Automatic segmentation of whole-slide H&amp;E stained breast histopathology images using a deep convolutional neural network architecture

Torres, Blanca Maria Priego; Morillo, Daniel; Fernández-Granero, Miguel Ángel; García-Rojo, Marcial

doi:10.1016/j.eswa.2020.113387

Cited by 62 publications

(30 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In another interesting study, Shahidi 24 proposed the method to use super-resolution generative adversarial networks to generate histopathology images, they tested their approach on CAMELYON16 dataset 25 . Priego-Torres et al 17 propose a segmentation pipeline for breast cancer images, using a patch-based approach, where the patches we extracted from all possible regions in an image and later merged with fully connected conditional random fields (CRF). Roy et al 19 presented a multi-resolution-based deep learning approach along with customized reconstruction loss for viable tumor segmentation in liver WSI images.…”

Section: Related Workmentioning

confidence: 99%

A generalized deep learning framework for whole-slide image segmentation and analysis

Khened

Kori

Rajkumar

et al. 2021

Sci Rep

137

View full text Add to dashboard Cite

Histopathology tissue analysis is considered the gold standard in cancer diagnosis and prognosis. Whole-slide imaging (WSI), i.e., the scanning and digitization of entire histology slides, are now being adopted across the world in pathology labs. Trained histopathologists can provide an accurate diagnosis of biopsy specimens based on WSI data. Given the dimensionality of WSIs and the increase in the number of potential cancer cases, analyzing these images is a time-consuming process. Automated segmentation of tumorous tissue helps in elevating the precision, speed, and reproducibility of research. In the recent past, deep learning-based techniques have provided state-of-the-art results in a wide variety of image analysis tasks, including the analysis of digitized slides. However, deep learning-based solutions pose many technical challenges, including the large size of WSI data, heterogeneity in images, and complexity of features. In this study, we propose a generalized deep learning-based framework for histopathology tissue analysis to address these challenges. Our framework is, in essence, a sequence of individual techniques in the preprocessing-training-inference pipeline which, in conjunction, improve the efficiency and the generalizability of the analysis. The combination of techniques we have introduced includes an ensemble segmentation model, division of the WSI into smaller overlapping patches while addressing class imbalances, efficient techniques for inference, and an efficient, patch-based uncertainty estimation framework. Our ensemble consists of DenseNet-121, Inception-ResNet-V2, and DeeplabV3Plus, where all the networks were trained end to end for every task. We demonstrate the efficacy and improved generalizability of our framework by evaluating it on a variety of histopathology tasks including breast cancer metastases (CAMELYON), colon cancer (DigestPath), and liver cancer (PAIP). Our proposed framework has state-of-the-art performance across all these tasks and is ranked within the top 5 currently for the challenges based on these datasets. The entire framework along with the trained models and the related documentation are made freely available at GitHub and PyPi. Our framework is expected to aid histopathologists in accurate and efficient initial diagnosis. Moreover, the estimated uncertainty maps will help clinicians to make informed decisions and further treatment planning or analysis.

show abstract

Section: Related Workmentioning

confidence: 99%

A generalized deep learning framework for whole-slide image segmentation and analysis

Khened

Kori

Rajkumar

et al. 2021

Sci Rep

137

View full text Add to dashboard Cite

show abstract

“…In [149], an automatic WSIs segmentation method based on DCNN is proposed. The method is effective regardless of the texture features in malignant tumors.…”

Section: B: Segmentationmentioning

confidence: 99%

A Comprehensive Review for Breast Histopathology Image Analysis Using Classical and Deep Neural Networks

et al. 2020

View full text Add to dashboard Cite

Breast cancer is one of the most common and deadliest cancers among women. Since histopathological images contain sufficient phenotypic information, they play an indispensable role in the diagnosis and treatment of breast cancers. To improve the accuracy and objectivity of Breast Histopathological Image Analysis (BHIA), Artificial Neural Network (ANN) approaches are widely used in the segmentation and classification tasks of breast histopathological images. In this review, we present a comprehensive overview of the BHIA techniques based on ANNs. First of all, we categorize the BHIA systems into classical and deep neural networks for in-depth investigation. Then, the relevant studies based on BHIA systems are presented. After that, we analyze the existing models to discover the most suitable algorithms. Finally, publicly accessible datasets, along with their download links, are provided for the convenience of future researchers.

show abstract

“…The proposed approach achieved an accuracy rate 55% while the pathologists achieved an accuracy rate of 65.44% for the 5 considered classes: non-proliferative changes, proliferative changes, atypical ductal hyperplasia, DC in situ, and IDCa. The same problem was investigated by Priego-Torres et al [4] who used fully connected Conditional Random Field (CRF) to merge the classification results of the different patches. Similarly, the authors of [17] considered WSIs while focusing only on the detection of IDC.…”

Section: Related Workmentioning

confidence: 99%

“…Recent research demonstrated that multiple DCNNs performs better than a single DCNN [3][4][5]. Thus, two DCNNs are trained independently and their feature vectors are combined by a concatenation and further processings to simplify the model complexity.…”

Section: Introductionmentioning

confidence: 99%

Deep Learning Models Combining for Breast Cancer Histopathology Image Classification

Elmannai¹,

Hamdi²,

Algarni³

2021

IJCIS

View full text Add to dashboard Cite

Breast cancer is one of the foremost reasons of death among women in the world. It has the largest mortality rate compared to the types of cancer accounting for 1.9 million per year in 2020. An early diagnosis may increase the survival rates. To this end, automating the analysis and the diagnosis allows to improve the accuracy and to reduce processing time. However, analyzing breast imagery's is non-trivial and may lead to experts' disagreements. In this research, we focus on breast cancer histopathological images acquired using the microscopic scan of breast tissues. We present combined two deep convolutional neural networks (DCNNs) to extract distinguished image features using transfer learning. The pre-trained Inception and the Xceptions models are used in parallel. Then, the feature maps are combined and reduced by dropout before being fed to the last fully connected layers for classification. We follow a sub-image classification then a whole image classification based on majority vote and maximum probability rules. Four tissue malignancy levels are considered: normal, benign, in situ carcinoma, and invasive carcinoma. The experimentations are performed to the Breast Cancer Histology (BACH) dataset. The overall accuracy for the sub-image classification is 97.29% and for the carcinoma cases the sensitivity achieved 99.58%. The whole image classification overall accuracy reaches 100% by majority vote and 95% by maximum probability fusion decision. The numerical results showed that our proposed approach outperforms the previous methods in terms of accuracy and sensitivity. The proposed design allows an extension to whole-slide histology images classification.

show abstract

Automatic segmentation of whole-slide H&E stained breast histopathology images using a deep convolutional neural network architecture

Cited by 62 publications

References 38 publications

A generalized deep learning framework for whole-slide image segmentation and analysis

A generalized deep learning framework for whole-slide image segmentation and analysis

A Comprehensive Review for Breast Histopathology Image Analysis Using Classical and Deep Neural Networks

Deep Learning Models Combining for Breast Cancer Histopathology Image Classification

Contact Info

Product

Resources

About