Discoursing Novel Procedure for Segmentation and Classification of Mammograms

SN Computer Science (2021) 2:61 https://doi.org/10.1007/s42979-021-00454-6 ORIGINAL RESEARCH Discoursing Novel Procedure for Segmentation and Classification of Mammograms Baljinder Singh1 · Ramandeep Kaur2 · Amandeep Kaur3 · Gagandeep Jagdev4 Received: 24 November 2020 / Accepted: 6 January 2021 © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. part of Springer Nature 2021 Abstract Mammography plays a significant role in the early detection of breast cancers since it can demonstrate changes in the breast, years before a patient or physician can feel them. The research work conducted in the research paper highlights the process of segmentation and classification of mammogram images intending to detect the presence of tumors in the breast at early stages and classifying it as benign (cancerous) or malignant (non-cancerous) so that the course of treatment could be decided to prevent further damage. The flowchart developed in the research paper defines a systematic approach adopted to perform segmentation on mammograms. This includes the use of techniques like Green Channel Complement, CLAHE (Contrast Limited Adaptive Histogram Equalization), Morphological operations, and FCM (Fuzzy C-Means). Mammogram images from the MIAS (Mammographic Image Analysis Society) database have been used for performing segmentation. The research paper features a detailed algorithm that discusses the detailed adopted approach. The GUI (Graphical User Interface) has been constructed with multiple windows to show the output received at each step after appropriate processing. The results have been obtained in the form of numerical readings using performance evaluation parameters like sensitivity, specificity, accuracy, positive predictive value, negative predictive value, false-negative rate, false-positive rate, etc. The obtained readings of different parameters prove the authenticity of the conducted work. Segmentation enables the scrutinizing of any region within an image. The conducted research work can prove helpful in enhancing the mammogram image and focusing on the segmented image which indicates the presence of microcalcifications. The effectively conducted segmentation enables the radiologist to classify the tumor and monitor the seriousness of caused damage. Based on the obtained results the further treatment of the patient can be decided upon. Keywords Breast cancer · CLAHE · Classification · Green channel · Mammography · Morphological operations · Segmentation Introduction This article is part of the topical collection “Applications of Cloud Computing, Data Analytics and Building Secure Networks” guest edited by Rajnish Sharma, Pao-Ann Hsiung and Sagar Juneja. * Gagandeep Jagdev 1 Department of Computer Science, Government Rajindra College, Bathinda, Punjab, India 2 Department of Computer Science, Bhai Behlo Khalsa Girls College, Phaphre Bhai Ke, Mansa, Punjab, India 3 Department of Computer Science, Guru Gobind Singh Khalsa College, Bhagta Bhai Ka, Bathinda, Punjab, India 4 Department of Computer Science, Punjabi University Guru Kashi College, Damdama Sahib, Talwandi Sabo, Punjab, India The major health problem faced by women throughout the world is breast cancer. Mammography was designed to find out the early detection of cancer. Mammography can also be used to identify and diagnose breast disease in women undergoing symptoms such as pain, skin dimpling, lump, or nipple discharge. Mammography is capable of detecting all kinds of cancer-related to the breast like invasive ductal cancer, invasive lobular cancer, etc. Mammography can detect even small abnormal tissue growths limited to milk ducts in women’s breasts. Mammography makes use of low energy X-rays to create images and inspect the human breast [1]. The X-rays are absorbed by different parts of the body in varying degrees. Soft tissues allow X-rays to pass through them whereas dense bones captivate much of the SN Computer Science Vol.:(0123456789) 61 Page 2 of 10 radiation [2]. Therefore, soft tissues appear in gray, bones in white, and air as black on X-ray. It is capable of detecting even minor calcium deposits. The breast is positioned by a trained radiologist on a special platform and a clear plastic paddle is used to compress it [3, 4]. A digital mammogram converts an X-ray into an electronic picture that is saved on a computer. The images obtained on the computer are even clearer than those visible on a regular mammogram. Greater is the degree of compression, the better will be the quality of the image [5]. The patient may be asked to hold her breath for a few seconds while the X-ray is been taken to reduce the likelihood of blurred image. After X-ray, no radiation is left behind in women’s bodies. Often there is a need to enhance images to reduce the level of noise of the image [6, 7]. With the advancement in technology, tumors are visible in mammograms. Tumors are classified into two categories; benign and malignant [8]. Benign tumors are not cancerous and are normal in appearance. In this case, the cells multiply very slowly but do not spread to other parts of the body [9]. Malignant tumors are cancerous and bears tend to spread to other parts of the body. Mammogram segmentation involves categorizing mammograms into separate regions like a nipple, breast border, and pectoral muscle [10, 11]. Several problems are associated with the accurate segmentation of the breast region. Because of the working nature of X-ray, pixels in mammograms embodies multiple tissues. It becomes difficult to differentiate between regions because of the superimposition of different tissue types [12]. Adopted Research Approach This section elaborates on the procedure adopted for performing segmentation and classification on the mammogram images. The images considered are obtained from the MIAS database. A mammogram image is given as input to the designed system. The input image is converted into GCC (Green Channel Complement). The output obtained from GCC is given as input to CLAHE. The output from CLAHE undergoes morphological operations. The 2D median filter is applied to the enhanced image and the optic disk is removed. Finally, segmentation is performed using Fuzzy C-Means and the numerical readings for relevant parameters are obtained. These parameters are briefly explained below. SN Computer Science • Positive Predictive Value (PPV)—PPV is the probability • • • • • • • that subjects with a positive screening test truly have the disease. Negative predictive value (NPV)—NPV is the probability that subjects with a negative screening test truly don’t have the disease [14]. False-negative rate (FNR)—FNR is the proportion of positives that yield negative test outcomes with the test [15]. False-positive rate (FPR)—FPR is the proportion of negatives which yield positive test outcomes with the test [16, 17]. False discovery rate (FDR)—FDR is the ratio of the number of false-positive results in t (...truncated)