Breast Tumour Classification using MobileNetV2
Shambhavi Kumar, Arohi Narang, Monica Parihar, Vidya Sawant · 2021 2nd Global Conference for Advancement in Technology (GCAT) · 2021
Breast Cancer is the most common cancer in Indian women. While a cure for cancer has not been discovered yet, it has been observed that the survival rate can reach 100% if it is detected early. Thus, our project aims to provide a model that gives an accurate and faster approach of classifying the tumor as benign or malignant present in the dataset of histopathology images which is a 3GB file. This model has used MobileNetV2 as the convolutional neural network which has fewer number of operations as compared to other architectures making the model training faster. MobileNetV2 is also ideal for mobile devices because of its compact size, fast computational and competitive performance in comparison to other models thus helping us achieve our aim better. The model is further automated by building an app using Flask to further improve the efficiency of cancer detection. The overall accuracy for the model is 89% and the sensitivity is 89.92%, specificity is 86.84% and AUC score is 88.38%.