PatchSorter: a high throughput deep learning digital pathology tool for object labeling.

The discovery of patterns associated with diagnosis, prognosis, and therapy response in digital pathology images often requires intractable labeling of large quantities of histological objects. Here we release an open-source labeling tool, PatchSorter, which integrates deep learning with an intuitive web interface. Using >100,000 objects, we demonstrate a >7x improvement in labels per second over unaided labeling, with minimal impact on labeling accuracy, thus enabling high-throughput labeling of large datasets.

Morphologic Features of Invasion in Lung Adenocarcinoma: Diagnostic Pitfalls.

Reproducibility of pulmonary invasive adenocarcinoma diagnosis is poor when applying the World Health Organization (WHO) classification. In this article, we aimed first to explain by 3-dimensional morphology why simple pattern recognition induces pitfalls for the assessment of invasion as applied in the current WHO classification of pulmonary adenocarcinomas. The underlying iatrogenic-induced morphologic alterations in collapsed adenocarcinoma in situ overlap with criteria for invasive adenocarcinoma. Pitfalls in seemingly acinar and papillary carcinoma are addressed with additional cytokeratin 7 and elastin stains. In addition, we provide more stringent criteria for a better reproducible and likely generalizable classification.

Pathology of Surgically Resected Lung Cancers Following Neoadjuvant Therapy.

In around 30% of patients, non-small cell lung cancer is diagnosed at an advanced but resectable stage. Adding systemic therapy has shown clear benefit over surgery alone in locally advanced disease, and currently, chemo-immunotherapy in the adjuvant or neoadjuvant setting is the new standard for patients without targetable mutations. One major advantage of the neoadjuvant approach is the possibility of an immediate evaluation of the treatment effect, highlighting the role of pathology as an important contributor at the forefront of clinical decision-making and research. This review provides a summary and an update on current guidelines for histological evaluation of treatment effect after neoadjuvant therapy, also known as regression grading, and discusses newer data focusing on areas of evolving questions and controversies, such as the gross examination of the tumor and tumor bed, weighted versus unweighted evaluation approaches, discussion of histologic tumor type-specific cut-offs for major pathologic response, assessment of lymph nodes and regression grading after immunotherapy and targeted therapy. As no data or recommendations exist on regression grading of multiple tumor nodules, a practical approach is recommended. Lastly, we will touch on additional tissue biomarkers and summarize recent advances in the ardently discussed field of using circulating tumor DNA for the evaluation of treatment response.

Genomic Staging of Multifocal Lung Squamous Cell Carcinomas Is Independent of the Comprehensive Morphologic Assessment.

INTRODUCTION: Morphologic and molecular data for staging of multifocal lung squamous cell carcinomas (LSCCs) are limited. In this study, whole exome sequencing (WES) was used as the gold standard to determine whether multifocal LSCC represented separate primary lung cancers (SPLCs) or intrapulmonary metastases (IPMs). Genomic profiles were compared with the comprehensive morphologic assessment. METHODS: WES was performed on 20 tumor pairs of multifocal LSCC and matched normal lymph nodes using the Illumina NovaSeq6000 S4-Xp (Illumina, San Diego, CA). WES clonal and subclonal analysis data were compared with histologic assessment by 16 thoracic pathologists. In addition, the immune gene profiling of the study cases was characterized by the HTG EdgeSeq Precision Immuno-Oncology Panel. RESULTS: By WES data, 11 cases were classified as SPLC and seven cases as IPM. Two cases were technically suboptimal. Analysis revealed marked genomic and immunogenic heterogeneity, but immune gene expression profiles highly correlated with mutation profiles. Tumors classified as IPM have a large number of shared mutations (ranging from 33.5% to 80.7%). The agreement between individual morphologic assessments for each case and WES was 58.3%. One case was unanimously interpreted morphologically as IPM and was in agreement with WES. In a further 17 cases, the number of pathologists whose morphologic interpretation was in agreement with WES ranged from two (one case) to 15 pathologists (one case) per case. Pathologists showed a fair interobserver agreement in the morphologic staging of multiple LSCCs, with an overall kappa of 0.232. CONCLUSIONS: Staging of multifocal LSCC based on morphologic assessment is unreliable. Comprehensive genomic analyses should be adopted for the staging of multifocal LSCC.

Impact of template-based synoptic reporting on completeness of surgical pathology reports.

Synoptic reporting increases completeness and standardization of surgical pathology reports and thereby contributes to an increased quality of clinical cancer care. Nevertheless, its widespread practical implementation remains a challenge, which is in part related to the effort required for setup and maintenance of database structures. This prompted us to assess the effect of a simple template-based, database-free system for synoptic reporting on completeness of surgical pathology reports. For this purpose, we analyzed 200 synoptic reports (100 colon and 100 lung cancer resections each) for completeness as required by the pertinent College of American Pathologists (CAP) protocols and compared these to a control dataset of 200 narrative reports. Introduction of template-based synoptic reporting resulted in improved completeness (98% of mandatory data elements) as compared to narrative reports (77%). Narrative reports showed a high degree of completeness for data elements covered by previously existing dictation templates. In conclusion, template-based synoptic reporting without underlying database structure can be a useful transitory phase in the implementation of synoptic reporting. It can result in a similar degree of completeness as reported in the literature for database solutions and provides other benefits of synoptic reporting while facilitating its implementation.

Current challenges and practical aspects of molecular pathology for non-small cell lung cancers.

The continuing evolution of treatment options in thoracic oncology requires the pathologist to regularly update diagnostic algorithms for management of tumor samples. It is essential to decide on the best way to use tissue biopsies, cytological samples, as well as liquid biopsies to identify the different mandatory predictive biomarkers of lung cancers in a short turnaround time. However, biological resources and laboratory member workforce are limited and may be not sufficient for the increased complexity of molecular pathological analyses and for complementary translational research development. In this context, the surgical pathologist is the only one who makes the decisions whether or not to send specimens to immunohistochemical and molecular pathology platforms. Moreover, the pathologist can rapidly contact the oncologist to obtain a new tissue biopsy and/or a liquid biopsy if he/she considers that the biological material is not sufficient in quantity or quality for assessment of predictive biomarkers. Inadequate control of algorithms and sampling workflow may lead to false negative, inconclusive, and incomplete findings, resulting in inappropriate choice of therapeutic strategy and potentially poor outcome for patients. International guidelines for lung cancer treatment are based on the results of the expression of different proteins and on genomic alterations. These guidelines have been established taking into consideration the best practices to be set up in clinical and molecular pathology laboratories. This review addresses the current predictive biomarkers and algorithms for use in thoracic oncology molecular pathology as well as the central role of the pathologist, notably in the molecular tumor board and her/his participation in the treatment decision-making. The perspectives in this setting will be discussed.

Pulmonary squamous cell carcinoma and lymphoepithelial carcinoma - morphology, molecular characteristics and differential diagnosis.

Squamous cell carcinoma (SCC) comprises one of the major groups of non-small-cell carcinoma of the lung, and is subtyped into keratinising, non-keratinising and basaloid SCC. SCC can readily be diagnosed using histomorphology alone in keratinising SCC. Confirmatory immunohistochemical analyses should always be applied in non-keratinising and basaloid tumours to exclude differential diagnoses, most prominently adenocarcinoma and high-grade neuroendocrine carcinoma, which may have important therapeutic consequences. According to the World Health Organisation (WHO) classification 2015, the diagnosis of SCC can be rendered in resections of morphologically ambiguous tumours with squamous immunophenotype. In biopsies and cytology preparations in the same setting the current guidelines propose a diagnosis of 'non-small-cell carcinoma, favour SCC' in TTF1-negative and p40-positive tumours to acknowledge a possible sampling bias and restrict extended immunohistochemical evaluation in order to preserve tissue for molecular testing. Most SCC feature a molecular 'tobacco-smoke signature' with enrichment in GG > TT mutations, in line with the strong epidemiological association of SCC with smoking. Targetable mutations are extremely rare but they do occur, in particular in younger and non- or light-smoking patients, warranting molecular investigations. Lymphoepithelial carcinoma (LEC) is a poorly differentiated SCC with a syncytial growth pattern and a usually prominent lymphoplasmacytic infiltrate and frequent Epstein-Barr virus (EBV) association. In this review, we describe the morphological and molecular characteristics of SCC and LEC and discuss the most pertinent differential diagnoses.

Swiss digital pathology recommendations: results from a Delphi process conducted by the Swiss Digital Pathology Consortium of the Swiss Society of Pathology.

Integration of digital pathology (DP) into clinical diagnostic workflows is increasingly receiving attention as new hardware and software become available. To facilitate the adoption of DP, the Swiss Digital Pathology Consortium (SDiPath) organized a Delphi process to produce a series of recommendations for DP integration within Swiss clinical environments. This process saw the creation of 4 working groups, focusing on the various components of a DP system (1) scanners, quality assurance and validation of scans, (2) integration of Whole Slide Image (WSI)-scanners and DP systems into the Pathology Laboratory Information System, (3) digital workflow-compliance with general quality guidelines, and (4) image analysis (IA)/artificial intelligence (AI), with topic experts for each recruited for discussion and statement generation. The work product of the Delphi process is 83 consensus statements presented here, forming the basis for "SDiPath Recommendations for Digital Pathology". They represent an up-to-date resource for national and international hospitals, researchers, device manufacturers, algorithm developers, and all supporting fields, with the intent of providing expectations and best practices to help ensure safe and efficient DP usage.

Spatial metabolomics identifies distinct tumor-specific and stroma-specific subtypes in patients with lung squamous cell carcinoma.

Molecular subtyping of lung squamous cell carcinoma (LUSC) has been performed at the genomic, transcriptomic, and proteomic level. However, LUSC stratification based on tissue metabolomics is still lacking. Combining high-mass-resolution imaging mass spectrometry with consensus clustering, four tumor- and four stroma-specific subtypes with distinct metabolite patterns were identified in 330 LUSC patients. The first tumor subtype T1 negatively correlated with DNA damage and immunological features including CD3, CD8, and PD-L1. The same features positively correlated with the tumor subtype T2. Tumor subtype T4 was associated with high PD-L1 expression. Compared with the status of subtypes T1 and T4, patients with subtype T3 had improved prognosis, and T3 was an independent prognostic factor with regard to UICC stage. Similarly, stroma subtypes were linked to distinct immunological features and metabolic pathways. Stroma subtype S4 had a better prognosis than S2. Subsequently, analyses based on an independent LUSC cohort treated by neoadjuvant therapy revealed that the S2 stroma subtype was associated with chemotherapy resistance. Clinically relevant patient subtypes as determined by tissue-based spatial metabolomics are a valuable addition to existing molecular classification systems. Metabolic differences among the subtypes and their associations with immunological features may contribute to the improvement of personalized therapy.

Digital image analysis and artificial intelligence in pathology diagnostics-the Swiss view.

Digital pathology (DP) is increasingly entering routine clinical pathology diagnostics. As digitization of the routine caseload advances, implementation of digital image analysis algorithms and artificial intelligence tools becomes not only attainable, but also desirable in daily sign out. The Swiss Digital Pathology Consortium (SDiPath) has initiated a Delphi process to generate best-practice recommendations for various phases of the process of digitization in pathology for the local Swiss environment, encompassing the following four topics: i) scanners, quality assurance, and validation of scans; ii) integration of scanners and systems into the pathology laboratory information system; iii) the digital workflow; and iv) digital image analysis (DIA)/artificial intelligence (AI). The current article focuses on the DIA-/AI-related recommendations generated and agreed upon by the working group and further verified by the Delphi process among the members of SDiPath. Importantly, they include the view and the currently perceived needs of practicing pathologists from multiple academic and cantonal hospitals as well as private practices.

International Association for the Study of Lung Cancer Study of Reproducibility in Assessment of Pathologic Response in Resected Lung Cancers After Neoadjuvant Therapy.

INTRODUCTION: Pathologic response has been proposed as an early clinical trial end point of survival after neoadjuvant treatment in clinical trials of NSCLC. The International Association for the Study of Lung Cancer (IASLC) published recommendations for pathologic evaluation of resected lung cancers after neoadjuvant therapy. The aim of this study was to assess pathologic response interobserver reproducibility using IASLC criteria. METHODS: An international panel of 11 pulmonary pathologists reviewed hematoxylin and eosin-stained slides from the lung tumors of resected NSCLC from 84 patients who received neoadjuvant immune checkpoint inhibitors in six clinical trials. Pathologic response was assessed for percent viable tumor, necrosis, and stroma. For each slide, tumor bed area was measured microscopically, and pre-embedded formulas calculated unweighted and weighted major pathologic response (MPR) averages to reflect variable tumor bed proportion. RESULTS: Unanimous agreement among pathologists for MPR was observed in 68 patients (81%), and inter-rater agreement (IRA) was 0.84 (95% confidence interval [CI]: 0.76-0.92) and 0.86 (95% CI: 0.79-0.93) for unweighted and weighted averages, respectively. Overall, unweighted and weighted methods did not reveal significant differences in the classification of MPR. The highest concordance by both methods was observed for cases with more than 95% viable tumor (IRA = 0.98, 95% CI: 0.96-1) and 0% viable tumor (IRA = 0.94, 95% CI: 0.89-0.98). The most common reasons for discrepancies included interpretations of tumor bed, presence of prominent stromal inflammation, distinction between reactive and neoplastic pneumocytes, and assessment of invasive mucinous adenocarcinoma. CONCLUSIONS: Our study revealed excellent reliability in cases with no residual viable tumor and good reliability for MPR with the IASLC recommended less than or equal to 10% cutoff for viable tumor after neoadjuvant therapy.

Pathologist Computer-Aided Diagnostic Scoring of Tumor Cell Fraction: A Swiss National Study.

Tumor cell fraction (TCF) estimation is a common clinical task with well-established large interobserver variability. It thus provides an ideal test bed to evaluate potential impacts of employing a tumor cell fraction computer-aided diagnostic (TCFCAD) tool to support pathologists' evaluation. During a National Slide Seminar event, pathologists (n = 69) were asked to visually estimate TCF in 10 regions of interest (ROIs) from hematoxylin and eosin colorectal cancer images intentionally curated for diverse tissue compositions, cellularity, and stain intensities. Next, they re-evaluated the same ROIs while being provided a TCFCAD-created overlay highlighting predicted tumor vs nontumor cells, together with the corresponding TCF percentage. Participants also reported confidence levels in their assessments using a 5-tier scale, indicating no confidence to high confidence, respectively. The TCF ground truth (GT) was defined by manual cell-counting by experts. When assisted, interobserver variability significantly decreased, showing estimates converging to the GT. This improvement remained even when TCFCAD predictions deviated slightly from the GT. The standard deviation (SD) of the estimated TCF to the GT across ROIs was 9.9% vs 5.8% with TCFCAD (P < .0001). The intraclass correlation coefficient increased from 0.8 to 0.93 (95% CI, 0.65-0.93 vs 0.86-0.98), and pathologists stated feeling more confident when aided (3.67 ± 0.81 vs 4.17 ± 0.82 with the computer-aided diagnostic [CAD] tool). TCFCAD estimation support demonstrated improved scoring accuracy, interpathologist agreement, and scoring confidence. Interestingly, pathologists also expressed more willingness to use such a CAD tool at the end of the survey, highlighting the importance of training/education to increase adoption of CAD systems.

Patterns of Carbon-Bound Exogenous Compounds Impact Disease Pathophysiology in Lung Cancer Subtypes in Different Ways.

Carbon-bound exogenous compounds, such as polycyclic aromatic hydrocarbons (PAHs), tobacco-specific nitrosamines, aromatic amines, and organohalogens, are known to affect both tumor characteristics and patient outcomes in lung squamous cell carcinoma (LUSC); however, the roles of these compounds in lung adenocarcinoma (LUAD) remain unclear. We analyzed 11 carbon-bound exogenous compounds in LUAD and LUSC samples using in situ high mass-resolution matrix-assisted laser desorption/ionization Fourier-transform ion cyclotron resonance mass spectrometry imaging and performed a cluster analysis to compare the patterns of carbon-bound exogenous compounds between these two lung cancer subtypes. Correlation analyses were conducted to investigate associations among exogenous compounds, endogenous metabolites, and clinical data, including patient survival outcomes and smoking behaviors. Additionally, we examined differences in exogenous compound patterns between normal and tumor tissues. Our analyses revealed that PAHs, aromatic amines, and organohalogens were more abundant in LUAD than in LUSC, whereas the tobacco-specific nitrosamine nicotine-derived nitrosamine ketone was more abundant in LUSC. Patients with LUAD and LUSC could be separated according to carbon-bound exogenous compound patterns detected in the tumor compartment. The same compounds had differential impacts on patient outcomes, depending on the cancer subtype. Correlation and network analyses indicated substantial differences between LUAD and LUSC metabolomes, associated with substantial differences in the patterns of the carbon-bound exogenous compounds. These data suggest that the contributions of these carcinogenic compounds to cancer biology may differ according to the cancer subtypes.

Standardized Classification of Lung Adenocarcinoma Subtypes and Improvement of Grading Assessment Through Deep Learning.

The histopathologic distinction of lung adenocarcinoma (LADC) subtypes is subject to high interobserver variability, which can compromise the optimal assessment of patient prognosis. Therefore, this study developed convolutional neural networks capable of distinguishing LADC subtypes and predicting disease-specific survival, according to the recently established LADC tumor grades. Consensus LADC histopathologic images were obtained from 17 expert pulmonary pathologists and one pathologist in training. Two deep learning models (AI-1 and AI-2) were trained to predict eight different LADC classes. Furthermore, the trained models were tested on an independent cohort of 133 patients. The models achieved high precision, recall, and F1 scores exceeding 0.90 for most of the LADC classes. Clear stratification of the three LADC grades was reached in predicting the disease-specific survival by the two models, with both Kaplan-Meier curves showing significance (P = 0.0017 and 0.0003). Moreover, both trained models showed high stability in the segmentation of each pair of predicted grades with low variation in the hazard ratio across 200 bootstrapped samples. These findings indicate that the trained convolutional neural networks improve the diagnostic accuracy of the pathologist and refine LADC grade assessment. Thus, the trained models are promising tools that may assist in the routine evaluation of LADC subtypes and grades in clinical practice.

Deep computational image analysis of immune cell niches reveals treatment-specific outcome associations in lung cancer.

The tumor immune composition influences prognosis and treatment sensitivity in lung cancer. The presence of effective adaptive immune responses is associated with increased clinical benefit after immune checkpoint blockers. Conversely, immunotherapy resistance can occur as a consequence of local T-cell exhaustion/dysfunction and upregulation of immunosuppressive signals and regulatory cells. Consequently, merely measuring the amount of tumor-infiltrating lymphocytes (TILs) may not accurately reflect the complexity of tumor-immune interactions and T-cell functional states and may not be valuable as a treatment-specific biomarker. In this work, we investigate an immune-related biomarker (PhenoTIL) and its value in associating with treatment-specific outcomes in non-small cell lung cancer (NSCLC). PhenoTIL is a novel computational pathology approach that uses machine learning to capture spatial interplay and infer functional features of immune cell niches associated with tumor rejection and patient outcomes. PhenoTIL's advantage is the computational characterization of the tumor immune microenvironment extracted from H&E-stained preparations. Association with clinical outcome and major non-small cell lung cancer (NSCLC) histology variants was studied in baseline tumor specimens from 1,774 lung cancer patients treated with immunotherapy and/or chemotherapy, including the clinical trial Checkmate 057 (NCT01673867).

MEK1 drives oncogenic signaling and interacts with PARP1 for genomic and metabolic homeostasis in malignant pleural mesothelioma.

Malignant pleural mesothelioma (MPM) is a lethal malignancy etiologically caused by asbestos exposure, for which there are few effective treatment options. Although asbestos carcinogenesis is associated with reactive oxygen species (ROS), the bona fide oncogenic signaling pathways that regulate ROS homeostasis and bypass ROS-evoked apoptosis in MPM are poorly understood. In this study, we demonstrate that the mitogen-activated protein kinase (MAPK) pathway RAS-RAF-MEK-ERK is hyperactive and a molecular driver of MPM, independent of histological subtypes and genetic heterogeneity. Suppression of MAPK signaling by clinically approved MEK inhibitors (MEKi) elicits PARP1 to protect MPM cells from the cytotoxic effects of MAPK pathway blockage. Mechanistically, MEKi induces impairment of homologous recombination (HR) repair proficiency and mitochondrial metabolic activity, which is counterbalanced by pleiotropic PARP1. Consequently, the combination of MEK with PARP inhibitors enhances apoptotic cell death in vitro and in vivo that occurs through coordinated upregulation of cytotoxic ROS in MPM cells, suggesting a mechanism-based, readily translatable strategy to treat this daunting disease. Collectively, our studies uncover a previously unrecognized scenario that hyperactivation of the MAPK pathway is an essential feature of MPM and provide unprecedented evidence that MAPK signaling cooperates with PARP1 to homeostatically maintain ROS levels and escape ROS-mediated apoptosis.

Immunogenomic analysis of human brain metastases reveals diverse immune landscapes across genetically distinct tumors.

Brain metastases (BrMs) are the most common form of brain tumors in adults and frequently originate from lung and breast primary cancers. BrMs are associated with high mortality, emphasizing the need for more effective therapies. Genetic profiling of primary tumors is increasingly used as part of the effort to guide targeted therapies against BrMs, and immune-based strategies for the treatment of metastatic cancer are gaining momentum. However, the tumor immune microenvironment (TIME) of BrM is extremely heterogeneous, and whether specific genetic profiles are associated with distinct immune states remains unknown. Here, we perform an extensive characterization of the immunogenomic landscape of human BrMs by combining whole-exome/whole-genome sequencing, RNA sequencing of immune cell populations, flow cytometry, immunofluorescence staining, and tissue imaging analyses. This revealed unique TIME phenotypes in genetically distinct lung- and breast-BrMs, thereby enabling the development of personalized immunotherapies tailored by the genetic makeup of the tumors.

Immunohistochemical Detection of the Autophagy Markers LC3 and p62/SQSTM1 in Formalin-Fixed and Paraffin-Embedded Tissue.

Autophagy is a highly conserved cellular mechanism of "self-digestion," ensuring cellular homeostasis and playing a role in many diseases including cancer. As a stress response mechanism, it may also be involved in cellular response to therapy. LC3 and Sequestosome 1 (p62/SQSTM1) are among the most widely used markers to monitor autophagy and can be visualized in formalin-fixed and paraffin-embedded tissue by immunohistochemistry. Here we describe a validated staining protocol using an automated staining system available in many routine pathology laboratories, enabling high-throughput staining under standardized conditions.

Immunohistochemical Detection of the Chaperone-Mediated Autophagy Markers LAMP2A and HSPA8 in Formalin-Fixed and Paraffin-Embedded Tissues.

Autophagy is crucial for maintaining cellular homeostasis and its deregulation is involved in disease development, including cancer. The key players of chaperone-mediated autophagy (CMA), a particular selective subtype of autophagy, are HSPA8 and LAMP2A. Both proteins can be immunohistochemically detected in formalin-fixed paraffin-embedded (FFPE) tissue. LAMP2A is frequently overexpressed in a variety of cancers where it likely supports cancer cell survival and resistance to anti-cancer therapies in a context-dependent manner. Here we present the immunohistochemical staining protocol of antibodies against LAMP2A and HSPA8, using an automated staining system, suitable for routine diagnostics. Additionally, we also suggest a staining evaluation method.

Overcoming the Interobserver Variability in Lung Adenocarcinoma Subtyping: A Clustering Approach to Establish a Ground Truth for Downstream Applications.

CONTEXT.­: The accurate identification of different lung adenocarcinoma histologic subtypes is important for determining prognosis but can be challenging because of overlaps in the diagnostic features, leading to considerable interobserver variability. OBJECTIVE.­: To provide an overview of the diagnostic agreement for lung adenocarcinoma subtypes among pathologists and to create a ground truth using the clustering approach for downstream computational applications. DESIGN.­: Three sets of lung adenocarcinoma histologic images with different evaluation levels (small patches, areas with relatively uniform histology, and whole slide images) were reviewed by 17 international expert lung pathologists and 1 pathologist in training. Each image was classified into one or several lung adenocarcinoma subtypes. RESULTS.­: Among the 4702 patches of the first set, 1742 (37%) had an overall consensus among all pathologists. The overall Fleiss κ score for the agreement of all subtypes was 0.58. Using cluster analysis, pathologists were hierarchically grouped into 2 clusters, with κ scores of 0.588 and 0.563 in clusters 1 and 2, respectively. Similar results were obtained for the second and third sets, with fair-to-moderate agreements. Patches from the first 2 sets that obtained the consensus of the 18 pathologists were retrieved to form consensus patches and were regarded as the ground truth of lung adenocarcinoma subtypes. CONCLUSIONS.­: Our observations highlight discrepancies among experts when assessing lung adenocarcinoma subtypes. However, a subsequent number of consensus patches could be retrieved from each cluster, which can be used as ground truth for the downstream computational pathology applications, with minimal influence from interobserver variability.