Xin Wang

Xin Wang

Auto-generated profile, based on publicly available information

Xin Wang

Airdoc, Beijing, China

Research Focus & Metrics

Academic Career Length

19Years (2008-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
7 January 2022 in IEEE Transactions on Medical Imaging
Verified
Journal Article
Closed access
LJLie Ju
Lie Ju
XWXin Wang
Xin Wang
LWLin Wang
Lin Wang
DMDwarikanath Mahapatra
Dwarikanath Mahapatra
XZXin Zhao
Xin Zhao
ZGZongyuan Ge
Zongyuan Ge
Deep neural networks are known to be data-driven and label noise can have a marked impact on model performance. Recent studies have shown great robustness to classic image recognition even under a high noisy rate. In medical applications, learning from datasets with label noise is more challenging since medical imaging datasets tend to have instance-dependent noise (IDN) and suffer from high observer variability. In this paper, we systematically discuss the two common types of label noise in medical images - disagreement label noise from inconsistency expert opinions and single-target label noise from biased aggregation of individual annotations. We then propose an uncertainty estimation-based framework to handle these two label noise amid the medical image classification task. We design a dual-uncertainty estimation approach to measure the disagreement label noise and single-target label noise via improved Direct Uncertainty Prediction and Monte- Carlo-Dropout. A boosting-based curriculum training procedure is later introduced for robust learning. We demonstrate the effectiveness of our method by conducting extensive experiments on three different diseases with synthesized and real-world label noise: skin lesions, prostate cancer, and retinal diseases. We also release a large re-engineered database that consists of annotations from more than ten ophthalmologists with an unbiased golden standard dataset for evaluation and benchmarking. The dataset is available at https://mmai.group/peoples/julie/.

2 February 2021 in IEEE Transactions on Medical Imaging
Verified
Journal Article
Closed access
LJLie Ju
Lie Ju
XWXin Wang
Xin Wang
XZXin Zhao
Xin Zhao
PBPaul Bonnington
Paul Bonnington
TDTom Drummond
Tom Drummond
ZGZongyuan Ge
Zongyuan Ge
Recently, ultra-widefield (UWF) 200° fundus imaging by Optos cameras has gradually been introduced because of its broader insights for detecting more information on the fundus than regular 30° - 60° fundus cameras. Compared with UWF fundus images, regular fundus images contain a large amount of high-quality and well-annotated data. Due to the domain gap, models trained by regular fundus images to recognize UWF fundus images perform poorly. Hence, given that annotating medical data is labor intensive and time consuming, in this paper, we explore how to leverage regular fundus images to improve the limited UWF fundus data and annotations for more efficient training. We propose the use of a modified cycle generative adversarial network (CycleGAN) model to bridge the gap between regular and UWF fundus and generate additional UWF fundus images for training. A consistency regularization term is proposed in the loss of the GAN to improve and regulate the quality of the generated data. Our method does not require that images from the two domains be paired or even that the semantic labels be the same, which provides great convenience for data collection. Furthermore, we show that our method is robust to noise and errors introduced by the generated unlabeled data with the pseudo-labeling technique. We evaluated the effectiveness of our methods on several common fundus diseases and tasks, such as diabetic retinopathy (DR) classification, lesion detection and tessellated fundus segmentation. The experimental results demonstrate that our proposed method simultaneously achieves superior generalizability of the learned representations and performance improvements in multiple tasks.

19 January 2021 in IEEE Journal of Biomedical and Health Informatics
Verified
Journal Article
Closed access
LJLie Ju
Lie Ju
XWXin Wang
Xin Wang
XZXin Zhao
Xin Zhao
HLHuimin Lu
Huimin Lu
DMDwarikanath Mahapatra
Dwarikanath Mahapatra
ZGZongyuan Ge
Zongyuan Ge
The need for comprehensive and automated screening methods for retinal image classification has long been recognized. Well-qualified doctors annotated images are very expensive and only a limited amount of data is available for various retinal diseases such as diabetic retinopathy (DR) and age-related macular degeneration (AMD). Some studies show that some retinal diseases such as DR and AMD share some common features like haemorrhages and exudation but most classification algorithms only train those disease models independently when the only single label for one image is available. Inspired by multi-task learning where additional monitoring signals from various sources is beneficial to train a robust model. We propose a method called synergic adversarial label learning (SALL) which leverages relevant retinal disease labels in both semantic and feature space as additional signals and train the model in a collaborative manner using knowledge distillation. Our experiments on DR and AMD fundus image classification task demonstrate that the proposed method can significantly improve the accuracy of the model for grading diseases by 5.91% and 3.69% respectively. In addition, we conduct additional experiments to show the effectiveness of SALL from the aspects of reliability and interpretability in the context of medical imaging application.

1 September 2021 in Information Fusion
Verified
Journal Article
Closed access
WHWanji He
Wanji He
XWXin Wang
Xin Wang
LWLin Wang
Lin Wang
YHYelin Huang
Yelin Huang
ZYZhiwen Yang
Zhiwen Yang
ZGZongyuan Ge
Zongyuan Ge

5 October 2022 in Computers in Biology and Medicine
Verified
Journal Article
Closed access
LWLin Wang
Lin Wang
Xiufen YeXiufen Ye
Xiufen Ye
DZDonghao Zhang
Donghao Zhang
WHWanji He
Wanji He
LJLie Ju
Lie Ju
ZGZongyuan Ge
Zongyuan Ge

1 January 2008
Verified
Book
Open access
XWXin Wang
Xin Wang
Image based rendering (IBR) has been the most Important and rapid developed techniques in the computer graphics and virtual reality fields these years. Image mosaic which is one of the hot topics of IBR is also becoming research interests of many researchers in the image processing and computer vision fields. Its application covers the areas of virtual scene construction, remote sensing, medical image and military affairs etc. However, some difficult issues need to be studied further, including new optimization methods for image registration, new accelerating methods for image stitching etc, which are the main topics of this thesis.\ud \ud First, as the precision and automatic degree of image mosaic suffers from the algorithm of image registration, a new image stitching optimization method based on maximum mutual information is presented in this thesis. The main idea of the new method is to combine PSO algorithm with wavelet multiresolution strategy and parameters of PSO are adapted along with the resolution of the images.The experiments show that this method can avoid registration process to get stuck into local extremes in image interpolation operations and finds the optimal exchange through limited iterations computation, and obtain subpixel registration accuracy in process of image stitching.\ud \ud Secondly, to solve the problem of image blur stitching when the geometric deformation and the changes of the scale factor among images are serious, a new method based robust features matching is proposed in this thesis. Firest, it searches overlap area between two images by using phase correlation, and then detects Harris comer points in the overlap areas which are reduced to different scale images by adopting multi resolution pyramid construction. This method can solve the Harris arithmetic operator robust characteristics inspection algorithm for the scale effects. In order to improve the running performance of image feature matching, a dimension reduction method of high dimension feature vector is proposed based on PCA. And last the globe parameters are optimized by Lmeds to realize images mosaic. The experiments prove that the methods proposed by this thesis can reduce the computation cost with guarantee of image mosaic quality