StainCNNs: An efficient stain feature learning method

Lei, Gaoyi; Xia, Yuanqing; Zhai, Di‐Hua; Zhang, Wei; Chen, Duanduan; Wang, Defeng

doi:10.1016/j.neucom.2020.04.008

Cited by 15 publications

(9 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…StainCNNs: Inspired by SPCN, Lei et al. proposed a deep neural network for stain separation to reduce the computational consumption of SNMF [110] . The proposed stainCNNs approach took the source images as input and learned to generate the stain colour appearance matrix.…”

Section: Data Harmonisation Strategies For Information Fusionmentioning

confidence: 99%

“…Feature similarity index (FSIM) utilises phase congruency and gradient magnitude features to evaluate the low-level features of image visual quality [150] . The QSSIM and FSIM were employed in [ 110 , 138 ] and [ 105 , 110 ], respectively, to assess the structural preserving conditions after the harmonisation process. Though most methods applied structural similarity related metrics for evaluation, studies have shown their limitations and weaknesses [151] .…”

Section: Evaluation Approaches Of the Data Harmonisation Strategiesmentioning

confidence: 99%

See 1 more Smart Citation

Data harmonisation for information fusion in digital healthcare: A state-of-the-art systematic review, meta-analysis and future research directions

et al. 2022

View full text Add to dashboard Cite

Section: Data Harmonisation Strategies For Information Fusionmentioning

confidence: 99%

Section: Evaluation Approaches Of the Data Harmonisation Strategiesmentioning

confidence: 99%

Data harmonisation for information fusion in digital healthcare: A state-of-the-art systematic review, meta-analysis and future research directions

et al. 2022

View full text Add to dashboard Cite

“…On the other hand, Cho et al (18), Salehi et al (7), and Tellez et al (17) reconstructed original images from the images with color augmentations, e.g., grayscale and Hue-Saturation-Value (HSV) transformation, and tried to normalize other color styles to the original. However, due to the complexity of deep neural networks and the instability of GANs, it is hard to preserve all source information; sometimes, it has a risk of introducing some artifacts, which has some adverse effects on subsequent analysis (19). At the same time, the network of deep learning-based methods usually contains millions of parameters, so it generally requires high-computing resources and the computing efficiency is generally low (14).…”

Section: Introductionmentioning

confidence: 99%

“…However, Pap stain used in cervical cytopathology involves not only Hematoxylin and Eosin but also Orange, Light Green, and Bismarck Brown [19], which is very difficult to separate each staining channel independently. Nevertheless, most conventional methods need a reference image to estimate stain parameters, but the information in an image patch could not cover the staining phenomena of the entire tissue section or represent all input images, which usually caused misestimation of stain parameters and thus delivered inaccurate normalization results [18,21].…”

Section: Introductionmentioning

confidence: 99%

StainNet: A Fast and Robust Stain Normalization Network

et al. 2021

View full text Add to dashboard Cite

Stain normalization often refers to transferring the color distribution to the target image and has been widely used in biomedical image analysis. The conventional stain normalization usually achieves through a pixel-by-pixel color mapping model, which depends on one reference image, and it is hard to achieve accurately the style transformation between image datasets. In principle, this difficulty can be well-solved by deep learning-based methods, whereas, its complicated structure results in low computational efficiency and artifacts in the style transformation, which has restricted the practical application. Here, we use distillation learning to reduce the complexity of deep learning methods and a fast and robust network called StainNet to learn the color mapping between the source image and the target image. StainNet can learn the color mapping relationship from a whole dataset and adjust the color value in a pixel-to-pixel manner. The pixel-to-pixel manner restricts the network size and avoids artifacts in the style transformation. The results on the cytopathology and histopathology datasets show that StainNet can achieve comparable performance to the deep learning-based methods. Computation results demonstrate StainNet is more than 40 times faster than StainGAN and can normalize a 100,000 × 100,000 whole slide image in 40 s.

show abstract

“…To tackle the issues of traditional methods, color normalization methods based on deep learning [26] , [28] , [29] , [30] , [31] , [32] , [33] , [34] , [35] especially generative adversarial networks (GANs) [36] are proposed. In contrast to traditional color normalization methods, GAN-based methods consider the overall dataset of target style as the template and approach the problem of color normalization as image-to-image translation [28] , [29] , [30] .…”

Section: Introductionmentioning

confidence: 99%