Breast Cancer Tumor Image Classification Using Deep Learning Image Data Generator
Aruna Kranthi Godishala, Hayati Yassin, Ramaswamy Veena, Daphne Teck Ching Lai · 2022 7th International Conference on Image, Vision and Computing (ICIVC) · 2022
Breast cancer is often encountered in females and is a prominent causative factor in mortality in women. Pathology needs time, but since technologies are limited, it is vital to design a system to effectively clear the uncertainty of breast cancer. For the categorization of benign and malignant tumors, many Deep Machine Learnings have been applied. Several of these techniques have already been assessed and compared in aspects of precision and accuracy. All approaches are programmed in Python and executed by either Google Collab, Jupyter Notebook, or Spyder which are Scientific Python Development Environment. Our problem statement deals with Deep learning for breast cancer risk prediction. It defines the prediction of the type of breast cancer using CNN as a deep learning technique that strengthens forecasting accuracy. It also demonstrates how deep learning technology can be used to diagnose breast cancer using Python. Our proposal undoubtedly explicates whether an individual is affected by a cancerous tumor (malignant) or non-cancerous tissue (benign). CNN can accept an input image, assign learnable weights and biases to various objects in the picture, and be able to differentiate one from the other. SVM and Random Forest Classifier were shown to be the most accurate for predictive modeling, with a 95.87 and 94.92 percent accuracy rate, respectively. In the instance of CNN, the utmost accuracy rate gained is 97.3 percent. In terms of statistics, activation functions like ReLu, and Sigmoid have been exploited to determine the upshots.