Contrast enhancement and micro-calcification detection using statistical and wavelet features in digital mammograms
Bhupendra Singh, Amit Verma, R. C. Tripathi · 2017
Breast cancer detection and diagnosis has been a difficult problem to be solved in the medical field. Recently, the field has seen the use of information technology and it's found that the diagnosis process has gained a lot of improvements. While most of the raw digital mammographic images cannot be read by radiologist through naked eye, the images processed by the proposed technique draws fine micro-calcifications. After the pre-processing, later stages i.e., feature extraction and classification have achieved significant performance results. In the proposed technique, input image is first enhanced through laplacian filtering, then wavelet, statistical and a feature extracted from Histogram is used for classification in to cancerous or non-cancerous image. On classification with Artificial Neural Network, Daubechies-8 with 50% low frequency components along with statistical features produced 92.8% true detection rate with 12.5% positive false rate.