A Conceptual Framework of Deploying a Trained CNN Model for Mobile Breast Self-Screening

In this article, we propose a conceptual framework of mobile self-screening to detect breast cancer at the initial stage using temperature characteristics. A trained convolutional neural network (CNN) is implemented to extract and classify the breast thermogram. Moreover, a histogram of temperature distribution is also provided to support a recommendation to the user. As we consider the user's data privacy, reliability, low latency, and low cost, we also suggest some recommendations in inferencing the CNN model on the mobile device. Finally, we describe the application overview to demonstrate the feature function.