Deep Neural Network for Transformation Zone Classification

Abstract

The type of the treatment that a patient undergoes depends on the type of transformation zone of the cervix of patient. It is difficult to identify the thin line difference between the various types of transformation zones using naked eyes. The application of deep learning can help medical practitioners for identifying the type with confidence. The survival rate of cancer patient will be higher if the affected cervix is diagnosed early in pre-cancerous stage. In this paper we present our work in developing a ConvNet (Convolutional Neural Network) to classify the transformation zones using cervix images provided by Kaggle. Our proposed model uses fine tuned transfer learning approach. We suspect more image augmentation methods can help to improve the overall performance of the model.