Predictive Diagnosis of Breast Cancer Using Deep Neural Network
Thankamony Devakhi Subha · Bioscience Biotechnology Research Communications · 2021
The ultimate aim of the study was to predict the overall survival rate for women affected by breast cancer by developing and validating a prognostic model.In this study mammography image were collector and the health records for at least 1 year before mammography.The algorithm was trained on mammogram and health records of women to make two breast cancer predictions: (i)to predict biopsy malignancy and (ii)to differentiate normal from abnormal examination.In this model,ANN's were used to complete the task.This model was developed using a large proportion of concealed layers for thedatageneralization.Alargeproportionclosetoa1000mammographyrecordsandresultswere used in the model to differentiate between benign and malignant tumors.Before being inputted, all the data was inspected by radiologists.The accuracy of the model was determined to be an AUC of 0.96 and was a huge success(area or curve used to determine the success of model).This model is highly consistent, effective and less prone to error when compared to diagnosis performed by pathologists.The study obtained very high accuracy of predictions such as 93% and 95%.This facilitates early and accurate diagnosis of breast cancer for patients and early diagnosis is integral in cancer diagnosis.This also contributes as a useful tool for new selective drugs for these tumors to be treated.In Today World, Technology has facilitated great improvements in various important aspects of the world such as healthcare, agriculture, business etc.Despite this tremendous growth and improvement various diseases are still on the rise world wide.Among these Cancer stands as one of the major causes of death and accounts for about 9.6million deaths world wide.Cancer has various forms.Among which Breast Cancer is one of the most dangerous and common reproductive cancers that mostly affects women.According to the World Health Organization, Breast cancer affects about 2.1 million women every year.In order to improve Breast cancer outcome and survival, Early detection is critical.Hence, the goal of our project is to use Machine learning techniques and deep learning to increase the proportion of breast cancers identified at an early stage, allowing for more effective treatment to be used and reduce the mortality rate.Since early detection of can ceriskey to effective treatment of breast cancer, we use various machine learning and deep learning techniques to predict if a tumour is benign or malignant, based on the features provided by the data in a more accurate,consistent and less prone toerror.