Comparison of anatomical or functional parameters is common theme in the evaluation of unilateral pathologies of the human knee joint, although the degree of bilateral symmetry is unknown. Currently, the evaluation is performed manually. Therefore, diversity of the evaluation happens due to the subjective judgment. To reduce the diversity, we propose an automated analysis method for anatomical knee structure from multi-detector row computerized tomography (MDCT) image. The experiment employed six different knees. Then, symmetry data were made from the six raw data by using mirroring processing. The six raw data and six mirroring data were evaluated by the proposed and manual method. The results of the proposed method indicated that there was no difference between the raw and mirroring data, while the results of the manual method clarified that there is almost two mm difference between them. It is concluded that the proposed method could identify the anatomical structure with no difference between bilateral knees.