Dataset Information

Piloting a Deep Learning Model for Predicting Nuclear BAP1 Immunohistochemical Expression of Uveal Melanoma from Hematoxylin-and-Eosin Sections.

ABSTRACT: Background:Uveal melanoma (UM) is the most common primary intraocular malignancy in adults. Monosomy 3 and BAP1 mutation are strong prognostic factors predicting metastatic risk in UM. Nuclear BAP1 (nBAP1) expression is a close immunohistochemical surrogate for both genetic alterations. Not all laboratories perform routine BAP1 immunohistochemistry or genetic testing, and rely mainly on clinical information and anatomic/morphologic analyses for UM prognostication. The purpose of our study was to pilot deep learning (DL) techniques to predict nBAP1 expression on whole slide images (WSIs) of hematoxylin and eosin (H&E) stained UM sections. Methods:One hundred forty H&E-stained UMs were scanned at 40 × magnification, using commercially available WSI image scanners. The training cohort comprised 66 BAP1+ and 74 BAP1- UM, with known chromosome 3 status and clinical outcomes. Nonoverlapping areas of three different dimensions (512 × 512, 1024 × 1024, and 2048 × 2048 pixels) for comparison were extracted from tumor regions in each WSI, and were resized to 256 × 256 pixels. Deep convolutional neural networks (Resnet18 pre-trained on Imagenet) and auto-encoder-decoders (U-Net) were trained to predict nBAP1 expression of these patches. Trained models were tested on the patches cropped from a test cohort of WSIs of 16 BAP1+ and 28 BAP1- UM cases. Results:The trained model with best performance achieved area under the curve values of 0.90 for patches and 0.93 for slides on the test set. Conclusions:Our results show the effectiveness of DL for predicting nBAP1 expression in UM on the basis of H&E sections only. Translational Relevance:Our pilot demonstrates a high capacity of artificial intelligence-related techniques for automated prediction on the basis of histomorphology, and may be translatable into routine histology laboratories.

SUBMITTER: Zhang H

PROVIDER: S-EPMC7476670 | biostudies-literature | 2020 Sep

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Piloting a Deep Learning Model for Predicting Nuclear BAP1 Immunohistochemical Expression of Uveal Melanoma from Hematoxylin-and-Eosin Sections.

Zhang Hongrun H Kalirai Helen H Acha-Sagredo Amelia A Yang Xiaoyun X Zheng Yalin Y Coupland Sarah E SE

Translational vision science & technology 20200901 2

<h4>Background</h4>Uveal melanoma (UM) is the most common primary intraocular malignancy in adults. Monosomy 3 and <i>BAP1</i> mutation are strong prognostic factors predicting metastatic risk in UM. Nuclear BAP1 (nBAP1) expression is a close immunohistochemical surrogate for both genetic alterations. Not all laboratories perform routine BAP1 immunohistochemistry or genetic testing, and rely mainly on clinical information and anatomic/morphologic analyses for UM prognostication. The purpose of o ...[more]

PMID: 32953248

Similar Datasets

Project description:Importance:Strategies to procure high-quality core-needle biopsy (CNB) specimens are critical for making basic tissue diagnoses and for ancillary testing. Objectives:To investigate acquisition of fluorescence confocal microscopy (FCM) images of interventional radiology (IR)-guided CNB in real time in the radiology suite and to compare the accuracy of FCM diagnoses with those of hematoxylin-eosin (H&E)-stained CNB sections. Design, Setting, and Participants:In this diagnostic study, FCM imaging of IR-guided CNBs was performed in the radiology suite at a major cancer center for patients with an imaging abnormality from August 1, 2016, to April 30, 2019. The time taken to acquire FCM images and the quality of FCM images based on percentage of interpretable tissue with optimal resolution was recorded. The FCM images were read by 2 pathologists and categorized as nondiagnostic, benign/atypical, or suspicious/malignant; these diagnoses were compared with those made using H&E-stained tissue sections. Cases with discrepant diagnosis were reassessed by the pathologists together for a consensus diagnosis. Data were analyzed from June 3 to July 19, 2019. Interventions:Each IR-guided CNB was stained with 0.6mM acridine orange, subjected to FCM imaging, and then processed to generate H&E-stained sections. Main Outcomes and Measures:Mean time taken for acquisition of FCM images, quality of FCM images based on interpretable percentage of the image, and accuracy of diagnostic categorization based on FCM images compared with H&E-stained sections. Results:A total of 105 patients (57 male [54.3%]; mean [SD] age, 63 [13] years) underwent IR-guided CNBs in a mean (SD) of 7 (2) minutes each. The FCM images showed at least 20% of the tissue with optimal quality in 101 CNB specimens (96.2%). The FCM images were accurately interpreted by the 2 pathologists in 100 of 105 cases (95.2%) (2 false-positive and 3 false-negative) and 90 of 105 cases (85.7%) (6 false-positive and 9 false-negative). A reassessment of 14 discordant diagnoses resulted in consensus diagnoses that were accurate in 101 of 105 cases (96.2%) (1 false-positive and 3 false-negative). Conclusions and Relevance:The ease of acquisition of FCM images of acceptable quality and the high accuracy of the diagnoses suggest that FCM may be useful for rapid evaluation of IR-guided CNBs. This approach warrants further investigation.

Project description:We introduce a machine learning-based analysis to predict the immunohistochemical (IHC) labeling index for the cell proliferation marker Ki67/MIB1 on cancer tissues based on morphometrical features extracted from hematoxylin and eosin (H&E)-stained formalin-fixed, paraffin-embedded (FFPE) tumor tissue samples. We provided a proof-of-concept prediction of the Ki67/MIB1 IHC positivity of cancer cells through the definition and quantitation of single nuclear features. In the first instance, we set our digital framework on Ki67/MIB1-stained OSCC (oral squamous cell carcinoma) tissue sample whole slide images, using QuPath as a working platform and its integrated algorithms, and we built a classifier in order to distinguish tumor and stroma classes and, within them, Ki67-positive and Ki67-negative cells; then, we sorted the morphometric features of tumor cells related to their Ki67 IHC status. Among the evaluated features, nuclear hematoxylin mean optical density (NHMOD) presented as the best one to distinguish Ki67/MIB1 positive from negative cells. We confirmed our findings in a single-cell level analysis of H&E staining on Ki67-immunostained/H&E-decolored tissue samples. Finally, we tested our digital framework on a case series of oral squamous cell carcinomas (OSCC), arranged in tissue microarrays; we selected two consecutive sections of each OSCC FFPE TMA (tissue microarray) block, respectively stained with H&E and immuno-stained for Ki67/MIB1. We automatically detected tumor cells in H&E slides and generated a "false color map" (FCM) based on NHMOD through the QuPath measurements map tool. FCM nearly coincided with the actual immunohistochemical result, allowing the prediction of Ki67/MIB1 positive cells in a direct visual fashion. Our proposed approach provides the pathologist with a fast method of identifying the proliferating compartment of the tumor through a quantitative assessment of the nuclear features on H&E slides, readily appreciable by visual inspection. Although this technique needs to be fine-tuned and tested on larger series of tumors, the digital analysis approach appears to be a promising tool to quickly forecast the tumor's proliferation fraction directly on routinely H&E-stained digital sections.

Project description:BACKGROUND:BAP1 is a histone deubiquitinase that acts as a tumor and metastasis suppressor associated with disease progression in human cancer. We have used the "Calling Card System" of transposase-directed transposon insertion mapping to identify the genomic targets of BAP1 in uveal melanoma (UM). This system was developed to identify the genomic loci visited by transcription factors that bind directly to DNA; our study is the first use of the system with a chromatin-remodeling factor that binds to histones but does not interact directly with DNA. METHODS:The transposase piggyBac (PBase) was fused to BAP1 and expressed in OCM-1A UM cells. The insertion of transposons near BAP1 binding sites in UM cells were identified by genomic sequencing. We also examined RNA expression in the same OCM-1A UM cells after BAP1 depletion to identify BAP1 binding sites associated with BAP1-responsive genes. Sets of significant genes were analyzed for common pathways, transcription factor binding sites, and ability to identify molecular tumor classes. RESULTS:We found a strong correlation between multiple calling-card transposon insertions targeted by BAP1-PBase and BAP1-responsive expression of adjacent genes. BAP1-bound genomic loci showed narrow distributions of insertions and were near transcription start sites, consistent with recruitment of BAP1 to these sites by specific DNA-binding proteins. Sequence consensus analysis of BAP1-bound sites showed enrichment of motifs specific for YY1, NRF1 and Ets transcription factors, which have been shown to interact with BAP1 in other cell types. Further, a subset of the BAP1 genomic target genes was able to discriminate aggressive tumors in published gene expression data from primary UM tumors. CONCLUSIONS:The calling card methodology works equally well for chromatin regulatory factors that do not interact directly with DNA as for transcription factors. This technique has generated a new and expanded list of BAP1 targets in UM that provides important insight into metastasis pathways and identifies novel potential therapeutic targets.

Dataset Information

Piloting a Deep Learning Model for Predicting Nuclear BAP1 Immunohistochemical Expression of Uveal Melanoma from Hematoxylin-and-Eosin Sections.

Publications

Piloting a Deep Learning Model for Predicting Nuclear BAP1 Immunohistochemical Expression of Uveal Melanoma from Hematoxylin-and-Eosin Sections.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets