Producing Histopathology Phantom Images using Generative Adversarial Networks to improve Tumor Detection

TL;DR

Problem: imbalanced histopathology datasets across cancer types hinder robust tumor detection. Approach: apply GAN-based data augmentation to synthesize phantom histology images and balance the dataset; Evaluation: a $50\%$ augmentation raises CNN accuracy from $80%$ to $87.0\%$ with $FID=35.495$ indicating realistic generation. Impact: demonstrates GANs as a practical tool for equalizing cancer-type representation and enhancing automated pathology analysis, while noting limitations and proposing broader validations across datasets and GAN architectures.

Abstract

Advance in medical imaging is an important part in deep learning research. One of the goals of computer vision is development of a holistic, comprehensive model which can identify tumors from histology slides obtained via biopsies. A major problem that stands in the way is lack of data for a few cancer-types. In this paper, we ascertain that data augmentation using GANs can be a viable solution to reduce the unevenness in the distribution of different cancer types in our dataset. Our demonstration showed that a dataset augmented to a 50% increase causes an increase in tumor detection from 80% to 87.5%

Figures (1)

• Figure 1: Overarching Structure: The Original Dataset $F_o$ is input into the Convolutional Network and the Augmented Dataset $F_a$ is input into the Convolutional Network