Understanding diagnostic variability in breast pathology: lessons learned from an expert consensus review panel

Multinational, Multicenter Evaluation of Prostate Cancer Tissue in Sub-Saharan Africa: Challenges and Opportunities

PURPOSE

Prostate cancer disproportionately affects men of African descent, yet their representation in tissue-based studies is limited. This multinational, multicenter pilot study aims to establish the groundwork for collaborative research on prostate cancer in sub-Saharan Africa.

METHODS

The Men of African Descent and Carcinoma of the Prostate network formed a pathologist working group representing eight institutions in five African countries. Formalin-fixed paraffin-embedded prostate tissue specimens were collected from Senegal, Nigeria, and Ghana. Histology slides were produced and digitally scanned. A central genitourinary pathologist (P.L.) and eight African general pathologists reviewed anonymized digital whole-slide images for International Society of Urological Pathology grade groups and other pathologic parameters. Discrepancies were re-evaluated, and consensus grading was assigned. A virtual training seminar on prostate cancer grading was followed by a second assessment on a subcohort of the same tissue set.

RESULTS

Of 134 tissue blocks, 133 had evaluable tissue; 13 lacked cancer evidence, and four were of insufficient quality. Post-training, interobserver agreement for grade groups improved to 56%, with a median Cohen's quadratic weighted kappa of 0.83 (mean, 0.74), compared with an initial 46% agreement and a quadratic weighted kappa of 0.77. Interobserver agreement between African pathologist groups was 40%, with a quadratic weighted kappa of 0.66 (95% CI, 0.51 to 0.76). African pathologists tended to overgrade (36%) more frequently than undergrade (18%) compared with the reference genitourinary pathologist. Interobserver variability tended to worsen with a decrease in tissue quality.

CONCLUSION

Tissue-based studies on prostate cancer in men of African descent are essential for a better understanding of this common disease. Standardized tissue handling protocols are crucial to ensure good tissue quality and data. The use of digital slide imaging can enhance collaboration among pathologists in multinational, multicenter studies.

Deep learning-based identification of esophageal cancer subtypes through analysis of high-resolution histopathology images

Esophageal cancer (EC) remains a significant health challenge globally, with increasing incidence and high mortality rates. Despite advances in treatment, there remains a need for improved diagnostic methods and understanding of disease progression. This study addresses the significant challenges in the automatic classification of EC, particularly in distinguishing its primary subtypes: adenocarcinoma and squamous cell carcinoma, using histopathology images. Traditional histopathological diagnosis, while being the gold standard, is subject to subjectivity and human error and imposes a substantial burden on pathologists. This study proposes a binary class classification system for detecting EC subtypes in response to these challenges. The system leverages deep learning techniques and tissue-level labels for enhanced accuracy. We utilized 59 high-resolution histopathological images from The Cancer Genome Atlas (TCGA) Esophageal Carcinoma dataset (TCGA-ESCA). These images were preprocessed, segmented into patches, and analyzed using a pre-trained ResNet101 model for feature extraction. For classification, we employed five machine learning classifiers: Support Vector Classifier (SVC), Logistic Regression (LR), Decision Tree (DT), AdaBoost (AD), Random Forest (RF), and a Feed-Forward Neural Network (FFNN). The classifiers were evaluated based on their prediction accuracy on the test dataset, yielding results of 0.88 (SVC and LR), 0.64 (DT and AD), 0.82 (RF), and 0.94 (FFNN). Notably, the FFNN classifier achieved the highest Area Under the Curve (AUC) score of 0.92, indicating its superior performance, followed closely by SVC and LR, with a score of 0.87. This suggested approach holds promising potential as a decision-support tool for pathologists, particularly in regions with limited resources and expertise. The timely and precise detection of EC subtypes through this system can substantially enhance the likelihood of successful treatment, ultimately leading to reduced mortality rates in patients with this aggressive cancer.

A platform-independent AI tumor lineage and site (ATLAS) classifier

Histopathologic diagnosis and classification of cancer plays a critical role in guiding treatment. Advances in next-generation sequencing have ushered in new complementary molecular frameworks. However, existing approaches do not independently assess both site-of-origin (e.g. prostate) and lineage (e.g. adenocarcinoma) and have minimal validation in metastatic disease, where classification is more difficult. Utilizing gradient-boosted machine learning, we developed ATLAS, a pair of separate AI Tumor Lineage and Site-of-origin models from RNA expression data on 8249 tumor samples. We assessed performance independently in 10,376 total tumor samples, including 1490 metastatic samples, achieving an accuracy of 91.4% for cancer site-of-origin and 97.1% for cancer lineage. High confidence predictions (encompassing the majority of cases) were accurate 98-99% of the time in both localized and remarkably even in metastatic samples. We also identified emergent properties of our lineage scores for tumor types on which the model was never trained (zero-shot learning). Adenocarcinoma/sarcoma lineage scores differentiated epithelioid from biphasic/sarcomatoid mesothelioma. Also, predicted lineage de-differentiation identified neuroendocrine/small cell tumors and was associated with poor outcomes across tumor types. Our platform-independent single-sample approach can be easily translated to existing RNA-seq platforms. ATLAS can complement and guide traditional histopathologic assessment in challenging situations and tumors of unknown primary.

Machine learning classification of diagnostic accuracy in pathologists interpreting breast biopsies

OBJECTIVE

This study explores the feasibility of using machine learning to predict accurate versus inaccurate diagnoses made by pathologists based on their spatiotemporal viewing behavior when evaluating digital breast biopsy images.

MATERIALS AND METHODS

The study gathered data from 140 pathologists of varying experience levels who each reviewed a set of 14 digital whole slide images of breast biopsy tissue. Pathologists' viewing behavior, including zooming and panning actions, was recorded during image evaluation. A total of 30 features were extracted from the viewing behavior data, and 4 machine learning algorithms were used to build classifiers for predicting diagnostic accuracy.

RESULTS

The Random Forest classifier demonstrated the best overall performance, achieving a test accuracy of 0.81 and area under the receiver-operator characteristic curve of 0.86. Features related to attention distribution and focus on critical regions of interest were found to be important predictors of diagnostic accuracy. Further including case-level and pathologist-level information incrementally improved classifier performance.

DISCUSSION

Results suggest that pathologists' viewing behavior during digital image evaluation can be leveraged to predict diagnostic accuracy, affording automated feedback and decision support systems based on viewing behavior to aid in training and, ultimately, clinical practice. They also carry implications for basic research examining the interplay between perception, thought, and action in diagnostic decision-making.

CONCLUSION

The classifiers developed herein have potential applications in training and clinical settings to provide timely feedback and support to pathologists during diagnostic decision-making. Further research could explore the generalizability of these findings to other medical domains and varied levels of expertise.

An interpretable decision-support model for breast cancer diagnosis using histopathology images

Microscopic examination of biopsy tissue slides is perceived as the gold-standard methodology for the confirmation of presence of cancer cells. Manual analysis of an overwhelming inflow of tissue slides is highly susceptible to misreading of tissue slides by pathologists. A computerized framework for histopathology image analysis is conceived as a diagnostic tool that greatly benefits pathologists, augmenting definitive diagnosis of cancer. Convolutional Neural Network (CNN) turned out to be the most adaptable and effective technique in the detection of abnormal pathologic histology. Despite their high sensitivity and predictive power, clinical translation is constrained by a lack of intelligible insights into the prediction. A computer-aided system that can offer a definitive diagnosis and interpretability is therefore highly desirable. Conventional visual explanatory techniques, Class Activation Mapping (CAM), combined with CNN models offers interpretable decision making. The major challenge in CAM is, it cannot be optimized to create the best visualization map. CAM also decreases the performance of the CNN models. To address this challenge, we introduce a novel interpretable decision-support model using CNN with a trainable attention mechanism using response-based feed-forward visual explanation. We introduce a variant of DarkNet19 CNN model for the classification of histopathology images. In order to achieve visual interpretation as well as boost the performance of the DarkNet19 model, an attention branch is integrated with DarkNet19 network forming Attention Branch Network (ABN). The attention branch uses a convolution layer of DarkNet19 and Global Average Pooling (GAP) to model the context of the visual features and generate a heatmap to identify the region of interest. Finally, the perception branch is constituted using a fully connected layer to classify images. We trained and validated our model using more than 7000 breast cancer biopsy slide images from an openly available dataset and achieved 98.7% accuracy in the binary classification of histopathology images. The observations substantiated the enhanced clinical interpretability of the DarkNet19 CNN model, supervened by the attention branch, besides delivering a 3%-4% performance boost of the baseline model. The cancer regions highlighted by the proposed model correlate well with the findings of an expert pathologist. The coalesced approach of unifying attention branch with the CNN model capacitates pathologists with augmented diagnostic interpretability of histological images with no detriment to state-of-art performance. The model's proficiency in pinpointing the region of interest is an added bonus that can lead to accurate clinical translation of deep learning models that underscore clinical decision support.

From Image to Diagnosis: Characterizing Sources of Error in Histopathologic Interpretation

An accurate histopathologic diagnosis on surgical biopsy material is necessary for the clinical management of patients and has important implications for research, clinical trial design/enrollment, and public health education. This study used a mixed methods approach to isolate sources of diagnostic error while residents and attending pathologists interpreted digitized breast biopsy slides. Ninety participants including pathology residents and attendings at major United States medical centers reviewed a set of 14 digitized whole slide images of breast biopsies. Each case had a consensus-defined diagnosis and critical region of interest (cROI) representing the most significant pathology on the slide. Participants were asked to view unmarked digitized slides, draw their own participant region of interest (pROI), describe its features, and render a diagnosis. Participants' review behavior was tracked using case viewer software and an eye-tracking device. Diagnostic accuracy was calculated in comparison to the consensus diagnosis. We measured the frequency of errors emerging during four interpretive phases: 1) detecting the cROI, 2) recognizing its relevance, 3) using the correct terminology to describe findings in the pROI, and 4) making a diagnostic decision. According to eye tracking data, both trainees and attending pathologists were very likely (about 94% of the time) to find the cROI when inspecting a slide. However, trainees were less likely to consider the cROI relevant to their diagnosis. Pathology trainees were more likely (41% of cases) to use incorrect terminology to describe pROI features than attending pathologists (21% of cases). Failure to accurately describe features was the only factor strongly associated with an incorrect diagnosis. Identifying where errors emerge in the interpretive and/or descriptive process and working on building organ-specific feature recognition and verbal fluency in describing those features are critical steps for achieving competency in diagnostic decision making.

Zoom behavior during visual search modulates pupil diameter and reflects adaptive control states

Adaptive gain theory proposes that the dynamic shifts between exploration and exploitation control states are modulated by the locus coeruleus-norepinephrine system and reflected in tonic and phasic pupil diameter. This study tested predictions of this theory in the context of a societally important visual search task: the review and interpretation of digital whole slide images of breast biopsies by physicians (pathologists). As these medical images are searched, pathologists encounter difficult visual features and intermittently zoom in to examine features of interest. We propose that tonic and phasic pupil diameter changes during image review may correspond to perceived difficulty and dynamic shifts between exploration and exploitation control states. To examine this possibility, we monitored visual search behavior and tonic and phasic pupil diameter while pathologists (N = 89) interpreted 14 digital images of breast biopsy tissue (1,246 total images reviewed). After viewing the images, pathologists provided a diagnosis and rated the level of difficulty of the image. Analyses of tonic pupil diameter examined whether pupil dilation was associated with pathologists' difficulty ratings, diagnostic accuracy, and experience level. To examine phasic pupil diameter, we parsed continuous visual search data into discrete zoom-in and zoom-out events, including shifts from low to high magnification (e.g., 1× to 10×) and the reverse. Analyses examined whether zoom-in and zoom-out events were associated with phasic pupil diameter change. Results demonstrated that tonic pupil diameter was associated with image difficulty ratings and zoom level, and phasic pupil diameter showed constriction upon zoom-in events, and dilation immediately preceding a zoom-out event. Results are interpreted in the context of adaptive gain theory, information gain theory, and the monitoring and assessment of physicians' diagnostic interpretive processes.

Pathologist pupil dilation reflects experience level and difficulty in diagnosing medical images

Purpose: Digital whole slide imaging allows pathologists to view slides on a computer screen instead of under a microscope. Digital viewing allows for real-time monitoring of pathologists' search behavior and neurophysiological responses during the diagnostic process. One particular neurophysiological measure, pupil diameter, could provide a basis for evaluating clinical competence during training or developing tools that support the diagnostic process. Prior research shows that pupil diameter is sensitive to cognitive load and arousal, and it switches between exploration and exploitation of a visual image. Different categories of lesions in pathology pose different levels of challenge, as indicated by diagnostic disagreement among pathologists. If pupil diameter is sensitive to the perceived difficulty in diagnosing biopsies, eye-tracking could potentially be used to identify biopsies that may benefit from a second opinion. Approach: We measured case onset baseline-corrected (phasic) and uncorrected (tonic) pupil diameter in 90 pathologists who each viewed and diagnosed 14 digital breast biopsy cases that cover the diagnostic spectrum from benign to invasive breast cancer. Pupil data were extracted from the beginning of viewing and interpreting of each individual case. After removing 122 trials ( < 10 % ) with poor eye-tracking quality, 1138 trials remained. We used multiple linear regression with robust standard error estimates to account for dependent observations within pathologists. Results: We found a positive association between the magnitude of phasic dilation and subject-centered difficulty ratings and between the magnitude of tonic dilation and untransformed difficulty ratings. When controlling for case diagnostic category, only the tonic-difficulty relationship persisted. Conclusions: Results suggest that tonic pupil dilation may indicate overall arousal differences between pathologists as they interpret biopsy cases and could signal a need for additional training, experience, or automated decision aids. Phasic dilation is sensitive to characteristics of biopsies that tend to elicit higher difficulty ratings and could indicate a need for a second opinion.

Deep learning in digital pathology for personalized treatment plans of cancer patients

Over the past decade, many new cancer treatments have been developed and made available to patients. However, in most cases, these treatments only benefit a specific subgroup of patients, making the selection of treatment for a specific patient an essential but challenging task for oncologists. Although some biomarkers were found to associate with treatment response, manual assessment is time-consuming and subjective. With the rapid developments and expanded implementation of artificial intelligence (AI) in digital pathology, many biomarkers can be quantified automatically from histopathology images. This approach allows for a more efficient and objective assessment of biomarkers, aiding oncologists in formulating personalized treatment plans for cancer patients. This review presents an overview and summary of the recent studies on biomarker quantification and treatment response prediction using hematoxylin-eosin (H&E) stained pathology images. These studies have shown that an AI-based digital pathology approach can be practical and will become increasingly important in improving the selection of cancer treatments for patients.

Changes in the Diagnoses of Breast Core Needle Biopsies on Second Review at a Tertiary Care Center: Implications for Surgical Management

Core needle biopsy (CNB) of breast lesions is routine for diagnosis and treatment planning. Despite refinement of diagnostic criteria, the diagnosis of breast lesions on CNB can be challenging. At many centers, including ours, confirmation of diagnoses rendered in other laboratories is required before treatment planning. We identified CNBs first diagnosed elsewhere that were reviewed in our department over the course of 1 year because the patients sought care at our center and in which a change in diagnosis had been recorded. The outside and in-house CNB diagnoses were then classified based on Breast WHO Fifth Edition diagnostic categories. The impact of the change in diagnosis was estimated based on the subsequent surgical management. Findings in follow-up surgical excisions (EXCs) were used for validation. In 2018, 4950 outside cases with CNB were reviewed at our center. A total of 403 CNBs diagnoses were discrepant. Of these, 147 had a change in the WHO diagnostic category: 80 (54%) CNBs had a more severe diagnosis and 44 (30%) a less severe diagnosis. In 23 (16%) CNBs, the change of diagnostic category had no impact on management. Intraductal proliferations (n=54), microinvasive carcinoma (n=18), and papillary lesions (n=35) were the most disputed diagnoses. The in-house CNB diagnosis was confirmed in most cases with available excisions. Following CNB reclassification, 22/147 (15%) lesions were not excised. A change affecting the surgical management at our center occurred in 2.5% of all CNBs. Our results support routine review of outside breast CNB as a clinically significant practice before definitive treatment.

Are Pathologists Self-Aware of Their Diagnostic Accuracy? Metacognition and the Diagnostic Process in Pathology

BACKGROUND

Metacognition is a cognitive process that involves self-awareness of thinking, understanding, and performance. This study assesses pathologists' metacognition by examining the association between their diagnostic accuracy and self-reported confidence levels while interpreting skin and breast biopsies.

DESIGN

We studied 187 pathologists from the Melanoma Pathology Study (M-Path) and 115 pathologists from the Breast Pathology Study (B-Path). We measured pathologists' metacognitive ability by examining the area under the curve (AUC), the area under each pathologist's receiver operating characteristic (ROC) curve summarizing the association between confidence and diagnostic accuracy. We investigated possible relationships between this AUC measure, referred to as metacognitive sensitivity, and pathologist attributes. We also assessed whether higher metacognitive sensitivity affected the association between diagnostic accuracy and a secondary diagnostic action such as requesting a second opinion.

RESULTS

We found no significant associations between pathologist clinical attributes and metacognitive AUC. However, we found that pathologists with higher AUC showed a stronger trend to request secondary diagnostic action for inaccurate diagnoses and not for accurate diagnoses compared with pathologists with lower AUC.

LIMITATIONS

Pathologists reported confidence in specific diagnostic terms, rather than the broader classes into which the diagnostic terms were later grouped to determine accuracy. In addition, while there is no gold standard for the correct diagnosis to determine the accuracy of pathologists' interpretations, our studies achieved a high-quality reference diagnosis by using the consensus diagnosis of 3 experienced pathologists.

CONCLUSIONS

Metacognition can affect clinical decisions. If pathologists have self-awareness that their diagnosis may be inaccurate, they can request additional tests or second opinions, providing the opportunity to correct inaccurate diagnoses.

HIGHLIGHTS

Metacognitive sensitivity varied across pathologists, with most showing higher sensitivity than expected by chance.None of the demographic or clinical characteristics we examined was significantly associated with metacognitive sensitivity.Pathologists with higher metacognitive sensitivity were more likely to request additional tests or second opinions for their inaccurate diagnoses.

Evaluation of a Probability-Based Predictive Tool on Pathologist Agreement Using Urinary Bladder as a Pilot Tissue

Simple Summary

There is a common joke in pathology—put three pathologists in a room and you will obtain three different answers. This saying comes from the fact that pathology can be subjective; pathologists’ diagnoses can be influenced by many different biases, and pathologists are also influenced by the presence or absence of animal information and medical history. Compared to pathology, statistics is a much more objective field. This study aimed to develop a probability-based tool using statistics obtained by analyzing 338 histopathology slides of canine and feline urinary bladders, then see if the tool affected agreement between the test pathologists. Four pathologists diagnosed 25 canine and feline bladder slides and they conducted this three times: without animal and clinical information, then with this information, and finally using the probability tool. Results showed large differences in the pathologists’ interpretation of bladder slides, with kappa agreement values (low value for digital slide images, high value for glass slides) of 7–37% without any animal or clinical information, 23–37% with animal signalment and history, and 31–42% when our probability tool was used. This study provides a starting point for the use of probability-based tools in standardizing pathologist agreement in veterinary pathology.

Abstract

Inter-pathologist variation is widely recognized across human and veterinary pathology and is often compounded by missing animal or clinical information on pathology submission forms. Variation in pathologist threshold levels of resident inflammatory cells in the tissue of interest can further decrease inter-pathologist agreement. This study applied a predictive modeling tool to bladder histology slides that were assessed by four pathologists: first without animal and clinical information, then with this information, and finally using the predictive tool. All three assessments were performed twice, using digital whole-slide images (WSI) and then glass slides. Results showed marked variation in pathologists’ interpretation of bladder slides, with kappa agreement values of 7–37% without any animal or clinical information, 23–37% with animal signalment and history, and 31–42% when our predictive tool was applied, for digital WSI and glass slides. The concurrence of test pathologists to the reference diagnosis was 60% overall. This study provides a starting point for the use of predictive modeling in standardizing pathologist agreement in veterinary pathology. It also highlights the importance of high-quality whole-slide imaging to limit the effect of digitization on inter-pathologist agreement and the benefit of continued standardization of tissue assessment in veterinary pathology.

Reproducibility and Feasibility of Classification and National Guidelines for Histological Diagnosis of Canine Mammary Gland Tumours: A Multi-Institutional Ring Study

Simple Summary

Tumours of the mammary gland are common in humans, as in canine species. They are very heterogenous with numerous morphological variants and different biologic behaviours. In the last few decades, several efforts have been made to classify these tumours histologically and establish the level of malignancy by using histologic grading systems. However, reproducibility and diagnostic agreement of such classification and grading have been only rarely assessed. In this study, we tested the variability in diagnoses performed by 15 pathologists using the same classification and grading system. Prior to the study, pathologists agreed on guidelines regarding how to apply these systems. Pathologists worked blindly on 36 digital histologic slides of canine mammary tumours. The agreement was statistically analysed using Cohen’s kappa coefficient that, when equal to 1, indicates perfect agreement. The overall agreement in the identification of hyperplastic-dysplastic/benign/malignant lesions was substantial (kappa 0.76), while outcomes on morphological classification had only a moderate agreement (k = 0.54). Tumour grade assigned by pathologists was the least concordant and kappa could not be calculated. Although promising, the results underline that each diagnostic/grading system should be assessed and optimized for standardization and high diagnostic agreement.

Abstract

Histological diagnosis of Canine Mammary Tumours (CMTs) provides the basis for proper treatment and follow-up. Nowadays, its accuracy is poorly understood and variable interpretation of histological criteria leads to a lack of standardisation and impossibility to compare studies. This study aimed to quantify the reproducibility of histological diagnosis and grading in CMTs. A blinded ring test on 36 CMTs was performed by 15 veterinary pathologists with different levels of education, after discussion of critical points on the Davis-Thompson Foundation Classification and providing consensus guidelines. Kappa statistics were used to compare the interobserver variability. The overall concordance rate of diagnostic interpretations of WP on identification of hyperplasia-dysplasia/benign/malignant lesions showed a substantial agreement (average k ranging from 0.66 to 0.82, with a k-combined of 0.76). Instead, outcomes on ICD-O-3.2 morphological code /diagnosis of histotype had only a moderate agreement (average k ranging from 0.44 and 0.64, with a k-combined of 0.54). The results demonstrated that standardised classification and consensus guidelines can produce moderate to substantial agreement; however, further efforts are needed to increase this agreement in distinguishing benign versus malignant lesions and in histological grading.

Pilot study to evaluate tools to collect pathologist annotations for validating machine learning algorithms

Purpose: Validation of artificial intelligence (AI) algorithms in digital pathology with a reference standard is necessary before widespread clinical use, but few examples focus on creating a reference standard based on pathologist annotations. This work assesses the results of a pilot study that collects density estimates of stromal tumor-infiltrating lymphocytes (sTILs) in breast cancer biopsy specimens. This work will inform the creation of a validation dataset for the evaluation of AI algorithms fit for a regulatory purpose. Approach: Collaborators and crowdsourced pathologists contributed glass slides, digital images, and annotations. Here, "annotations" refer to any marks, segmentations, measurements, or labels a pathologist adds to a report, image, region of interest (ROI), or biological feature. Pathologists estimated sTILs density in 640 ROIs from hematoxylin and eosin stained slides of 64 patients via two modalities: an optical light microscope and two digital image viewing platforms. Results: The pilot study generated 7373 sTILs density estimates from 29 pathologists. Analysis of annotations found the variability of density estimates per ROI increases with the mean; the root mean square differences were 4.46, 14.25, and 26.25 as the mean density ranged from 0% to 10%, 11% to 40%, and 41% to 100%, respectively. The pilot study informs three areas of improvement for future work: technical workflows, annotation platforms, and agreement analysis methods. Upgrades to the workflows and platforms will improve operability and increase annotation speed and consistency. Conclusions: Exploratory data analysis demonstrates the need to develop new statistical approaches for agreement. The pilot study dataset and analysis methods are publicly available to allow community feedback. The development and results of the validation dataset will be publicly available to serve as an instructive tool that can be replicated by developers and researchers.

Usability of deep learning and H&E images predict disease outcome-emerging tool to optimize clinical trials

Understanding factors that impact prognosis for cancer patients have high clinical relevance for treatment decisions and monitoring of the disease outcome. Advances in artificial intelligence (AI) and digital pathology offer an exciting opportunity to capitalize on the use of whole slide images (WSIs) of hematoxylin and eosin (H&E) stained tumor tissue for objective prognosis and prediction of response to targeted therapies. AI models often require hand-delineated annotations for effective training which may not be readily available for larger data sets. In this study, we investigated whether AI models can be trained without region-level annotations and solely on patient-level survival data. We present a weakly supervised survival convolutional neural network (WSS-CNN) approach equipped with a visual attention mechanism for predicting overall survival. The inclusion of visual attention provides insights into regions of the tumor microenvironment with the pathological interpretation which may improve our understanding of the disease pathomechanism. We performed this analysis on two independent, multi-center patient data sets of lung (which is publicly available data) and bladder urothelial carcinoma. We perform univariable and multivariable analysis and show that WSS-CNN features are prognostic of overall survival in both tumor indications. The presented results highlight the significance of computational pathology algorithms for predicting prognosis using H&E stained images alone and underpin the use of computational methods to improve the efficiency of clinical trial studies.

General Roadmap and Core Steps for the Development of AI Tools in Digital Pathology

Integrating Artificial Intelligence (AI) tools in the tissue diagnostic workflow will benefit the pathologist and, ultimately, the patient. The generation of such AI tools has two parallel yet interconnected processes, namely the definition of the pathologist’s task to be delivered in silico, and the software development requirements. In this review paper, we demystify this process, from a viewpoint that joins experienced pathologists and data scientists, by proposing a general pathway and describing the core steps to build an AI digital pathology tool. In doing so, we highlight the importance of the collaboration between AI scientists and pathologists, from the initial formulation of the hypothesis to the final, ready-to-use product.

End-to-End Diagnosis of Breast Biopsy Images with Transformers

Diagnostic disagreements among pathologists occur throughout the spectrum of benign to malignant lesions. A computer-aided diagnostic system capable of reducing uncertainties would have important clinical impact. To develop a computer-aided diagnosis method for classifying breast biopsy images into a range of diagnostic categories (benign, atypia, ductal carcinoma in situ, and invasive breast cancer), we introduce a transformer-based hollistic attention network called HATNet. Unlike state-of-the-art histopathological image classification systems that use a two pronged approach, i.e., they first learn local representations using a multi-instance learning framework and then combine these local representations to produce image-level decisions, HATNet streamlines the histopathological image classification pipeline and shows how to learn representations from gigapixel size images end-to-end. HATNet extends the bag-of-words approach and uses self-attention to encode global information, allowing it to learn representations from clinically relevant tissue structures without any explicit supervision. It outperforms the previous best network Y-Net, which uses supervision in the form of tissue-level segmentation masks, by 8%. Importantly, our analysis reveals that HATNet learns representations from clinically relevant structures, and it matches the classification accuracy of 87 U.S. pathologists for this challenging test set.

Breast cancer or surrounding normal tissue? A successful discrimination by FTIR or Raman microspectroscopy

Breast cancer is a type of cancer with the highest incidence worldwide in 2021, with early diagnosis and rapid treatment intervention being the reasons for the decreasing mortality rate associated with the disease.

Recommendations on compiling test datasets for evaluating artificial intelligence solutions in pathology

Artificial intelligence (AI) solutions that automatically extract information from digital histology images have shown great promise for improving pathological diagnosis. Prior to routine use, it is important to evaluate their predictive performance and obtain regulatory approval. This assessment requires appropriate test datasets. However, compiling such datasets is challenging and specific recommendations are missing. A committee of various stakeholders, including commercial AI developers, pathologists, and researchers, discussed key aspects and conducted extensive literature reviews on test datasets in pathology. Here, we summarize the results and derive general recommendations on compiling test datasets. We address several questions: Which and how many images are needed? How to deal with low-prevalence subsets? How can potential bias be detected? How should datasets be reported? What are the regulatory requirements in different countries? The recommendations are intended to help AI developers demonstrate the utility of their products and to help pathologists and regulatory agencies verify reported performance measures. Further research is needed to formulate criteria for sufficiently representative test datasets so that AI solutions can operate with less user intervention and better support diagnostic workflows in the future.

A Pyramid Architecture-Based Deep Learning Framework for Breast Cancer Detection

Breast cancer diagnosis is a critical step in clinical decision making, and this is achieved by making a pathological slide and gives a decision by the doctors, which is the method of final decision making for cancer diagnosis. Traditionally, the doctors usually check the pathological images by visual inspection under the microscope. Whole-slide images (WSIs) have supported the state-of-the-art diagnosis results and have been admitted as the gold standard clinically. However, this task is time-consuming and labour-intensive, and all of these limitations make low efficiency in decision making. Medical image processing protocols have been used for this task during the last decades and have obtained satisfactory results under some conditions; especially in the deep learning era, it has exhibited the advantages than those in the shallow learning period. In this paper, we proposed a novel breast cancer region mining framework based on deep pyramid architecture from multilevel and multiscale breast pathological WSIs. We incorporate the tissue- and cell-level information together and integrate these into a LSTM model for the final sequence modelling, which successfully keeps the WSIs' integration and is not mentioned by the prevalence frameworks. The experiment results demonstrated that our proposed framework greatly improved the detection accuracy than that only using tissue-level information.

More scanning, but not zooming, is associated with diagnostic accuracy in evaluating digital breast pathology slides

Diagnoses of medical images can invite strikingly diverse strategies for image navigation and visual search. In computed tomography screening for lung nodules, distinct strategies, termed scanning and drilling, relate to both radiologists' clinical experience and accuracy in lesion detection. Here, we examined associations between search patterns and accuracy for pathologists (N = 92) interpreting a diverse set of breast biopsy images. While changes in depth in volumetric images reveal new structures through movement in the z-plane, in digital pathology changes in depth are associated with increased magnification. Thus, "drilling" in radiology may be more appropriately termed "zooming" in pathology. We monitored eye-movements and navigation through digital pathology slides to derive metrics of how quickly the pathologists moved through XY (scanning) and Z (zooming) space. Prior research on eye-movements in depth has categorized clinicians as either "scanners" or "drillers." In contrast, we found that there was no reliable association between a clinician's tendency to scan or zoom while examining digital pathology slides. Thus, in the current work we treated scanning and zooming as continuous predictors rather than categorizing as either a "scanner" or "zoomer." In contrast to prior work in volumetric chest images, we found significant associations between accuracy and scanning rate but not zooming rate. These findings suggest fundamental differences in the relative value of information types and review behaviors across two image formats. Our data suggest that pathologists gather critical information by scanning on a given plane of depth, whereas radiologists drill through depth to interrogate critical features.

Development and reliability of the histological THROMBEX-classification rule for thrombotic emboli of acute ischemic stroke patients

Background

Thrombus histology has become a potential diagnostic tool for the etiology assessment of patients with ischemic stroke caused by embolic proximal vessel occlusion. We validated a classification rule that differentiates between cardiac and arteriosclerotic emboli in individual stroke patients. We aim to describe in detail the development of this classification rule and disclose its reliability.

Methods

The classification rule is based on the hypothesis that cardiac emboli arise out of separation thrombi and arteriosclerotic emboli result from agglutinative thrombi. 125 emboli recovered by thrombectomy from stroke patients and 11 thrombi serving as references for cardiac (n = 5) and arteriosclerotic emboli (n = 6) were Hematoxylin and eosin, Elastica-van Gieson and CD61 stained and rated independently by two histopathologists blinded to the presumed etiology by several pre-defined criteria. Intra- and interobserver reliabilities of all criteria were determined. Out of the different criteria, three criteria with the most satisfactory reliability values were selected to compose the classification rule that was finally adjusted to the reference thrombi. Reliabilities of the classification rule were calculated by using the emboli of stroke patients.

Results

The classification rule reached intraobserver reliabilities for the two raters of 92.9% and 68.2%, respectively. Interobserver reliability was 69.9%.

Conclusions

A new classification rule for emboli obtained from thrombectomy was established. Within the limitations of histological investigations, it is reliable and able to distinguish between cardioembolic and arteriosclerotic emboli.

Supplementary Information

The online version contains supplementary material available at 10.1186/s42466-021-00149-6.

Improving Generalization of Deep Learning Models for Diagnostic Pathology by Increasing Variability in Training Data: Experiments on Osteosarcoma Subtypes

BACKGROUND

Artificial intelligence has an emerging progress in diagnostic pathology. A large number of studies of applying deep learning models to histopathological images have been published in recent years. While many studies claim high accuracies, they may fall into the pitfalls of overfitting and lack of generalization due to the high variability of the histopathological images.

AIMS AND OBJECTS

Use the model training of osteosarcoma as an example to illustrate the pitfalls of overfitting and how the addition of model input variability can help improve model performance.

MATERIALS AND METHODS

We use the publicly available osteosarcoma dataset to retrain a previously published classification model for osteosarcoma. We partition the same set of images into the training and testing datasets differently than the original study: the test dataset consists of images from one patient while the training dataset consists images of all other patients. We also show the influence of training data variability on model performance by collecting a minimal dataset of 10 osteosarcoma subtypes as well as benign tissues and benign bone tumors of differentiation.

RESULTS

The performance of the re-trained model on the test set using the new partition schema declines dramatically, indicating a lack of model generalization and overfitting. We show the additions of more and moresubtypes into the training data step by step under the same model schema yield a series of coherent models with increasing performances.

CONCLUSIONS

In conclusion, we bring forward data preprocessing and collection tactics for histopathological images of high variability to avoid the pitfalls of overfitting and build deep learning models of higher generalization abilities.

Diagnostic concordance among pathologists interpreting breast core biopsies on secondary review over a 1-year period at an Australian tertiary hospital

This study provides data on the diagnostic concordance between initial and review diagnoses of all breast core biopsy cases at a single tertiary hospital in Western Australia over a 1-year period. A retrospective review of all breast core biopsy cases between January 1 and December 31, 2016, was carried out at PathWest, Fiona Stanley Hospital in Perth, Western Australia. Each biopsy is reported by a single pathologist and then reviewed within 1 week by a panel of intradepartmental subspecialist breast pathologists, who either agree with the original diagnosis, have a minor discordant diagnosis, or a major discordant diagnosis. Records for 2036 core biopsies were available between January 1 and December 31, 2016. Of these, 56.0% (n = 1141) were classified as benign, 34.3% (n = 699) as malignant, 7.2% (n = 147) as indeterminate, 2.3% (n = 46) as nondiagnostic, and 0.1% (n = 3) as suspicious for malignancy. In 99.1% (n = 2018) of cases, there was agreement between initial and review diagnoses. In total, 0.9% (n = 18) were disagreements: 0.49% (n = 10) were major discordant disagreements and 0.39% (n = 8) were minor discordant disagreements. All cases of major discordant disagreements would have resulted in significant changes to clinical management. This study demonstrates that an Australian institution is providing a high-quality pathology service with a low error rate between initial and review diagnoses of breast core biopsies. It reinforces the importance of secondary review of biopsies in a timely fashion for detecting potentially serious misdiagnoses that could lead to inappropriate management.

A tree-based multiclassification of breast tumor histopathology images through deep learning

Worldwide, the burden of cancer is drastically increasing over the past few years. Among all types of cancers in women, breast cancer (BrC) is the main cause of unnatural deaths. For early diagnosis, histopathology (Hp) imaging is a gold standard for positive and detailed (at tissue level) diagnosis of breast tumor (BrT) compared to mammogram images. A large number of studies used BrT Hp images to solve binary or multiclassification problems using high computational resources. However, classification models' performance may be compromised due to the high correlation among various types of BrT in Hp images, which raises the misclassification rate. Thus, this paper aims to develop a tree-based BrT multiclassification model via deep learning (DL) to extract discriminative features to solve the multiclassification problem with better performance using less computational resources. The main contributions of this work are to create an ensemble, tree-based DL model that is pre-trained on the BreakHis dataset, and implementation of a misclassification reduction algorithm. The ensemble, tree-based DL model, extracts discriminative BrT features from Hp images. The target dataset (i.e., Bioimaging challenge 2015 breast histology) is small in size; thus, to avoid overfitting of the proposed model, pretraining is performed on the BreakHis dataset. Whereas, misclassification reduction algorithm is implemented to enhance the performance of the classification model. The experimental results show that the proposed model outperformed the existing state-of-the-art baseline studies. The achieved classification accuracy is ranging from 87.50 % to 100 % for four subtypes of BrT. Thus, the proposed model can assist doctors as the second opinion in any healthcare centre.

Scale-Aware Transformers for Diagnosing Melanocytic Lesions

Diagnosing melanocytic lesions is one of the most challenging areas of pathology with extensive intra- and inter-observer variability. The gold standard for a diagnosis of invasive melanoma is the examination of histopathological whole slide skin biopsy images by an experienced dermatopathologist. Digitized whole slide images offer novel opportunities for computer programs to improve the diagnostic performance of pathologists. In order to automatically classify such images, representations that reflect the content and context of the input images are needed. In this paper, we introduce a novel self-attention-based network to learn representations from digital whole slide images of melanocytic skin lesions at multiple scales. Our model softly weighs representations from multiple scales, allowing it to discriminate between diagnosis-relevant and -irrelevant information automatically. Our experiments show that our method outperforms five other state-of-the-art whole slide image classification methods by a significant margin. Our method also achieves comparable performance to 187 practicing U.S. pathologists who interpreted the same cases in an independent study. To facilitate relevant research, full training and inference code is made publicly available at https://github.com/meredith-wenjunwu/ScATNet.

Deep learning-based cross-classifications reveal conserved spatial behaviors within tumor histological images

Histopathological images are a rich but incompletely explored data type for studying cancer. Manual inspection is time consuming, making it challenging to use for image data mining. Here we show that convolutional neural networks (CNNs) can be systematically applied across cancer types, enabling comparisons to reveal shared spatial behaviors. We develop CNN architectures to analyze 27,815 hematoxylin and eosin scanned images from The Cancer Genome Atlas for tumor/normal, cancer subtype, and mutation classification. Our CNNs are able to classify TCGA pathologist-annotated tumor/normal status of whole slide images (WSIs) in 19 cancer types with consistently high AUCs (0.995 ± 0.008), as well as subtypes with lower but significant accuracy (AUC 0.87 ± 0.1). Remarkably, tumor/normal CNNs trained on one tissue are effective in others (AUC 0.88 ± 0.11), with classifier relationships also recapitulating known adenocarcinoma, carcinoma, and developmental biology. Moreover, classifier comparisons reveal intra-slide spatial similarities, with an average tile-level correlation of 0.45 ± 0.16 between classifier pairs. Breast cancers, bladder cancers, and uterine cancers have spatial patterns that are particularly easy to detect, suggesting these cancers can be canonical types for image analysis. Patterns for TP53 mutations can also be detected, with WSI self- and cross-tissue AUCs ranging from 0.65-0.80. Finally, we comparatively evaluate CNNs on 170 breast and colon cancer images with pathologist-annotated nuclei, finding that both cellular and intercellular regions contribute to CNN accuracy. These results demonstrate the power of CNNs not only for histopathological classification, but also for cross-comparisons to reveal conserved spatial behaviors across tumors.

Deep computational pathology in breast cancer

Deep Learning (DL) algorithms are a set of techniques that exploit large and/or complex real-world datasets for cross-domain and cross-discipline prediction and classification tasks. DL architectures excel in computer vision tasks, and in particular image processing and interpretation. This has prompted a wave of disruptingly innovative applications in medical imaging, where DL strategies have the potential to vastly outperform human experts. This is particularly relevant in the context of histopathology, where whole slide imaging (WSI) of stained tissue in conjuction with DL algorithms for their interpretation, selection and cancer staging are beginning to play an ever increasing role in supporting human operators in visual assessments. This has the potential to reduce everyday workload as well as to increase precision and reproducibility across observers, centers, staining techniques and even pathologies. In this paper we introduce the most common DL architectures used in image analysis, with a focus on histopathological image analysis in general and in breast histology in particular. We briefly review how, state-of-art DL architectures compare to human performance on across a number of critical tasks such as mitotic count, tubules analysis and nuclear pleomorphism analysis. Also, the development of DL algorithms specialized to pathology images have been enormously fueled by a number of world-wide challenges based on large, multicentric image databases which are now publicly available. In turn, this has allowed most recent efforts to shift more and more towards semi-supervised learning methods, which provide greater flexibility and applicability. We also review all major repositories of manually labelled pathology images in breast cancer and provide an in-depth discussion of the challenges specific to training DL architectures to interpret WSI data, as well as a review of the state-of-the-art methods for interpretation of images generated from immunohistochemical analysis of breast lesions. We finally discuss the future challenges and opportunities which the adoption of DL paradigms is most likely to pose in the field of pathology for breast cancer detection, diagnosis, staging and prognosis. This review is intended as a comprehensive stepping stone into the field of modern computational pathology for a transdisciplinary readership across technical and medical disciplines.

Deep learned tissue "fingerprints" classify breast cancers by ER/PR/Her2 status from H&E images

Because histologic types are subjective and difficult to reproduce between pathologists, tissue morphology often takes a back seat to molecular testing for the selection of breast cancer treatments. This work explores whether a deep-learning algorithm can learn objective histologic H&E features that predict the clinical subtypes of breast cancer, as assessed by immunostaining for estrogen, progesterone, and Her2 receptors (ER/PR/Her2). Translating deep learning to this and related problems in histopathology presents a challenge due to the lack of large, well-annotated data sets, which are typically required for the algorithms to learn statistically significant discriminatory patterns. To overcome this limitation, we introduce the concept of “tissue fingerprints,” which leverages large, unannotated datasets in a label-free manner to learn H&E features that can distinguish one patient from another. The hypothesis is that training the algorithm to learn the morphological differences between patients will implicitly teach it about the biologic variation between them. Following this training internship, we used the features the network learned, which we call “fingerprints,” to predict ER, PR, and Her2 status in two datasets. Despite the discovery dataset being relatively small by the standards of the machine learning community (n = 939), fingerprints enabled the determination of ER, PR, and Her2 status from whole slide H&E images with 0.89 AUC (ER), 0.81 AUC (PR), and 0.79 AUC (Her2) on a large, independent test set (n = 2531). Tissue fingerprints are concise but meaningful histopathologic image representations that capture biological information and may enable machine learning algorithms that go beyond the traditional ER/PR/Her2 clinical groupings by directly predicting theragnosis.

Artificial intelligence in digital breast pathology: Techniques and applications

Breast cancer is the most common cancer and second leading cause of cancer-related death worldwide. The mainstay of breast cancer workup is histopathological diagnosis - which guides therapy and prognosis. However, emerging knowledge about the complex nature of cancer and the availability of tailored therapies have exposed opportunities for improvements in diagnostic precision. In parallel, advances in artificial intelligence (AI) along with the growing digitization of pathology slides for the primary diagnosis are a promising approach to meet the demand for more accurate detection, classification and prediction of behaviour of breast tumours. In this article, we cover the current and prospective uses of AI in digital pathology for breast cancer, review the basics of digital pathology and AI, and outline outstanding challenges in the field.

Assessment of Machine Learning of Breast Pathology Structures for Automated Differentiation of Breast Cancer and High-Risk Proliferative Lesions

IMPORTANCE

Following recent US Food and Drug Administration approval, adoption of whole slide imaging in clinical settings may be imminent, and diagnostic accuracy, particularly among challenging breast biopsy specimens, may benefit from computerized diagnostic support tools.

OBJECTIVE

To develop and evaluate computer vision methods to assist pathologists in diagnosing the full spectrum of breast biopsy samples, from benign to invasive cancer.

DESIGN, SETTING, AND PARTICIPANTS

In this diagnostic study, 240 breast biopsies from Breast Cancer Surveillance Consortium registries that varied by breast density, diagnosis, patient age, and biopsy type were selected, reviewed, and categorized by 3 expert pathologists as benign, atypia, ductal carcinoma in situ (DCIS), and invasive cancer. The atypia and DCIS cases were oversampled to increase statistical power. High-resolution digital slide images were obtained, and 2 automated image features (tissue distribution feature and structure feature) were developed and evaluated according to the consensus diagnosis of the expert panel. The performance of the automated image analysis methods was compared with independent interpretations from 87 practicing US pathologists. Data analysis was performed between February 2017 and February 2019.

MAIN OUTCOMES AND MEASURES

Diagnostic accuracy defined by consensus reference standard of 3 experienced breast pathologists.

RESULTS

The accuracy of machine learning tissue distribution features, structure features, and pathologists for classification of invasive cancer vs noninvasive cancer was 0.94, 0.91, and 0.98, respectively; the accuracy of classification of atypia and DCIS vs benign tissue was 0.70, 0.70, and 0.81, respectively; and the accuracy of classification of DCIS vs atypia was 0.83, 0.85, and 0.80, respectively. The sensitivity of both machine learning features was lower than that of the pathologists for the invasive vs noninvasive classification (tissue distribution feature, 0.70; structure feature, 0.49; pathologists, 0.84) but higher for the classification of atypia and DCIS vs benign cases (tissue distribution feature, 0.79; structure feature, 0.85; pathologists, 0.72) and the classification of DCIS vs atypia (tissue distribution feature, 0.88; structure feature, 0.89; pathologists, 0.70). For the DCIS vs atypia classification, the specificity of the machine learning feature classification was similar to that of the pathologists (tissue distribution feature, 0.78; structure feature, 0.80; pathologists, 0.82).

CONCLUSION AND RELEVANCE

The computer-based automated approach to interpreting breast pathology showed promise, especially as a diagnostic aid in differentiating DCIS from atypical hyperplasia.

Breast Cancer Prognostic Factors in the Digital Era: Comparison of Nottingham Grade using Whole Slide Images and Glass Slides

Background:

To assess reproducibility and accuracy of overall Nottingham grade and component scores using digital whole slide images (WSIs) compared to glass slides.

Methods:

Two hundred and eight pathologists were randomized to independently interpret 1 of 4 breast biopsy sets using either glass slides or digital WSI. Each set included 5 or 6 invasive carcinomas (22 total invasive cases). Participants interpreted the same biopsy set approximately 9 months later following a second randomization to WSI or glass slides. Nottingham grade, including component scores, was assessed on each interpretation, providing 2045 independent interpretations of grade. Overall grade and component scores were compared between pathologists (interobserver agreement) and for interpretations by the same pathologist (intraobserver agreement). Grade assessments were compared when the format (WSI vs. glass slides) changed or was the same for the two interpretations.

Results:

Nottingham grade intraobserver agreement was highest using glass slides for both interpretations (73%, 95% confidence interval [CI]: 68%, 78%) and slightly lower but not statistically different using digital WSI for both interpretations (68%, 95% CI: 61%, 75%; P= 0.22). The agreement was lowest when the format changed between interpretations (63%, 95% CI: 59%, 68%). Interobserver agreement was significantly higher (P < 0.001) using glass slides versus digital WSI (68%, 95% CI: 66%, 70% versus 60%, 95% CI: 57%, 62%, respectively). Nuclear pleomorphism scores had the lowest inter- and intra-observer agreement. Mitotic scores were higher on glass slides in inter- and intra-observer comparisons.

Conclusions:

Pathologists’ intraobserver agreement (reproducibility) is similar for Nottingham grade using glass slides or WSI. However, slightly lower agreement between pathologists suggests that verification of grade using digital WSI may be more challenging.

MuDeRN: Multi-category classification of breast histopathological image using deep residual networks

MOTIVATION

Identifying carcinoma subtype can help to select appropriate treatment options and determining the subtype of benign lesions can be beneficial to estimate the patients' risk of developing cancer in the future. Pathologists' assessment of lesion subtypes is considered as the gold standard, however, sometimes strong disagreements among pathologists for distinction among lesion subtypes have been previously reported in the literature.

OBJECTIVE

To propose a framework for classifying hematoxylin-eosin stained breast digital slides either as benign or cancer, and then categorizing cancer and benign cases into four different subtypes each.

MATERIALS AND METHODS

We used data from a publicly available database (BreakHis) of 81 patients where each patient had images at four magnification factors (×40, ×100, ×200, and ×400) available, for a total of 7786 images. The proposed framework, called MuDeRN (MUlti-category classification of breast histopathological image using DEep Residual Networks) consisted of two stages. In the first stage, for each magnification factor, a deep residual network (ResNet) with 152 layers has been trained for classifying patches from the images as benign or malignant. In the next stage, the images classified as malignant were subdivided into four cancer subcategories and those categorized as benign were classified into four subtypes. Finally, the diagnosis for each patient was made by combining outputs of ResNets' processed images in different magnification factors using a meta-decision tree.

RESULTS

For the malignant/benign classification of images, MuDeRN's first stage achieved correct classification rates (CCR) of 98.52%, 97.90%, 98.33%, and 97.66% in ×40, ×100, ×200, and ×400 magnification factors respectively. For eight-class categorization of images based on the output of MuDeRN's both stages, CCRs in four magnification factors were 95.40%, 94.90%, 95.70%, and 94.60%. Finally, for making patient-level diagnosis, MuDeRN achieved a CCR of 96.25% for eight-class categorization.

CONCLUSIONS

MuDeRN can be helpful in the categorization of breast lesions.

Localization of Diagnostically Relevant Regions of Interest in Whole Slide Images: a Comparative Study

Whole slide digital imaging technology enables researchers to study pathologists' interpretive behavior as they view digital slides and gain new understanding of the diagnostic medical decision-making process. In this study, we propose a simple yet important analysis to extract diagnostically relevant regions of interest (ROIs) from tracking records using only pathologists' actions as they viewed biopsy specimens in the whole slide digital imaging format (zooming, panning, and fixating). We use these extracted regions in a visual bag-of-words model based on color and texture features to predict diagnostically relevant ROIs on whole slide images. Using a logistic regression classifier in a cross-validation setting on 240 digital breast biopsy slides and viewport tracking logs of three expert pathologists, we produce probability maps that show 74 % overlap with the actual regions at which pathologists looked. We compare different bag-of-words models by changing dictionary size, visual word definition (patches vs. superpixels), and training data (automatically extracted ROIs vs. manually marked ROIs). This study is a first step in understanding the scanning behaviors of pathologists and the underlying reasons for diagnostic errors.

Association between mammographic breast density and histologic features of benign breast disease

BACKGROUND

Over 40% of women undergoing breast screening have mammographically dense breasts. Elevated mammographic breast density (MBD) is an established breast cancer risk factor and is known to mask tumors within the dense tissue. However, the association of MBD with high risk benign breast disease (BBD) is unknown.

METHOD

We analyzed data for 3400 women diagnosed with pathologically confirmed BBD in the Mayo Clinic BBD cohort from 1985-2001, with a clinical MBD measure (either parenchymal pattern (PP) or Breast Imaging Reporting and Data Systems (BI-RADS) density) and expert pathology review. Risk factor information was collected from medical records and questionnaires. MBD was dichotomized as dense (PP classification P2 or DY, or BI-RADS classification c or d) or non-dense (PP classification N1 or P1, or BI-RADS classification a or b). Associations of clinical and histologic characteristics with MBD were examined using logistic regression analysis to estimate odds ratios (ORs) with 95% confidence intervals (CIs).

RESULTS

Of 3400 women in the study, 2163 (64%) had dense breasts. Adjusting for age and body mass index (BMI), there were positive associations of dense breasts with use of hormone therapy (HT), lack of lobular involution, presence of atypical lobular hyperplasia (ALH), histologic fibrosis, columnar cell hyperplasia/flat epithelia atypia (CCH/FEA), sclerosing adenosis (SA), cyst, usual ductal hyperplasia, and calcifications. In fully adjusted multivariate models, HT (1.3, 95% CI 1.1-1.5), ALH (1.5, 95% CI 1.0-2.2), lack of lobular involution (OR 1.6, 95% CI 1.2-2.1, compared to complete involution), fibrosis (OR 2.2, 95% CI 1.9-2.6) and CCH/FEA (OR 1.3, 95% CI 1.0-1.6) remained significantly associated with high MBD.

CONCLUSION

Our findings support an association between high risk BBD and high MBD, suggesting that risks associated with the latter may act early in breast carcinogenesis.

Nanoscale imaging of clinical specimens using pathology-optimized expansion microscopy

Expansion microscopy (ExM), a method for improving the resolution of light microscopy by physically expanding the specimen, has not been applied to clinical tissue samples. Here we report a clinically optimized form of ExM that supports nanoscale imaging of human tissue specimens that have been fixed with formalin, embedded in paraffin, stained with hematoxylin and eosin (H&E), and/or fresh frozen. The method, which we call expansion pathology (ExPath), converts clinical samples into an ExM-compatible state, then applies an ExM protocol with protein anchoring and mechanical homogenization steps optimized for clinical samples. ExPath enables ~70 nm resolution imaging of diverse biomolecules in intact tissues using conventional diffraction-limited microscopes, and standard antibody and fluorescent DNA in situ hybridization reagents. We use ExPath for optical diagnosis of kidney minimal-change disease, which previously required electron microscopy (EM), and demonstrate high-fidelity computational discrimination between early breast neoplastic lesions that to date have challenged human judgment. ExPath may enable the routine use of nanoscale imaging in pathology and clinical research.

The diagnostic challenge of low-grade ductal carcinoma in situ

BACKGROUND

Diagnostic agreement among pathologists is 84% for ductal carcinoma in situ (DCIS). Studies of interpretive variation according to grade are limited.

METHODS

A national sample of 115 pathologists interpreted 240 breast pathology test set cases in the Breast Pathology Study and their interpretations were compared to expert consensus interpretations. We assessed agreement of pathologists' interpretations with a consensus reference diagnosis of DCIS dichotomised into low- and high-grade lesions. Generalised estimating equations were used in logistic regression models of rates of under- and over-interpretation of DCIS by grade.

RESULTS

We evaluated 2097 independent interpretations of DCIS (512 low-grade DCIS and 1585 high-grade DCIS). Agreement with reference diagnoses was 46% (95% confidence interval [CI] 42-51) for low-grade DCIS and 83% (95% CI 81-86) for high-grade DCIS. The proportion of reference low-grade DCIS interpretations over-interpreted by pathologists (i.e. categorised as either high-grade DCIS or invasive cancer) was 23% (95% CI 19-28); 30% (95% CI 26-34) were interpreted as a lower diagnostic category (atypia or benign proliferative). Reference high-grade DCIS was under-interpreted in 14% (95% CI 12-16) of observations and only over-interpreted 3% (95% CI 2-4).

CONCLUSION

Grade is a major factor when examining pathologists' variability in diagnosing DCIS, with much lower agreement for low-grade DCIS cases compared to high-grade. These findings support the hypothesis that low-grade DCIS poses a greater interpretive challenge than high-grade DCIS, which should be considered when developing DCIS management strategies.

Surgical implications and variability in the use of the flat epithelial atypia diagnosis on breast biopsy specimens

OBJECTIVES

Flat epithelial atypia (FEA) is a relatively new diagnostic term with uncertain clinical significance for surgical management. Any implied risk of invasive breast cancer associated with FEA is contingent upon diagnostic reproducibility, yet little is known regarding its use.

MATERIALS AND METHODS

Pathologists in the Breast Pathology Study interpreted one of four 60-case test sets, one slide per case, constructed from 240 breast biopsy specimens. An electronic data form with standardized diagnostic categories was used; participants were instructed to indicate all diagnoses present. We assessed participants' use of FEA as a diagnostic term within: 1) each test set; 2) 72 cases classified by reference as benign without FEA; and 3) six cases classified by reference as FEA. 115 pathologists participated, providing 6900 total independent assessments.

RESULTS

Notation of FEA ranged from 0% to 35% of the cases interpreted, with most pathologists noting FEA on 4 or more test cases. At least one participant noted FEA in 34 of the 72 benign non-FEA cases. For the 6 reference FEA cases, participant agreement with the case reference FEA diagnosis ranged from 17% to 52%; diagnoses noted by participating pathologists for these FEA cases included columnar cell hyperplasia, usual ductal hyperplasia, atypical lobular hyperplasia, and atypical ductal hyperplasia.

CONCLUSIONS

We observed wide variation in the diagnosis of FEA among U.S. pathologists. This suggests that perceptions of diagnostic criteria and any implied risk associated with FEA may also vary. Surgical excision following a core biopsy diagnosis of FEA should be reconsidered and studied further.

Diagnostic Reproducibility: What Happens When the Same Pathologist Interprets the Same Breast Biopsy Specimen at Two Points in Time?

BACKGROUND

Surgeons may receive a different diagnosis when a breast biopsy is interpreted by a second pathologist. The extent to which diagnostic agreement by the same pathologist varies at two time points is unknown.

METHODS

Pathologists from eight U.S. states independently interpreted 60 breast specimens, one glass slide per case, on two occasions separated by ≥9 months. Reproducibility was assessed by comparing interpretations between the two time points; associations between reproducibility (intraobserver agreement rates); and characteristics of pathologists and cases were determined and also compared with interobserver agreement of baseline interpretations.

RESULTS

Sixty-five percent of invited, responding pathologists were eligible and consented; 49 interpreted glass slides in both study phases, resulting in 2940 interpretations. Intraobserver agreement rates between the two phases were 92% [95% confidence interval (CI) 88-95] for invasive breast cancer, 84% (95% CI 81-87) for ductal carcinoma-in-situ, 53% (95% CI 47-59) for atypia, and 84% (95% CI 81-86) for benign without atypia. When comparing all study participants' case interpretations at baseline, interobserver agreement rates were 89% (95% CI 84-92) for invasive cancer, 79% (95% CI 76-81) for ductal carcinoma-in-situ, 43% (95% CI 41-45) for atypia, and 77% (95% CI 74-79) for benign without atypia.

CONCLUSIONS

Interpretive agreement between two time points by the same individual pathologist was low for atypia and was similar to observed rates of agreement for atypia between different pathologists. Physicians and patients should be aware of the diagnostic challenges associated with a breast biopsy diagnosis of atypia when considering treatment and surveillance decisions.

A Randomized Study Comparing Digital Imaging to Traditional Glass Slide Microscopy for Breast Biopsy and Cancer Diagnosis

Background:

Digital whole slide imaging may be useful for obtaining second opinions and is used in many countries. However, the U.S. Food and Drug Administration requires verification studies.

Methods:

Pathologists were randomized to interpret one of four sets of breast biopsy cases during two phases, separated by ≥9 months, using glass slides or digital format (sixty cases per set, one slide per case, n = 240 cases). Accuracy was assessed by comparing interpretations to a consensus reference standard. Intraobserver reproducibility was assessed by comparing the agreement of interpretations on the same cases between two phases. Estimated probabilities of confirmation by a reference panel (i.e., predictive values) were obtained by incorporating data on the population prevalence of diagnoses.

Results:

Sixty-five percent of responding pathologists were eligible, and 252 consented to randomization; 208 completed Phase I (115 glass, 93 digital); and 172 completed Phase II (86 glass, 86 digital). Accuracy was slightly higher using glass compared to digital format and varied by category: invasive carcinoma, 96% versus 93% (P = 0.04); ductal carcinoma in situ (DCIS), 84% versus 79% (P < 0.01); atypia, 48% versus 43% (P = 0.08); and benign without atypia, 87% versus 82% (P < 0.01). There was a small decrease in intraobserver agreement when the format changed compared to when glass slides were used in both phases (P = 0.08). Predictive values for confirmation by a reference panel using glass versus digital were: invasive carcinoma, 98% and 97% (not significant [NS]); DCIS, 70% and 57% (P = 0.007); atypia, 38% and 28% (P = 0.002); and benign without atypia, 97% and 96% (NS).

Conclusions:

In this large randomized study, digital format interpretations were similar to glass slide interpretations of benign and invasive cancer cases. However, cases in the middle of the spectrum, where more inherent variability exists, may be more problematic in digital format. Future studies evaluating the effect these findings exert on clinical practice and patient outcomes are required.

High-selectivity cytology via lab-on-a-disc western blotting of individual cells

Cytology of sparingly available cell samples from both clinical and experimental settings would benefit from high-selectivity protein tools. To minimize cell handling losses in sparse samples, we design a multi-stage assay using a lab-on-a-disc that integrates cell handling and subsequent single-cell western blotting (scWestern). As the two-layer microfluidic device rotates, the induced centrifugal force directs dissociated cells to dams, which in turn localize the cells over microwells. Cells then sediment into the microwells, where the cells are lysed and subjected to scWestern. Taking into account cell losses from loading, centrifugation, and lysis-buffer exchange, our lab-on-a-disc device handles cell samples with as few as 200 cells with 75% cell settling efficiencies. Over 70% of microwells contain single cells after the centrifugation. In addition to cell settling efficiency, cell-size filtration from a mixed population of two cell lines is also realized by tuning the cell time-of-flight during centrifugation (58.4% settling efficiency with 6.4% impurity). Following the upstream cell handling, scWestern analysis detects four proteins (GFP, β-TUB, GAPDH, and STAT3) in a glioblastoma cell line. By integrating the lab-on-a-disc cell preparation and scWestern analysis, our platform measures proteins from sparse cell samples at single-cell resolution.

The BIG 2.04 MRC/EORTC SUPREMO Trial: pathology quality assurance of a large phase 3 randomised international clinical trial of postmastectomy radiotherapy in intermediate-risk breast cancer

Introduction

SUPREMO is a phase 3 randomised trial evaluating radiotherapy post-mastectomy for intermediate-risk breast cancer. 1688 patients were enrolled from 16 countries between 2006 and 2013. We report the results of central pathology review carried out for quality assurance.

Patients and methods

A single recut haematoxylin and eosin (H&E) tumour section was assessed by one of two reviewing pathologists, blinded to the originally reported pathology and patient data. Tumour type, grade and lymphovascular invasion were reviewed to assess if they met the inclusion criteria. Slides from potentially ineligible patients on central review were scanned and reviewed online together by the two pathologists and a consensus reached. A subset of 25 of these cases was double-reported independently by the pathologists prior to the online assessment.

Results

The major contributors to the trial were the UK (75%) and the Netherlands (10%). There is a striking difference in lymphovascular invasion (LVi) rates (41.6 vs. 15.1% (UK); p = <0.0001) and proportions of grade 3 carcinomas (54.0 vs. 42.0% (UK); p = <0.0001) on comparing local reporting with central review. There was no difference in the locally reported frequency of LVi rates in node-positive (N+) and node-negative (N−) subgroups (40.3 vs. 38.0%; p = 0.40) but a significant difference in the reviewed frequency (16.9 vs. 9.9%; p = 0.004). Of the N− cases, 104 (25.1%) would have been ineligible by initial central review by virtue of grade and/or lymphovascular invasion status. Following online consensus review, this fell to 70 cases (16.3% of N− cases, 4.1% of all cases).

Conclusions

These data have important implications for the design, powering and interpretation of outcomes from this and future clinical trials. If critical pathology criteria are determinants for trial entry, serious consideration should be given to up-front central pathology review.

Electronic supplementary material

The online version of this article (doi:10.1007/s10549-017-4145-4) contains supplementary material, which is available to authorized users.

Accuracy is in the eyes of the pathologist: The visual interpretive process and diagnostic accuracy with digital whole slide images

Digital whole slide imaging is an increasingly common medium in pathology, with application to education, telemedicine, and rendering second opinions. It has also made it possible to use eye tracking devices to explore the dynamic visual inspection and interpretation of histopathological features of tissue while pathologists review cases. Using whole slide images, the present study examined how a pathologist's diagnosis is influenced by fixed case-level factors, their prior clinical experience, and their patterns of visual inspection. Participating pathologists interpreted one of two test sets, each containing 12 digital whole slide images of breast biopsy specimens. Cases represented four diagnostic categories as determined via expert consensus: benign without atypia, atypia, ductal carcinoma in situ (DCIS), and invasive cancer. Each case included one or more regions of interest (ROIs) previously determined as of critical diagnostic importance. During pathologist interpretation we tracked eye movements, viewer tool behavior (zooming, panning), and interpretation time. Models were built using logistic and linear regression with generalized estimating equations, testing whether variables at the level of the pathologists, cases, and visual interpretive behavior would independently and/or interactively predict diagnostic accuracy and efficiency. Diagnostic accuracy varied as a function of case consensus diagnosis, replicating earlier research. As would be expected, benign cases tended to elicit false positives, and atypia, DCIS, and invasive cases tended to elicit false negatives. Pathologist experience levels, case consensus diagnosis, case difficulty, eye fixation durations, and the extent to which pathologists' eyes fixated within versus outside of diagnostic ROIs, all independently or interactively predicted diagnostic accuracy. Higher zooming behavior predicted a tendency to over-interpret benign and atypia cases, but not DCIS cases. Efficiency was not predicted by pathologist- or visual search-level variables. Results provide new insights into the medical interpretive process and demonstrate the complex interactions between pathologists and cases that guide diagnostic decision-making. Implications for training, clinical practice, and computer-aided decision aids are considered.

Histological features associated with diagnostic agreement in atypical ductal hyperplasia of the breast: illustrative cases from the B-Path study

AIMS

This study examined the case-specific characteristics associated with interobserver diagnostic agreement in atypical ductal hyperplasia (ADH) of the breast.

METHODS AND RESULTS

Seventy-two test set cases with a consensus diagnosis of ADH from the B-Path study were evaluated. Cases were scored for 17 histological features, which were then correlated with the participant agreement with the consensus ADH diagnosis. Participating pathologists' perceptions of case difficulty, borderline features or whether they would obtain a second opinion were also examined for associations with agreement. Of the 2070 participant interpretations of the 72 consensus ADH cases, 48% were scored by participants as difficult and 45% as borderline between two diagnoses; the presence of both of these features was significantly associated with increased agreement (P < 0.001). A second opinion would have been obtained in 80% of interpretations, and this was associated with increased agreement (P < 0.001). Diagnostic agreement ranged from 10% to 89% on a case-by-case basis. Cases with papillary lesions, cribriform architecture and obvious cytological monotony were associated with higher agreement. Lower agreement rates were associated with solid or micropapillary architecture, borderline cytological monotony, or cases without a diagnostic area that was obvious on low power.

CONCLUSIONS

The results of this study suggest that pathologists frequently recognize the challenge of ADH cases, with some cases being more prone to diagnostic variability. In addition, there are specific histological features associated with diagnostic agreement on ADH cases. Multiple example images from cases in this test set are provided to serve as educational illustrations of these challenges.