Automated Gleason grading of prostate cancer tissue microarrays via deep learning

By Eirini Arvaniti, Kim S. Fricker, Michael Moret, N. Rupp, T. Hermanns, C. Fankhauser, Norbert Wey, P. Wild, J. Rüschoff, Manfred Claassen. Scientific Reports, 2018

425 citations recorded by the source corpus.

The Gleason grading system remains the most powerful prognostic predictor for patients with prostate cancer since the 1960s. Its application requires highly-trained pathologists, is tedious and yet suffers from limited inter-pathologist reproducibility, especially for the intermediate Gleason score 7. Automated annotation procedures constitute a viable solution to remedy these limitations. In this study, we present a deep learning approach for automated Gleason grading of prostate cancer tissue microarrays with Hematoxylin and Eosin (H&E) staining. Our system was trained using detailed Gleason annotations on a discovery cohort of 641 patients and was then evaluated on an independent test cohort of 245 patients annotated by two pathologists. On the test cohort, the inter-annotator agreements between the model and each pathologist, quantified via Cohen’s quadratic kappa statistic, were 0.75 and 0.71 respectively, comparable with the inter-pathologist agreement (kappa = 0.71). Furthermore, the model’s Gleason score assignments achieved pathology expert-level stratification of patients into prognostically distinct groups, on the basis of disease-specific survival data available for the test cohort. Overall, our study shows promising results regarding the applicability of deep learning-based solutions towards more objective and reproducible prostate cancer grading, especially for cases with heterogeneous Gleason patterns.

Subfields: Artificial Intelligence, Health Informatics, Radiology, Nuclear Medicine and Imaging

View Paper: https://www.semanticscholar.org/paper/57a892b9576baeba70277179712d5b09e19224b9

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Radiology, Nuclear Medicine and Imaging

Citations in the corpus

Survey on Deep Neural Networks in Speech and Vision Systems (2019, background): “…in deep neural networks have shown tremendous potential in almost all areas of medical imaging such as ophthalmology [215], dental radiography [216], skin cancer imaging [217], brain imaging [218], cardiac imaging [219, 220], urology [221], lung imaging [222], stroke imaging [223], and so on.”
Artificial intelligence as the next step towards precision pathology (2020, methodology): “Arvaniti et al. [93] used 641 TMA images of prostate cancer to train CNN algorithm for automated Gleason scoring.”
The impact of pre- and post-image processing techniques on deep learning frameworks: A comprehensive review for digital pathology image analysis (2020, methodology): “Using this strategy, Arvaniti et al. [39] obtained a global accuracy comparable with the inter-pathologist agreement.”
Deep neural network models for computational histopathology: A survey (2019, methodology): “allenge (400 WSIs) Bychkov et al. (2018) Colon H&E Outcome prediction of colorectal cancer A 3-layer LSTM + VGG16 pre-trained features to predict colorectal cancer outcome Private set (420 cases) Arvaniti et al. (2018) (X) Prostate H&E Predicting Gleason score Pre-trained MobileNet architect”
Computational pathology definitions, best practices, and recommendations for regulatory guidance: a white paper from the Digital Pathology Association (2019, methodology): “For example, algorithms have been described that identify images likely to contain tumor cells [21], compute mitotic counts [22], improve the accuracy and precision of immunohistochemistry scoring [23,24], or apply standardized histological scoring criteria such as the Gleason score [25], which can ”
Dense Steerable Filter CNNs for Exploiting Rotational Symmetry in Histology Images (2020, background): “[16]–[18]; mitosis detection [19]; cancer type prediction [20] and cancer grading [21], [22].”
Deep learning with noisy labels: exploring techniques and remedies in medical image analysis (2019, methodology): “2) Methods: We used a MobileNet CNN architecture, which had been shown to be a good choice for this application by [165], [166] and used patches of size 768× 768 pixels at 40X magnification as suggested by [165].”