A Hybrid Convolutional and Recurrent Deep Neural Network for Breast Cancer Pathological Image Classification

Yan, Rui; Ren, Fei; Wang, Zihao; Wang, Lihua; Ren, Yanzhi; Liu, Yudong; Rao, Xiaosong; Zheng, Chun-Hou; Zhang, Fa

doi:10.1109/bibm.2018.8621429

Cited by 19 publications

(14 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Extended Bioimaging2015 dataset [21] This dataset was an extension of the Bioimaging2015 dataset and contained 1319 hematoxylin and eosin color images that were high resolution (2048 × 1536 pixels), uncompressed, and annotated. All images were digital, and the acquisition conditions were the same: the magnification is 200×, and the pixel size is 0.42 × 0.42 µm.…”

Section: Datasetsmentioning

confidence: 99%

“…Our Single Model 97.5% Our Ensemble Model 97.5% CNN + BiLSTM (serial architecture) [21] 82.1% Figure 6. The schematic diagram of the Acc-Loss training process of the model and the confusion matrix of the best results on the extended Bioimaging2015 dataset [21]. Table 8.…”

Section: Accuracy On the Test Setmentioning

confidence: 99%

“…The most well-known and influential human breast cancer datasets include BACH2018 [19], Bioimaging2015 [20], and extended Bioimaging2015 [21]. Meng et al [22] proposed a gravitational loss (G-loss) as a new loss function to maximize interclass differences and minimize intraclass variance.…”

Section: Introductionmentioning

confidence: 99%

“…To date, this is the best result on the Bioimaging2015 dataset. Yan [21] fine-tuned the Inception V3 network as an image feature extractor and inputted the extracted features into an LSTM model. Yan's method achieved the best (82.1%) patch-wise accuracy on the extended Bioimaging2015 dataset.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Parallel Structure Deep Neural Network Using CNN and RNN with an Attention Mechanism for Breast Cancer Histology Image Classification

Yao

Zhang

Zhou

et al. 2019

Cancers

View full text Add to dashboard Cite

In this paper, we present a new deep learning model to classify hematoxylin–eosin-stained breast biopsy images into four classes (normal tissues, benign lesions, in situ carcinomas, and invasive carcinomas). Our model uses a parallel structure consist of a convolutional neural network (CNN) and a recurrent neural network (RNN) for image feature extraction, which is greatly different from the common existed serial method of extracting image features by CNN and then inputting them into RNN. Then, we introduce a special perceptron attention mechanism, which is derived from the natural language processing (NLP) field, to unify the features extracted by the two different neural network structures of the model. In the convolution layer, general batch normalization is replaced by the new switchable normalization method. And the latest regularization technology, targeted dropout, is used to substitute for the general dropout in the last three fully connected layers of the model. In the testing phase, we use the model fusion method and test time augmentation technology on three different datasets of hematoxylin–eosin-stained breast biopsy images. The results demonstrate that our model significantly outperforms state-of-the-art methods.

show abstract

Section: Datasetsmentioning

confidence: 99%

Section: Accuracy On the Test Setmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Parallel Structure Deep Neural Network Using CNN and RNN with an Attention Mechanism for Breast Cancer Histology Image Classification

Yao

Zhang

Zhou

et al. 2019

Cancers

View full text Add to dashboard Cite

show abstract

“…A hematoxilinaé responsável pela coloração do DNA do núcleo celular, do RNA do citoplasma e da cartilagem, produzindo uma cor azulada nas estruturas [Roy et al 2019]. Enquanto a eosina tinge estruturas do citoplasma e o colágeno de tons rosas [Yan et al 2019]. O processo de coloraçãoé necessário por que são a maior parte das células e material extracelular serem incolores [Mescher 2018].…”

Section: Introductionunclassified

Análise comparativa da influência dos espaços de cores na segmentação multi-classe de Whole Slide Imaging do câncer de mama utilizando deep learning

Zeiser¹,

Costa²,

Ramos³

et al. 2020

Anais Principais Do Simpósio Brasileiro De Computação Aplicada À Saúde (SBCAS 2020)

View full text Add to dashboard Cite

O diagnóstico em estágios iniciais normalmente resulta em um prognóstico melhor nos casos de câncer. Atualmente, a análise histopatológica e o padrão ouro para o diagnóstico, estadiamento e definição do tratamento de neoplasias de mama. Contudo, a técnica possui algumas restrições que dificultam o processo de análise pelo patologista, como diferentes protocolos de coloração, variações de coloração nas lâminas, e sobreposição de tecidos. Assim, a representação computacional das cores pode exercer uma influência significativa no comportamento dos modelos convolucionais. Desta forma, este estudo propõe uma análise comparativa da influência de quatro espaços de cores (RGB, HSV, YCrCb e LAB) para a segmentação multi-classe de Whole Slide Imaging (WSI) do câncer de mama. A metodologia proposta e composta por uma etapa de pré-processamento das WSI, data augmentation e multi- segmentação utilizando a arquitetura convolucional U-Net com uma ResNet-50 pré-treinada como codificador. Os resultados obtidos demonstram que o espaço de cores HSV apresentou de maneira geral índices melhores para a segmentação das WSI utilizando a metodologia proposta neste estudo.

show abstract

Computer Aided System for Nuclei Localization in Histopathological Images Using CNN

Kanojia

Ansari²,

Gandhi

et al. 2020

Advances in Intelligent Systems and Computing

View full text Add to dashboard Cite

A Hybrid Convolutional and Recurrent Deep Neural Network for Breast Cancer Pathological Image Classification

Cited by 19 publications

References 12 publications

Parallel Structure Deep Neural Network Using CNN and RNN with an Attention Mechanism for Breast Cancer Histology Image Classification

Parallel Structure Deep Neural Network Using CNN and RNN with an Attention Mechanism for Breast Cancer Histology Image Classification

Análise comparativa da influência dos espaços de cores na segmentação multi-classe de Whole Slide Imaging do câncer de mama utilizando deep learning

Computer Aided System for Nuclei Localization in Histopathological Images Using CNN

Contact Info

Product

Resources

About