Ultrasonography is one of the safest methods used as an effective diagnostic tool. The non-radioactive attribute makes it stand out in evaluating cancer treatment when the application of CT and MRI may have deteriorating effects. In this paper, for ultrasonic liver image analysis, approaches based on multiresolution fractal feature vector is explored to detect residual tumor tissue after the treatment of radio frequency ablation (RFA). This technique is applied on three sets of ultrasonic liver images, normal, cancer, and post-treatment coagulation necrosis in different periods, all histologically proven. In all images, 32*32 pixel rectangular regions of interest were selected by specialized physicians and used in the analysis. Our experiment demonstrates the feasibility of differentiation residual tumor from post-treatment necrosis and normal tissue by using multiresolution fractal (MF) feature vector, which may become a promising auxiliary tool for clinic evaluation of the RAF treatment effects and guidance providing for prospective therapy.