Active Appearance Model Induced Generative Adversarial Network for Controlled Data Augmentation

Abstract

Data augmentation is an important strategy for enlarging training datasets in deep learning-based medical image analysis. This is because large, annotated medical datasets are not only difficult and costly to generate, but also quickly become obsolete due to rapid advances in imaging technology. Image-to-image conditional generative adversarial networks (C-GAN) provide a potential solution for data augmentation. However, annotations used as inputs to C-GAN are typically based only on shape information, which can result in undesirable intensity distributions in the resulting artificially-created images. In this paper, we introduce an active cell appearance model (ACAM) that can measure statistical distributions of shape and intensity and use this ACAM model to guide C-GAN to generate more realistic images, which we call A-GAN. A-GAN provides an effective means for conveying anisotropic intensity information to C-GAN. A-GAN incorporates a statistical model (ACAM) to determine how transformations are applied for data augmentation. Traditional approaches for data augmentation that are based on arbitrary transformations might lead to unrealistic shape variations in an augmented dataset that are not representative of real data. A-GAN is designed to ameliorate this. To validate the effectiveness of using A-GAN for data augmentation, we assessed its performance on cell analysis in adaptive optics retinal imaging, which is a rapidly-changing medical imaging modality. Compared to C-GAN, A-GAN achieved stability in fewer iterations. The cell detection and segmentation accuracy when assisted by A-GAN augmentation was higher than that achieved with C-GAN. These findings demonstrate the potential for A-GAN to substantially improve existing data augmentation methods in medical image analysis.

GanglionNet: Objectively assess the density and distribution of ganglion cells with NABLA-N network

“…An active cell appearance model (ACAM) has been proposed which can measure the statistical distribution of shape and intensity of the cells. ACAM, when used with a guide Conditional Generative Adversarial Model (CGAN) and referred to as "AGAN", was evaluated for cells analysis in adaptive optics retinal imaging and achieved substantially improved results [9].…”

A multitask dual‐stream attention network for the identification of KRAS mutation in colorectal cancer

“…For the limitations of small data set, the most intuitive idea is to expand the data set to get more training samples. In addition to typical data expansion methods such as translation, rotation, and scaling, 17 there are many improved data generation methods based on generative adversarial networks 18–20 . However, in most cases, the low quality and high cost of medical image annotations are the real factors leading to small‐scale data set.…”

Realistic Adversarial Data Augmentation for MR Image Segmentation

“…affine, elastic transformations). Recently, there is a growing interest in developing generative network-based methods for data augmentation [5,6,7,8], which have been found effective for one-shot brain segmentation [5] and low-shot cardiac segmentation [7]. Unlike conventional data augmentation, which generates new examples in an uninformative fashion and does not account for complex variations in data, this generative network-based method is data-driven, learning optimal image transformations from the underlying data distribution in the real world [7].…”