Abstract:Data augmentation is widely applied to medical image analysis tasks in limited datasets with imbalanced classes and insufficient annotations. However, traditional augmentation techniques cannot supply extra information, making the performance of diagnosis unsatisfactory. GAN-based generative methods have thus been proposed to obtain additional useful information to realize more effective data augmentation; but existing generative data augmentation techniques mainly encounter two problems: (i) Current generative data augmentation lacks of the capability in using cross-domain differential information to extend limited datasets. (ii) The existing generative methods cannot provide effective supervised information in medical image segmentation tasks. To solve these problems, we propose an attention-guided cross-domain tumor image generation model (CDA-GAN) with an information enhancement strategy. The CDA-GAN can generate diverse samples to expand the scale of datasets, improving the performance of medical image diagnosis and treatment tasks. In particular, we incorporate channel attention into a CycleGAN-based cross-domain generation network that captures inter-domain information and generates positive or negative samples of brain tumors. In addition, we propose a semi-supervised spatial attention strategy to guide spatial information of features at the pixel level in tumor generation. Furthermore, we add spectral normalization to prevent the discriminator from mode collapse and stabilize the training procedure. Finally, to resolve an inapplicability problem in the segmentation task, we further propose an application strategy of using this data augmentation model to achieve more accurate medical image segmentation with limited data. Experimental studies on two public brain tumor datasets (BraTS and TCIA) show that the proposed CDA-GAN model greatly outperforms the state-of-the-art generative data augmentation in both practical medical image classification tasks and segmentation tasks; e.g. CDA-GAN is 0.50%, 1.72%, 2.05%, and 0.21% better than the best SOTA baseline in terms of ACC, AUC, Recall, and F1, respectively, in the classification task of BraTS, while its improvements w.r.t. the best SOTA baseline in terms of Dice, Sens, HD95, and mIOU, in the segmentation task of TCIA are 2.50%, 0.90%, 14.96%, and 4.18%, respectively.

End-To-End Chromosome Karyotyping with Data Augmentation Using GAN

A New Multiple-Distribution GAN Model to Solve Complexity in End-to-End Chromosome Karyotyping

Enhanced Rotated Mask R-CNN for Chromosome Segmentation.

A novel end-to-end chromosome classification approach using deep neural network with triple attention mechanism

Deep Anomaly Generation: An Image Translation Approach of Synthesizing Abnormal Banded Chromosome Images

Statistical Karyotype Analysis Using Cnn And Geometric Optimization

Chromosome classification via deep learning and its application to patients with structural abnormalities of chromosomes

Attention-Based Densely Connected Convolutional Network for Chromosome Classification.

An End-to-End Combinatorial Optimization Method for R-band Chromosome Recognition with Grouping Guided Attention

CellGAN: Conditional Cervical Cell Synthesis for Augmenting Cytopathological Image Classification

Chromosome Segmentation Analysis Using Image Processing Techniques and Autoencoders

Conditional Infilling GANs for Data Augmentation in Mammogram Classification

Population-scale Genomic Data Augmentation Based on Conditional Generative Adversarial Networks

Adversarial Multiscale Feature Learning Framework for Overlapping Chromosome Segmentation

Classification of Metaphase Chromosomes Using Deep Convolutional Neural Network.

Cross-domain attention-guided generative data augmentation for medical image analysis with limited data

KaryoNet: Chromosome Recognition with End-to-End Combinatorial Optimization Network.

DeepACEv2: Automated Chromosome Enumeration in Metaphase Cell Images Using Deep Convolutional Neural Networks

Enhanced label constrained contrastive learning for chromosome optical microscopic image classification

Emotion Classification with Data Augmentation Using Generative Adversarial Networks

Additional Look into GAN-based Augmentation for Deep Learning COVID-19 Image Classification