This paper presents a fiber tracking method that solves fiber crossing and avoids anatomically improbable fiber tracts using clinical-quality diffusion tensor images. The method tracks fiber pathways under the guidance of global information from volumetric tract segmentation obtained from the algorithm of diffusion oriented tract segmentation (DOTS). New primary diffusion directions (PDD) are calculated in crossing areas labeled by DOTS, by solving the Euler-Lagrange equation. We replace the PDD field obtained from the tensors with the new PDD field, which we refer to as DOTS-PDD, in crossing areas in tractography. Results on synthetic data and real data demonstrate that the proposed method manages to propagate fiber pathways through crossing area and produce anatomically desirable fibers.