Prediction of NAC Response in Breast Cancer Patients Using Neural Network

Main Article Content

Susmitha Uddaraju
G. P. Saradhi Varma
M. R. Narasingarao

Abstract

Breast cancer is now the most prominent female cancer in both developing and developed nations, and that it is the largest risk factor for mortality worldwide. Notwithstanding the well-documented declines in breast cancer mortality during the last twenty years, occurrence rates continue to rise, and do so more rapidly in nations where rates were previously low. This has highlighted the significance of survival concerns and illness duration treatment. Patient data after first chemotherapy is collected from the hospital and this data is then analysed using neural network. Proposed architecture gives result as the patient is responding to the chemotherapy or not. Moreover, it also gives the risk factor in surgery. Early prediction of such things gives broader idea about how treatment should go. Once the Breast cancer is detected and if chemotherapy is done, then it becomes very important to check whether patient is responding to the chemotherapy or not. So, the proposed system architecture is designed in such a way that it detects if the patient is responding to the chemotherapy or not. And if patient is not responding to the chemotherapy, then patient should go to the surgery. The proposed system is also compared with the existing algorithms machine learning and neural network techniques like support vector machine (SVM) and Decision Tree(DT) algorithms. The proposed neural network architecture gives 99.19% accuracy where SVM and DT gives 89.15% and 74.82%. Bosom disease is known to have asymptomatic stages, which is distinguished simply by mammography and around 10% of patients getting mammography recovers further assessments, and among them 8 to 10% require bosom biopsy. Alert the cautious consideration of the radiologist to peruse mammograms to perceive mammograms is generally 30 to 60 seconds for every picture. In any case, the weakness and explicitness of human radiologist's mammography was controlled by 77-87% and 89-97%, individually. As of late, twofold peruses are allowed with most screening programs, yet this will additionally disintegrate the time heap of human radiologists. As of late, the headway of man-made brainpower (AI) has made it conceivable to recognize programmed infection on clinical pictures in radiology, pathology, and even gastrointestinalities. For bosom malignant growth screening, all the more profound examinations have additionally been led, 86.1 to 9.0% responsiveness and 79.0 to 90.0% exceptional elements. By and by, there are a couple of distributions for built up disease location of mammography under Asian with higher bosom thickness contrasted with white individuals. Bosom thickness can influence the malignant growth pace of mammography pictures. Hence, the motivation behind this study was to create and approve a profound learning model that consequently recognizes threatening bosom sores in Asian advanced mammograms and to inspect the exhibition of the model by bosom thickness level. We have acquainted our own pretreatment technique with expand the exhibition of the model. Furthermore, we tried to lead a meta-examination to contrast and accessible investigations on AI-based bosom malignant growth recognition. Apparently, this is probably the greatest review done on Asians.


 

Article Details

Section
Special Issue Papers