A qualitative and quantitative comparative study of different denoising and enhancement techniques for breast mammograms, ultrasound and magnetic resonance images
Abhinav Kumar, Subodh P. Srivastava · IET conference proceedings. · 2023
Breast cancer ranks as the second highest-diagnosed malignancy in women. Mammography, ultrasound, and magnetic resonance imaging (MRI) are the most frequently used imaging modalities to identify breast cancer. These imaging modalities produce a digital image that helps to highlight breast cancer-associated abnormalities like calcification, masses and lumps. However, these digital images are ordinarily low-contrast images and are affected by noise. Furthermore, the presence of noise limits radiologists' ability to accurately interpret breast cancer. So, with pre-processing, the noise present in mammograms, Ultrasounds and MRIs is removed to predict breast cancer accurately. Over the years, various pre-processing techniques have been developed to overcome the noise and enhance the image. The pre-processing helps in denoising as well as the classification of breast cancer by providing a processed image. This paper offers a qualitative and quantitative comparative study of the most recent and efficient denoising and enhancement techniques for Mammography, Ultrasound and MRI imaging modalities. The quantitative assessments are provided in terms of full reference, no reference and human visual system-based image assessment metrics. The datasets from the mammographic image analysis society, ultrasound web and RIDER have been used for Mammography, Ultrasound and MRI, respectively.