary exploration of feasibility verification of using near-infrared spectroscopy (NIRS) in clinical diagnosis of BMLs by Monte Carlo simulation study. We simulated and visualized the light propagation in the bone marrow quantitatively with a Monte Carlo simulation software for 3D voxelized media and Visible Chinese Human data set, which faithfully represents human anatomy. The results indicate that bone marrow actually has significant effects on light propagation. According to a sequence of simulation and data analysis, the optimal source-detector separation was suggested to be narrowed down to 2.8–3.2cm, at which separation the spatial sensitivity distribution of NIRS cover the most region of bone marrow with high signal-to-noise ratio. The display of the sources and detectors were optimized as well. This study investigated the light transport in spine addressing to the BMLs detection issue and reported the feasibility of NIRS detection of BMLs noninvasively in theory. The optimized probe design of the coming NIRS-based BMLs detector is also provided.