Augmenting data with GANs to segment melanoma skin lesions

Abstract

This paper presents a novel strategy that employs Generative Adversarial Networks (GANs) to augment data in the skin lesion segmentation task, which is a fundamental first step in the automated melanoma detection process. The proposed framework generates both skin lesion images and their segmentation masks, making the data augmentation process extremely straightforward. In order to thoroughly analyze how the quality and diversity of synthetic images impact the efficiency of the method, we remodel two different well known GANs: a Deep Convolutional GAN (DCGAN) and a Laplacian GAN (LAPGAN). Experimental results reveal that, by introducing such kind of synthetic data into the training process, the overall accuracy of a stateof-the-art Convolutional/Deconvolutional Neural Network for melanoma skin lesion segmentation is increased.

Skin Lesion Segmentation Ensemble with Diverse Training Strategies

“…Generative Adversarial Networks (GANs) are often used to create unlabeled examples, which cannot be directly employed for the training of a supervised algorithm [38]. Following the approach introduced in [27,28], we improve the role of GANs in the training process by designing an architecture able to generate both the skin lesion image and its segmentation mask, making it extremely easy to exploit new synthetic images as additional training data. We modify the GAN proposed by Karras et al [19] in order to feed it 4-channels images: the first three channels are the R, G and B components and the fourth one is the binary segmentation mask.…”

How Does Connected Components Labeling with Decision Trees Perform on GPUs?

“…Efficiency of connected components labeling is critical in many real-time applications [13,29,30], and this is the reason why many strategies have been proposed for efficiently addressing the problem [10]. Traditionally, on sequential machines a two scan algorithm is employed.…”

Optimized Block-Based Algorithms to Label Connected Components on GPUs

“…Originally introduced by Rosenfeld and Pfaltz in 1966 [1], CCL has been in use for more than 50 years in multiple image processing and computer vision pipelines, including Object Tracking [2], Video Surveillance [3], Image Segmentation [4], [5], [6], Medical Imaging Applications [7], [8], [9], [10], Document Restoration [11], [12], Graph Analysis [13], [14], and Environmental Applications [15].…”