Cytokeratin-Supervised Deep Learning for Automatic Recognition of Epithelial Cells in Breast Cancers Stained for ER, PR, and Ki-67

Valkonen, Mira; Isola, Jorma; Ylinen, Onni; Muhonen, Ville; Saxlin, Anna; Tolonen, Teemu; Nykter, Matti; Ruusuvuori, Pekka

doi:10.1109/tmi.2019.2933656

Cited by 47 publications

(42 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our choice of the base deep neural network architecture was motivated by the relatively simple adaptability of the VGG-16 architecture, which can be done simply by adding problem domain spedific layers into a generic VGG-16 network. We have successfully used similar transfer learning strategy of extending the generic VGG-16 network into a specific problem domain in histopathology earlier in a study where imageto-image transform from immunohistochemichal staining to cytokeratin staning mask was done using VGG-16 based architecture [33]. In the current study, the use of a generic, well tested architecture underlines the applicability of the proposed domain adaptation approach.…”

Section: B Nuclei Detection Modelmentioning

confidence: 78%

See 1 more Smart Citation

Generalized Fixation Invariant Nuclei Detection Through Domain Adaptation Based Deep Learning

Valkonen

Högnäs

Bova

et al. 2021

IEEE J. Biomed. Health Inform.

Self Cite

View full text Add to dashboard Cite

Nucleus detection is a fundamental task in histological image analysis and an important tool for many follow up analyses. It is known that sample preparation and scanning procedure of histological slides introduce a great amount of variability to the histological images and poses challenges for automated nucleus detection. Here, we studied the effect of histopathological sample fixation on the accuracy of a deep learning based nuclei detection model trained with hematoxylin and eosin stained images. We experimented with training data that includes three methods of fixation; PAXgene, formalin and frozen, and studied the detection accuracy results of various convolutional neural networks. Our results indicate that the variability introduced during sample preparation affects the generalization of a model and should be considered when building accurate and robust nuclei detection algorithms. Our dataset includes over 67 000 annotated nuclei locations from 16 patients and three different sample fixation types. The dataset provides excellent basis for building an accurate and robust nuclei detection model, and combined with unsupervised domain adaptation, the workflow allows generalization to images from unseen domains, including different tissues and images from different labs.

show abstract

Section: B Nuclei Detection Modelmentioning

confidence: 78%

“…new domain (see Supplementary Figure 3 for such examples using the algorithm presented here and IHC data from [33].…”

Section: Conclusion and Discussionmentioning

confidence: 99%

Generalized Fixation Invariant Nuclei Detection Through Domain Adaptation Based Deep Learning

Valkonen

Högnäs

Bova

et al. 2021

IEEE J. Biomed. Health Inform.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Indeed, CNNs have already shown promising results in quantifying TILs in hematoxylin and eosin (HE) stained tissue samples [21]. Furthermore, reproducibility can even further be improved when using leukocyte-specific immunohistochemical (IHC) stains as a reference when annotating leukocytes in HE stained samples [22], [23]. However, the tissue morphology might significantly change in consecutive tissue sections, particularly on cell level.…”

Section: Convolutional Neuralmentioning

confidence: 99%

Antibody Supervised Training of a Deep Learning Based Algorithm for Leukocyte Segmentation in Papillary Thyroid Carcinoma

Stenman

Bychkov

Kücükel

et al. 2021

IEEE J. Biomed. Health Inform.

View full text Add to dashboard Cite

The quantity of leukocytes in papillary thyroid carcinoma (PTC) potentially have prognostic and treatment predictive value. Here, we propose a novel method for training a convolutional neural network (CNN) algorithm for segmenting leukocytes in PTCs. Tissue samples from two retrospective PTC cohort were obtained and representative tissue slides from twelve patients were stained with hematoxylin and eosin (HE) and digitized. Then, the HE slides were destained and restained immunohistochemically (IHC) with antibodies to the pan-leukocyte anti CD45 antigen and scanned again. The two stain-pairs of all representative tissue slides were registered, and image tiles of regions of interests were exported. The image tiles were processed and the 3,3-diaminobenzidine (DAB) stained areas representing anti CD45 expression were turned into binary masks. These binary masks were applied as annotations on the HE image tiles and used in the training of a CNN algorithm. Ten whole slide images (WSIs) were used for training using a five-fold cross-validation and the remaining two slides were used as an independent test set for the trained model. For visual

show abstract

“…Convolution neural networks (CNNs) are a fundamental class of deep learning networks that can be trained to detect, segment, and classify objects using large learning data sets [ 1 , 3 ]. CNNs are well-suited to perform complex visual recognition tasks, such as tumor detection, Gleason grading [ 4 , 5 ], scoring of tissue stains [ 6 , 7 ], as well as determining prognosis [ 8 ], and are emerging as a core method in medical image analysis.…”

Section: Introductionmentioning

confidence: 99%

PTEN and DNA Ploidy Status by Machine Learning in Prostate Cancer

Cyll

Kleppe

Kalsnes

et al. 2021

Cancers

View full text Add to dashboard Cite

Machine learning (ML) is expected to improve biomarker assessment. Using convolution neural networks, we developed a fully-automated method for assessing PTEN protein status in immunohistochemically-stained slides using a radical prostatectomy (RP) cohort (n = 253). It was validated according to a predefined protocol in an independent RP cohort (n = 259), alone and by measuring its prognostic value in combination with DNA ploidy status determined by ML-based image cytometry. In the primary analysis, automatically assessed dichotomized PTEN status was associated with time to biochemical recurrence (TTBCR) (hazard ratio (HR) = 3.32, 95% CI 2.05 to 5.38). Patients with both non-diploid tumors and PTEN-low had an HR of 4.63 (95% CI 2.50 to 8.57), while patients with one of these characteristics had an HR of 1.94 (95% CI 1.15 to 3.30), compared to patients with diploid tumors and PTEN-high, in univariable analysis of TTBCR in the validation cohort. Automatic PTEN scoring was strongly predictive of the PTEN status assessed by human experts (area under the curve 0.987 (95% CI 0.968 to 0.994)). This suggests that PTEN status can be accurately assessed using ML, and that the combined marker of automatically assessed PTEN and DNA ploidy status may provide an objective supplement to the existing risk stratification factors in prostate cancer.

show abstract

Cytokeratin-Supervised Deep Learning for Automatic Recognition of Epithelial Cells in Breast Cancers Stained for ER, PR, and Ki-67

Cited by 47 publications

References 21 publications

Generalized Fixation Invariant Nuclei Detection Through Domain Adaptation Based Deep Learning

Generalized Fixation Invariant Nuclei Detection Through Domain Adaptation Based Deep Learning

Antibody Supervised Training of a Deep Learning Based Algorithm for Leukocyte Segmentation in Papillary Thyroid Carcinoma

PTEN and DNA Ploidy Status by Machine Learning in Prostate Cancer

Contact Info

Product

Resources

About