ABSTRACT :

Breast cancer is a disease characterized by the rapid and uncontrolled growth of cells in the breast. It occurs when a malignant (cancerous) tumor originates in the breast cells. Breast cancer is the most commonly occurring cancer in women and the second most common cancer overall, with around 2 million cases observed in 2018. Early diagnosis of breast cancer can significantly improve prognosis and survival rates by enabling timely clinical treatment. Accurate classification of benign tumors can also prevent patients from undergoing unnecessary treatments. Thus, the correct diagnosis of breast cancer and the classification of patients into malignant or benign groups is the focus of much research. Due to its unique advantages in detecting critical features from complex breast cancer datasets, machine learning (ML) is widely recognized as the methodology of choice in breast cancer pattern classification. This project involves a study on the implementation of a Convolutional Neural Network (CNN) model using a dataset from the UCI repository. We have obtained an accuracy of 93% with the CNN model.

INTRODUCTION:

Breast cancer (BC) stands as the most prevalent cancer among women, impacting approximately 10 percent of women at some stage in their lives. With modern advancements, the incidence rate continues to rise. However, significant improvements in early detection and treatment have led to increased survival rates, with data indicating an 88 percent survival rate five years post-diagnosis and 80 percent after ten years. Early prediction and diagnosis have been pivotal, contributing to a 39 percent reduction in breast cancer mortality since 1989.
The diverse nature of breast cancer symptoms necessitates a comprehensive diagnostic approach, involving a variety of tests such as mammography, ultrasound, and biopsies. Among these, biopsy remains the most definitive, involving the extraction of sample cells or tissues through procedures like Fine Needle Aspiration (FNA). These samples are then examined microscopically, with numerical features such as radius, texture, perimeter, and area measured from the images. The data obtained from FNA, combined with other imaging data, aid in predicting the likelihood of malignancy in breast cancer tumors.
Automating this diagnostic process holds significant potential to expedite and enhance the accuracy of predictions, reducing the need for multiple, often painful, diagnostic tests such as mammography, ultrasound, and MRI, which also expose patients to radiation. By leveraging data mining methods and classification techniques, the number of false positive and false negative decisions can be minimized. This has led to the increasing adoption of data discovery in databases as a preferred tool for medical researchers.
In this study, we investigate the implementation of a Convolutional Neural Network (CNN) model for breast cancer prediction using the Wisconsin Breast Cancer (original) Datasets. The CNN model is evaluated based on its performance in accurately classifying malignant and benign tumors, with the aim of identifying the most effective method for early breast cancer detection.

PROBLEM STATEMENT:

Breast cancer, a leading cause of mortality among women, requires early and accurate diagnosis to improve survival rates. Traditional diagnostic methods, involving multiple tests, can be uncomfortable, time-consuming, and expose patients to radiation. There is a need for an automated, efficient system that accurately classifies tumors as malignant or benign. This project aims to develop a Convolutional Neural Network (CNN) model to predict breast cancer from Fine Needle Aspiration (FNA) data, reducing diagnostic burden and enhancing accuracy. The CNN model's performance will be benchmarked against traditional methods using the Wisconsin Breast Cancer (original) Datasets.

OBJECTIVE:

The primary objective of this project is to develop and evaluate a Convolutional Neural Network (CNN) model for the early prediction of breast cancer using data obtained from Fine Needle Aspiration (FNA) procedures. By leveraging advanced machine learning techniques, particularly CNNs, we aim to automate the diagnostic process, thereby reducing the need for multiple diagnostic tests such as mammography, ultrasound, and biopsies. This automation not only minimizes patient discomfort but also decreases their exposure to radiation. Additionally, we seek to enhance the accuracy of breast cancer diagnosis by utilizing the CNN model to decrease the rates of false positive and false negative diagnoses. To validate the effectiveness of our approach, we will benchmark the performance of the CNN model against traditional classification methods, including Decision Tree, K-Neighbors, Logistic Regression, Random Forest, and Support Vector Machine, using the Wisconsin Breast Cancer (original) Datasets. Through this comprehensive evaluation, we aim to contribute to the development of more efficient and accurate breast cancer diagnostic tools, ultimately improving patient care and outcomes.