Automated detection of cancer metastases in lymph nodes has the potential to improve assessment of prognosis for patients. To enable fair comparison between the algorithms for this purpose, we set up the CAMELYON17 challenge in conjunction with the IEEE International Symposium on Biomedical Imaging 2017 conference in Melbourne. Over 300 participants registered on the challenge website, of which 23 teams submitted a total of 37 algorithms before the initial deadline. Participants were provided with 899 whole-slide images for developing their algorithms. The developed algorithms were evaluated based on the test set encompassing 100 patients and 500 whole-slide images. The evaluation metric used was a quadratic weighted Cohen’s kappa. We discuss the algorithmic details of the ten best pre-conference and two post-conference submissions. All these participants used convolutional neural networks in combination with pre-and postprocessing steps. Algorithms differed mostly in neural network architecture, training strategy and pre-and postprocessing methodology. Overall, the kappa metric ranged from 0.89 to -0.13 across all submissions. The best results were obtained with pre-trained architectures such as ResNet. Confusion matrix analysis revealed that all participants struggled with reliably identifying isolated tumor cells, the smallest type of metastasis, with detection rates below 40%. Qualitative inspection of the results of the top participants showed categories of false positives, such as nerves or contamination, which could be targets for further optimization. Last, we show that simple combinations of the top algorithms result in higher kappa metric values than any algorithm individually, with 0.93 for the best combination.

Artificial intelligence (AI) has shown promise for diagnosing prostate cancer in biopsies. However, results have been limited to individual studies, lacking validation in multinational settings. Competitions have been shown to be accelerators for medical imaging innovations, but their impact is hindered by lack of reproducibility and independent validation. With this in mind, we organized the PANDA challenge—the largest histopathology competition to date, joined by 1,290 developers—to catalyze development of reproducible AI algorithms for Gleason grading using 10,616 digitized prostate biopsies. We validated that a diverse set of submitted algorithms reached pathologist-level performance on independent cross-continental cohorts, fully blinded to the algorithm developers. On United States and European external validation sets, the algorithms achieved agreements of 0.862 (quadratically weighted κ, 95% confidence interval (CI), 0.840–0.884) and 0.868 (95% CI, 0.835–0.900) with expert uropathologists. Successful generalization across different patient populations, laboratories and reference standards, achieved by a variety of algorithmic approaches, warrants evaluating AI-based Gleason grading in prospective clinical trials.

Significance Statement: Histopathologic assessment of kidney tissue currently relies on manual scoring or traditional image-processing techniques to quantify and classify tissue features, time-consuming approaches that have limited reproducibility. The authors present an alternative approach, featuring a convolutional neural network for multiclass segmentation of kidney tissue in sections stained by periodic acid–Schiff. Their findings demonstrate applicability of convolutional neural networks for tissue from multiple centers, for biopsies and nephrectomy samples, and for the analysis of both healthy and pathologic tissues. In addition, they validated the network’s results with components from the Banff classification system. Their convolutional neural network may have utility for quantitative studies involving kidney histopathology across centers and potential for application in routine diagnostics. Background: The development of deep neural networks is facilitating more advanced digital analysis of histopathologic images. We trained a convolutional neural network for multiclass segmentation of digitized kidney tissue sections stained with periodic acid–Schiff (PAS). Methods: We trained the network using multiclass annotations from 40 whole-slide images of stained kidney transplant biopsies and applied it to four independent data sets. We assessed multiclass segmentation performance by calculating Dice coefficients for ten tissue classes on ten transplant biopsies from the Radboud University Medical Center in Nijmegen, The Netherlands, and on ten transplant biopsies from an external center for validation. We also fully segmented 15 nephrectomy samples and calculated the network’s glomerular detection rates and compared network-based measures with visually scored histologic components (Banff classification) in 82 kidney transplant biopsies. Results: The weighted mean Dice coefficients of all classes were 0.80 and 0.84 in ten kidney transplant biopsies from the Radboud center and the external center, respectively. The best segmented class was “glomeruli” in both data sets (Dice coefficients, 0.95 and 0.94, respectively), followed by “tubuli combined” and “interstitium.” The network detected 92.7% of all glomeruli in nephrectomy samples, with 10.4% false positives. In whole transplant biopsies, the mean intraclass correlation coefficient for glomerular counting performed by pathologists versus the network was 0.94. We found significant correlations between visually scored histologic components and network-based measures. Conclusions: This study presents the first convolutional neural network for multiclass segmentation of PAS-stained nephrectomy samples and transplant biopsies. Our network may have utility for quantitative studies involving kidney histopathology across centers and provide opportunities for deep learning applications in routine diagnostics.

Manual counting of mitotic tumor cells in tissue sections constitutes one of the strongest prognostic markers for breast cancer. This procedure, however, is time-consuming and error-prone. We developed a method to automatically detect mitotic figures in breast cancer tissue sections based on convolutional neural networks (CNNs). Application of CNNs to hematoxylin and eosin (H&E) stained histological tissue sections is hampered by noisy and expensive reference standards established by pathologists, lack of generalization due to staining variation across laboratories, and high computational requirements needed to process gigapixel whole-slide images (WSIs). In this paper, we present a method to train and evaluate CNNs to specifically solve these issues in the context of mitosis detection in breast cancer WSIs. First, by combining image analysis of mitotic activity in phosphohistone-H3 restained slides and registration, we built a reference standard for mitosis detection in entire H&E WSIs requiring minimal manual annotation effort. Second, we designed a data augmentation strategy that creates diverse and realistic H&E stain variations by modifying H&E color channels directly. Using it during training combined with network ensembling resulted in a stain invariant mitosis detector. Third, we applied knowledge distillation to reduce the computational requirements of the mitosis detection ensemble with a negligible loss of performance. The system was trained in a single-center cohort and evaluated in an independent multicenter cohort from the cancer genome atlas on the three tasks of the tumor proliferation assessment challenge. We obtained a performance within the top three best methods for most of the tasks of the challenge.

We propose Neural Image Compression (NIC), a two-step method to build convolutional neural networks for gigapixel image analysis solely using weak image-level labels. First, gigapixel images are compressed using a neural network trained in an unsupervised fashion, retaining high-level information while suppressing pixel-level noise. Second, a convolutional neural network (CNN) is trained on these compressed image representations to predict image-level labels, avoiding the need for fine-grained manual annotations. We compared several encoding strategies, namely reconstruction error minimization, contrastive training and adversarial feature learning, and evaluated NIC on a synthetic task and two public histopathology datasets. We found that NIC can exploit visual cues associated with image-level labels successfully, integrating both global and local visual information. Furthermore, we visualized the regions of the input gigapixel images where the CNN attended to, and confirmed that they overlapped with annotations from human experts.