Ruchika Verma

Nuclear segmentation in digital microscopic tissue images can enable extraction of high-quality features for nuclear morphometrics and other analysis in computational pathology. Conventional image processing techniques, such as Otsu thresholding and watershed segmentation, do not work effectively on challenging cases, such as chromatin-sparse and crowded nuclei. In contrast, machine learning-based segmentation can generalize across various nuclear appearances. However, training machine learning algorithms requires data sets of images, in which a vast number of nuclei have been annotated. Publicly accessible and annotated data sets, along with widely agreed upon metrics to compare techniques, have catalyzed tremendous innovation and progress on other image classification problems, particularly in object recognition. Inspired by their success, we introduce a large publicly accessible data set of hematoxylin and eosin (H&E)-stained tissue images with more than 21000 painstakingly annotated nuclear boundaries, whose quality was validated by a medical doctor. Because our data set is taken from multiple hospitals and includes a diversity of nuclear appearances from several patients, disease states, and organs, techniques trained on it are likely to generalize well and work right out-of-the-box on other H&E-stained images. We also propose a new metric to evaluate nuclear segmentation results that penalizes object- and pixel-level errors in a unified manner, unlike previous metrics that penalize only one type of error. We also propose a segmentation technique based on deep learning that lays a special emphasis on identifying the nuclear boundaries, including those between the touching or overlapping nuclei, and works well on a diverse set of test images.

Generalized nucleus segmentation techniques can contribute greatly to reducing the time to develop and validate visual biomarkers for new digital pathology datasets. We summarize the results of MoNuSeg 2018 Challenge whose objective was to develop generalizable nuclei segmentation techniques in digital pathology. The challenge was an official satellite event of the MICCAI 2018 conference in which 32 teams with more than 80 participants from geographically diverse institutes participated. Contestants were given a training set with 30 images from seven organs with annotations of 21,623 individual nuclei. A test dataset with 14 images taken from seven organs, including two organs that did not appear in the training set was released without annotations. Entries were evaluated based on average aggregated Jaccard index (AJI) on the test set to prioritize accurate instance segmentation as opposed to mere semantic segmentation. More than half the teams that completed the challenge outperformed a previous baseline. Among the trends observed that contributed to increased accuracy were the use of color normalization as well as heavy data augmentation. Additionally, fully convolutional networks inspired by variants of U-Net, FCN, and Mask-RCNN were popularly used, typically based on ResNet or VGG base architectures. Watershed segmentation on predicted semantic segmentation maps was a popular post-processing strategy. Several of the top techniques compared favorably to an individual human annotator and can be used with confidence for nuclear morphometrics.

Abstract Purpose: To (i) create a survival risk score using radiomic features from the tumor habitat on routine MRI to predict progression-free survival (PFS) in glioblastoma and (ii) obtain a biological basis for these prognostic radiomic features, by studying their radiogenomic associations with molecular signaling pathways. Experimental Design: Two hundred three patients with pretreatment Gd-T1w, T2w, T2w-FLAIR MRI were obtained from 3 cohorts: The Cancer Imaging Archive (TCIA; n = 130), Ivy GAP (n = 32), and Cleveland Clinic (n = 41). Gene-expression profiles of corresponding patients were obtained for TCIA cohort. For every study, following expert segmentation of tumor subcompartments (necrotic core, enhancing tumor, peritumoral edema), 936 3D radiomic features were extracted from each subcompartment across all MRI protocols. Using Cox regression model, radiomic risk score (RRS) was developed for every protocol to predict PFS on the training cohort (n = 130) and evaluated on the holdout cohort (n = 73). Further, Gene Ontology and single-sample gene set enrichment analysis were used to identify specific molecular signaling pathway networks associated with RRS features. Results: Twenty-five radiomic features from the tumor habitat yielded the RRS. A combination of RRS with clinical (age and gender) and molecular features (MGMT and IDH status) resulted in a concordance index of 0.81 (P < 0.0001) on training and 0.84 (P = 0.03) on the test set. Radiogenomic analysis revealed associations of RRS features with signaling pathways for cell differentiation, cell adhesion, and angiogenesis, which contribute to chemoresistance in GBM. Conclusions: Our findings suggest that prognostic radiomic features from routine Gd-T1w MRI may also be significantly associated with key biological processes that affect response to chemotherapy in GBM.

Purpose To identify radiomic features extracted from the tumor habitat on routine MR images that are prognostic for progression-free survival (PFS) and to assess their morphologic basis with corresponding histopathologic attributes in glioblastoma (GBM). Materials and Methods In this retrospective study, 156 pretreatment GBM MR images (gadolinium-enhanced T1-weighted, T2-weighted, and fluid-attenuated inversion recovery [FLAIR] images) were curated. Of these 156 images, 122 were used for training (90 from The Cancer Imaging Archive and 32 from the Cleveland Clinic, acquired between December 1, 2011, and May 1, 2018) and 34 were used for validation. The validation set was obtained from the Ivy Glioblastoma Atlas Project database, for which the percentage extent of 11 histologic attributes was available on corresponding histopathologic specimens of the resected tumor. Following expert annotations of the tumor habitat (necrotic core, enhancing tumor, and FLAIR-hyperintense subcompartments), 1008 radiomic descriptors (eg, Haralick texture features, Laws energy features, co-occurrence of local anisotropic gradient orientations [CoLIAGe]) were extracted from the three MRI sequences. The top radiomic features were obtained from each subcompartment in the training set on the basis of their ability to risk-stratify patients according to PFS. These features were then concatenated to create a radiomics risk score (RRS). The RRS was independently validated on a holdout set. In addition, correlations (P < .05) of RRS features were computed, with the percentage extent of the 11 histopathologic attributes, using Spearman correlation analysis. Results RRS yielded a concordance index of 0.80 on the validation set and constituted radiomic features, including Laws (capture edges, waves, ripple patterns) and CoLIAGe (capture disease heterogeneity) from enhancing tumor and FLAIR hyperintensity. These radiomic features were correlated with histopathologic attributes associated with disease aggressiveness in GBM, particularly tumor infiltration (P = .0044) and hyperplastic blood vessels (P = .0005). Conclusion Preliminary findings demonstrated significant associations of prognostic radiomic features with disease-specific histologic attributes, with implications for risk-stratifying patients with GBM for personalized treatment decisions. Supplemental material is available for this article. Keywords: Brain/Brain Stem, Prognosis, Radiomics © RSNA, 2020

The use of self-supervised learning to train pathology foundation models has increased substantially in the past few years. Notably, several models trained on large quantities of clinical data have been made publicly available in recent months. This will significantly enhance scientific research in computational pathology and help bridge the gap between research and clinical deployment. With the increase in availability of public foundation models of different sizes, trained using different algorithms on different datasets, it becomes important to establish a benchmark to compare the performance of such models on a variety of clinically relevant tasks spanning multiple organs and diseases. In this work, we present a collection of pathology datasets comprising clinical slides associated with clinically relevant endpoints including cancer diagnoses and a variety of biomarkers generated during standard hospital operation from three medical centers. We leverage these datasets to systematically assess the performance of public pathology foundation models and provide insights into best practices for training foundation models and selecting appropriate pretrained models. To enable the community to evaluate their models on our clinical datasets, we make available an automated benchmarking pipeline for external use. Self-supervised learning (SSL) is increasingly used to train pathology foundation models. Here, the authors introduce a pathology benchmark set generated during standard clinical workflows that includes multiple cancer and disease types; then leverage it to assess the performance of multiple public SSL pathology foundation models and to provide best practices for model training and selection.