A Cancer Cell Image Classification Program : Based on CNN Model

Abstract

Breast Cancer, as a deadly existence, has negatively influenced women’s health. To determine whether a cell is benign or malignant is critical for doctors to diagnose breast cancer. However, simply judging whether the cells are malignant by the appearance of the X-ray’s outcome greatly reduces the efficiency of diagnosis and the probability of misdiagnosis. This paper proposed a simulation model to tackle this issue, which can be utilized to analyze and determine whether the cell is benign or malignant. With the help of CNN and Text-to-speech, I have researched a solution for doctors to identify Breast Cancer with the help of machine learning. The experiment consists of Image classification built upon CNN and a user interface for upload functionality by Streamlit framework, combined with an NLP speech synthesis interface to communicate the result done with gTTS. The result of the experiment has brought up a fast, efficient, accurate result after the model has been trained measured by sensitivity and specificitysignflcantly reduced the amount of error rate. Building an interface that interacts with the result allows the result to be more visualized and straightforward.