A major source of potential pitfalls in CT-based attenuation correction (CTAC) of PET data is the use of integral mode of CT detectors in the presence of polychromatic x-rays resulting in limited information for determination of exact tissue content. The wide range of bone mineral contents and densities in the human body makes it difficult to map the CT number to Linear Attenuation Coefficient (LAC) at the PET energy when merely using one or two scaling factors. In this study we proposed an alternative approach in order to use energy sensitive CT imaging techniques as opposed to integrating CT imaging. The multi-energy strategy would promise significant improvements in tissue determination and leads to accurate energy mapping results in CTAC, which can be especially critical and useful in the presence of varying bone tissues. In order to accurately validate our method a novel bone model based on cortical and marrow mixtures is proposed. Furthermore, a two-step energy mapping algorithm is implemented. For validation, tomographic projections of phantom in five energy bin were acquired and reconstructed. The proposed energy mapping technique was used to estimate the LAC of different bone tissues at 511 keV. The results had 1.1% error at maximum compared to true values. To test the precision, the effect of 10% variation in effective energy was investigated. In different bone tissues, maximum errors induced by the pricewise linear and hybrid methods were 8.0% and 14.6%, respectively; whereas in the proposed multi-energy method, errors was 1.6%, at maximum.