Computer-aided diagnosis for detection and classification of breast abnormalities on digital mammograms is an active area of. research. In the recent years, there have been many research developments in all aspects of the mammography. However, it still faces many challenges. The current detection accuracy of lesions such as mass and architectural distortion is considerably low. This research is focused on computerised texture analysis of mass and architectural distortion, and shape analysis of mass in digital mammograms. In texture analysis, we investigate fractal-based methods in texture characterisation of mammographic masses and architectural distortion.The individual ability of the different fractal-based features in the task of discriminating between abnormal lesion and normal breast parenchyma tissue are evaluated and compared using statistical analysis and receiver operating characteristics analysis.