Precision histology: how deep learning is poised to revitalize histomorphology for personalized cancer care. NPJ Precis Oncol. 2017;1:22. [PMID: 29872706 PMCID: PMC5871847 DOI: 10.1038/s41698-017-0022-1]

Non-invasive screening of bladder cancer using digital microfluidics and FLIM technology combined with deep learning

Non-invasive screening for bladder cancer is crucial for treatment and postoperative follow-up. This study combines digital microfluidics (DMF) technology with fluorescence lifetime imaging microscopy (FLIM) for urine analysis and introduces a novel non-invasive bladder cancer screening technique. Initially, the DMF was utilized to perform preliminary screening and enrichment of urine exfoliated cells from 54 participants, followed by cell staining and FLIM analysis to assess the viscosity of the intracellular microenvironment. Subsequently, a deep learning residual convolutional neural network was employed to automatically classify FLIM images, achieving a three-class prediction of high-risk (malignant), low-risk (benign), and minimal risk (normal) categories. The results demonstrated a high consistency with pathological diagnosis, with an accuracy of 91% and a precision of 93%. Notably, the method is sensitive for both high-grade and low-grade bladder cancer cases. This highly accurate non-invasive screening method presents a promising approach for bladder cancer screening with significant clinical application potential.

Mitosis detection, fast and slow: Robust and efficient detection of mitotic figures

Counting of mitotic figures is a fundamental step in grading and prognostication of several cancers. However, manual mitosis counting is tedious and time-consuming. In addition, variation in the appearance of mitotic figures causes a high degree of discordance among pathologists. With advances in deep learning models, several automatic mitosis detection algorithms have been proposed but they are sensitive to domain shift often seen in histology images. We propose a robust and efficient two-stage mitosis detection framework, which comprises mitosis candidate segmentation (Detecting Fast) and candidate refinement (Detecting Slow) stages. The proposed candidate segmentation model, termed EUNet, is fast and accurate due to its architectural design. EUNet can precisely segment candidates at a lower resolution to considerably speed up candidate detection. Candidates are then refined using a deeper classifier network, EfficientNet-B7, in the second stage. We make sure both stages are robust against domain shift by incorporating domain generalization methods. We demonstrate state-of-the-art performance and generalizability of the proposed model on the three largest publicly available mitosis datasets, winning the two mitosis domain generalization challenge contests (MIDOG21 and MIDOG22). Finally, we showcase the utility of the proposed algorithm by processing the TCGA breast cancer cohort (1,124 whole-slide images) to generate and release a repository of more than 620K potential mitotic figures (not exhaustively validated).

Deep learning in cancer genomics and histopathology

Histopathology and genomic profiling are cornerstones of precision oncology and are routinely obtained for patients with cancer. Traditionally, histopathology slides are manually reviewed by highly trained pathologists. Genomic data, on the other hand, is evaluated by engineered computational pipelines. In both applications, the advent of modern artificial intelligence methods, specifically machine learning (ML) and deep learning (DL), have opened up a fundamentally new way of extracting actionable insights from raw data, which could augment and potentially replace some aspects of traditional evaluation workflows. In this review, we summarize current and emerging applications of DL in histopathology and genomics, including basic diagnostic as well as advanced prognostic tasks. Based on a growing body of evidence, we suggest that DL could be the groundwork for a new kind of workflow in oncology and cancer research. However, we also point out that DL models can have biases and other flaws that users in healthcare and research need to know about, and we propose ways to address them.

Data Privacy in Healthcare: In the Era of Artificial Intelligence

Data Privacy has increasingly become a matter of concern in the era of large public digital respositories of data. This is particularly true in healthcare where data can be misused if traced back to patients, and brings with itself a myriad of possibilities. Bring custodians of data, as well as being at the helm of disigning studies and products that can potentially benefit products, healthcare professionals often find themselves unsure about ethical and legal constraints that undelie data sharing. In this review we touch upon the concerns, leal frameworks as well as some common practices in these respects.

H&E image analysis pipeline for quantifying morphological features

Detecting cell types from histopathological images is essential for various digital pathology applications. However, large number of cells in whole-slide images (WSIs) necessitates automated analysis pipelines for efficient cell type detection. Herein, we present hematoxylin and eosin (H&E) Image Processing pipeline (HEIP) for automatied analysis of scanned H&E-stained slides. HEIP is a flexible and modular open-source software that performs preprocessing, instance segmentation, and nuclei feature extraction. To evaluate the performance of HEIP, we applied it to extract cell types from ovarian high-grade serous carcinoma (HGSC) patient WSIs. HEIP showed high precision in instance segmentation, particularly for neoplastic and epithelial cells. We also show that there is a significant correlation between genomic ploidy values and morphological features, such as major axis of the nucleus.

Artificial Intelligence-Based PTEN Loss Assessment as an Early Predictor of Prostate Cancer Metastasis After Surgery: A Multicenter Retrospective Study

Phosphatase and tensin homolog (PTEN) loss is associated with adverse outcomes in prostate cancer and can be measured via immunohistochemistry. The purpose of the study was to establish the clinical application of an in-house developed artificial intelligence (AI) image analysis workflow for automated detection of PTEN loss on digital images for identifying patients at risk of early recurrence and metastasis. Postsurgical tissue microarray sections from the Canary Foundation (n = 1264) stained with anti-PTEN antibody were evaluated independently by pathologist conventional visual scoring (cPTEN) and an automated AI-based image analysis pipeline (AI-PTEN). The relationship of PTEN evaluation methods with cancer recurrence and metastasis was analyzed using multivariable Cox proportional hazard and decision curve models. Both cPTEN scoring by the pathologist and quantification of PTEN loss by AI (high-risk AI-qPTEN) were significantly associated with shorter metastasis-free survival (MFS) in univariable analysis (cPTEN hazard ratio [HR], 1.54; CI, 1.07-2.21; P = .019; AI-qPTEN HR, 2.55; CI, 1.83-3.56; P < .001). In multivariable analyses, AI-qPTEN showed a statistically significant association with shorter MFS (HR, 2.17; CI, 1.49-3.17; P < .001) and recurrence-free survival (HR, 1.36; CI, 1.06-1.75; P = .016) when adjusting for relevant postsurgical clinical nomogram (Cancer of the Prostate Risk Assessment [CAPRA] postsurgical score [CAPRA-S]), whereas cPTEN does not show a statistically significant association (HR, 1.33; CI, 0.89-2; P = .2 and HR, 1.26; CI, 0.99-1.62; P = .063, respectively) when adjusting for CAPRA-S risk stratification. More importantly, AI-qPTEN was associated with shorter MFS in patients with favorable pathological stage and negative surgical margins (HR, 2.72; CI, 1.46-5.06; P = .002). Workflow also demonstrated enhanced clinical utility in decision curve analysis, more accurately identifying men who might benefit from adjuvant therapy postsurgery. This study demonstrates the clinical value of an affordable and fully automated AI-powered PTEN assessment for evaluating the risk of developing metastasis or disease recurrence after radical prostatectomy. Adding the AI-qPTEN assessment workflow to clinical variables may affect postoperative surveillance or management options, particularly in low-risk patients.

Computerized Diagnosis of Liver Tumors From CT Scans Using a Deep Neural Network Approach

The liver is a frequent site of benign and malignant, primary and metastatic tumors. Hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (ICC) are the most common primary liver cancers, and colorectal liver metastasis (CRLM) is the most common secondary liver cancer. Although the imaging characteristic of these tumors is central to optimal clinical management, it relies on imaging features that are often non-specific, overlap, and are subject to inter-observer variability. Thus, in this study, we aimed to categorize liver tumors automatically from CT scans using a deep learning approach that objectively extracts discriminating features not visible to the naked eye. Specifically, we used a modified Inception v3 network-based classification model to classify HCC, ICC, CRLM, and benign tumors from pretreatment portal venous phase computed tomography (CT) scans. Using a multi-institutional dataset of 814 patients, this method achieved an overall accuracy rate of 96%, with sensitivity rates of 96%, 94%, 99%, and 86% for HCC, ICC, CRLM, and benign tumors, respectively, using an independent dataset. These results demonstrate the feasibility of the proposed computer-assisted system as a novel non-invasive diagnostic tool to classify the most common liver tumors objectively.

Calibration and Uncertainty in Neural Time-to-Event Modeling

Models for predicting the time of a future event are crucial for risk assessment, across a diverse range of applications. Existing time-to-event (survival) models have focused primarily on preserving pairwise ordering of estimated event times (i.e., relative risk). We propose neural time-to-event models that account for calibration and uncertainty while predicting accurate absolute event times. Specifically, an adversarial nonparametric model is introduced for estimating matched time-to-event distributions for probabilistically concentrated and accurate predictions. We also consider replacing the discriminator of the adversarial nonparametric model with a survival-function matching estimator that accounts for model calibration. The proposed estimator can be used as a means of estimating and comparing conditional survival distributions while accounting for the predictive uncertainty of probabilistic models. Extensive experiments show that the distribution matching methods outperform existing approaches in terms of both calibration and concentration of time-to-event distributions.

QuPath Analysis for CD30+ Cutaneous T-Cell Lymphoma

BACKGROUND

Mycosis fungoides is the most common subtype of cutaneous T-cell lymphoma, in which the expression of cluster of differentiation 30 (CD30)+ subtype can now be treated with the CD30 antibody conjugate brentuximab vedotin. Diagnostic methods are based on immunohistochemical (IHC) staining followed by manual assessment by pathologists, which is always a subjective calculation. QuPath, an open-source software for digital pathology image analysis, satisfies the requirements of objective approaches.

METHODS

Ten samples from mycosis fungoides patients with CD30 expression at different stages were stained for CD3 and CD30 by IHC staining, scanned, and quantitative analysis was performed using QuPath (version 2.1). Each slide was independently assessed by 3 board-certified dermatopathologists.

RESULTS

Individual estimates for CD30+/CD3+ cells varied among the individual histopathologists (mean coefficient of variation, 0.46; range, 0-0.78). QuPath analysis showed excellent separation between the positively stained cells for CD3 and CD30 IHC and other cells and tissue structures, although the results correlated strongly with the respective mean estimates of the 3 histopathologists (Pearson-R 0.93).

CONCLUSIONS

The results show a high interobserver variability evaluation of IHC markers, although quantitative image analysis offer a significant advantage for comparison. This is not only relevant for clinical routine but also especially critical in therapeutic studies addressing targeted molecules.

Artificial Intelligence You Can Trust: What Matters Beyond Performance When Applying Artificial Intelligence to Renal Histopathology?

Although still in its infancy, artificial intelligence (AI) analysis of kidney biopsy images is anticipated to become an integral aspect of renal histopathology. As these systems are developed, the focus will understandably be on developing ever more accurate models, but successful translation to the clinic will also depend upon other characteristics of the system.In the extreme, deployment of highly performant but "black box" AI is fraught with risk, and high-profile errors could damage future trust in the technology. Furthermore, a major factor determining whether new systems are adopted in clinical settings is whether they are "trusted" by clinicians. Key to unlocking trust will be designing platforms optimized for intuitive human-AI interactions and ensuring that, where judgment is required to resolve ambiguous areas of assessment, the workings of the AI image classifier are understandable to the human observer. Therefore, determining the optimal design for AI systems depends on factors beyond performance, with considerations of goals, interpretability, and safety constraining many design and engineering choices.In this article, we explore challenges that arise in the application of AI to renal histopathology, and consider areas where choices around model architecture, training strategy, and workflow design may be influenced by factors beyond the final performance metrics of the system.

A dual decoder U-Net-based model for nuclei instance segmentation in hematoxylin and eosin-stained histological images

Even in the era of precision medicine, with various molecular tests based on omics technologies available to improve the diagnosis process, microscopic analysis of images derived from stained tissue sections remains crucial for diagnostic and treatment decisions. Among other cellular features, both nuclei number and shape provide essential diagnostic information. With the advent of digital pathology and emerging computerized methods to analyze the digitized images, nuclei detection, their instance segmentation and classification can be performed automatically. These computerized methods support human experts and allow for faster and more objective image analysis. While methods ranging from conventional image processing techniques to machine learning-based algorithms have been proposed, supervised convolutional neural network (CNN)-based techniques have delivered the best results. In this paper, we propose a CNN-based dual decoder U-Net-based model to perform nuclei instance segmentation in hematoxylin and eosin (H&E)-stained histological images. While the encoder path of the model is developed to perform standard feature extraction, the two decoder heads are designed to predict the foreground and distance maps of all nuclei. The outputs of the two decoder branches are then merged through a watershed algorithm, followed by post-processing refinements to generate the final instance segmentation results. Moreover, to additionally perform nuclei classification, we develop an independent U-Net-based model to classify the nuclei predicted by the dual decoder model. When applied to three publicly available datasets, our method achieves excellent segmentation performance, leading to average panoptic quality values of 50.8%, 51.3%, and 62.1% for the CryoNuSeg, NuInsSeg, and MoNuSAC datasets, respectively. Moreover, our model is the top-ranked method in the MoNuSAC post-challenge leaderboard.

K-RAS Associated Gene-Mutation-Based Algorithm for Prediction of Treatment Response of Patients with Subtypes of Breast Cancer and Especially Triple-Negative Cancer

Purpose: There is an urgent need for developing new biomarker tools to accurately predict treatment response of breast cancer, especially the deadly triple-negative breast cancer. We aimed to develop gene-mutation-based machine learning (ML) algorithms as biomarker classifiers to predict treatment response of first-line chemotherapy with high precision. Methods: Random Forest ML was applied to screen the algorithms of various combinations of gene mutation profiles of primary tumors at diagnosis using a TCGA Cohort (n = 399) with up to 150 months follow-up as a training set and validated in a MSK Cohort (n = 807) with up to 220 months follow-up. Subtypes of breast cancer including triple-negative and luminal A (ER+, PR+ and HER2−) were also assessed. The predictive performance of the candidate algorithms as classifiers was further assessed using logistic regression, Kaplan−Meier progression-free survival (PFS) plot, and univariate/multivariate Cox proportional hazard regression analyses. Results: A novel algorithm termed the 12-Gene Algorithm based on mutation profiles of KRAS, PIK3CA, MAP3K1, MAP2K4, PTEN, TP53, CDH1, GATA3, KMT2C, ARID1A, RunX1, and ESR1, was identified. The performance of this algorithm to distinguish non-progressed (responder) vs. progressed (non-responder) to treatment in the TCGA Cohort as determined using AUC was 0.96 (95% CI 0.94−0.98). It predicted progression-free survival (PFS) with hazard ratio (HR) of 21.6 (95% CI 11.3−41.5) (p < 0.0001) in all patients. The algorithm predicted PFS in the triple-negative subgroup with HR of 19.3 (95% CI 3.7−101.3) (n = 42, p = 0.000). The 12-Gene Algorithm was validated in the MSK Cohort with a similar AUC of 0.97 (95% CI 0.96−0.98) to distinguish responder vs. non-responder patients, and had a HR of 18.6 (95% CI 4.4−79.2) to predict PFS in the triple-negative subgroup (n = 75, p < 0.0001). Conclusions: The novel 12-Gene algorithm based on multitude gene-mutation profiles identified through ML has a potential to predict breast cancer treatment response to therapies, especially in triple-negative subgroups patients, which may assist personalized therapies and reduce mortality.

Deep learning-based tumor microenvironment segmentation is predictive of tumor mutations and patient survival in non-small-cell lung cancer

BACKGROUND

Despite the fact that tumor microenvironment (TME) and gene mutations are the main determinants of progression of the deadliest cancer in the world - lung cancer, their interrelations are not well understood. Digital pathology data provides a unique insight into the spatial composition of the TME. Various spatial metrics and machine learning approaches were proposed for prediction of either patient survival or gene mutations from this data. Still, these approaches are limited in the scope of analyzed features and in their explainability, and as such fail to transfer to clinical practice.

METHODS

Here, we generated 23,199 image patches from 26 hematoxylin-and-eosin (H&E)-stained lung cancer tissue sections and annotated them into 9 different tissue classes. Using this dataset, we trained a deep neural network ARA-CNN. Next, we applied the trained network to segment 467 lung cancer H&E images from The Cancer Genome Atlas (TCGA) database. We used the segmented images to compute human-interpretable features reflecting the heterogeneous composition of the TME, and successfully utilized them to predict patient survival and cancer gene mutations.

RESULTS

We achieved per-class AUC ranging from 0.72 to 0.99 for classifying tissue types in lung cancer with ARA-CNN. Machine learning models trained on the proposed human-interpretable features achieved a c-index of 0.723 in the task of survival prediction and AUC up to 73.5% for PDGFRB in the task of mutation classification.

CONCLUSIONS

We presented a framework that accurately predicted survival and gene mutations in lung adenocarcinoma patients based on human-interpretable features extracted from H&E slides. Our approach can provide important insights for designing novel cancer treatments, by linking the spatial structure of the TME in lung adenocarcinoma to gene mutations and patient survival. It can also expand our understanding of the effects that the TME has on tumor evolutionary processes. Our approach can be generalized to different cancer types to inform precision medicine strategies.

A Deep Multi-Label Segmentation Network For Eosinophilic Esophagitis Whole Slide Biopsy Diagnostics

Eosinophilic esophagitis (EoE) is an allergic inflammatory condition of the esophagus associated with elevated numbers of eosinophils. Disease diagnosis and monitoring require determining the concentration of eosinophils in esophageal biopsies, a time-consuming, tedious and somewhat subjective task currently performed by pathologists. Here, we developed a machine learning pipeline to identify, quantitate and diagnose EoE patients' at the whole slide image level. We propose a platform that combines multi-label segmentation deep network decision support system with dynamics convolution that is able to process whole biopsy slide. Our network is able to segment both intact and not-intact eosinophils with a mean intersection over union (mIoU) of 0.93. This segmentation enables the local quantification of intact eosinophils with a mean absolute error of 0.611 eosinophils. We examined a cohort of 1066 whole slide images from 400 patients derived from multiple institutions. Using this set, our model achieved a global accuracy of 94.75%, sensitivity of 94.13%, and specificity of 95.25% in reporting EoE disease activity. Our work provides state-of-the-art performances on the largest EoE cohort to date, and successfully addresses two of the main challenges in EoE diagnostics and digital pathology, the need to detect several types of small features simultaneously, and the ability to analyze whole slides efficiently. Our results pave the way for an automated diagnosis of EoE and can be utilized for other conditions with similar challenges.

Recent Applications of Artificial Intelligence from Histopathologic Image-Based Prediction of Microsatellite Instability in Solid Cancers: A Systematic Review

Cancers with high microsatellite instability (MSI-H) have a better prognosis and respond well to immunotherapy. However, MSI is not tested in all cancers because of the additional costs and time of diagnosis. Therefore, artificial intelligence (AI)-based models have been recently developed to evaluate MSI from whole slide images (WSIs). Here, we aimed to assess the current state of AI application to predict MSI based on WSIs analysis in MSI-related cancers and suggest a better study design for future studies. Studies were searched in online databases and screened by reference type, and only the full texts of eligible studies were reviewed. The included 14 studies were published between 2018 and 2021, and most of the publications were from developed countries. The commonly used dataset is The Cancer Genome Atlas dataset. Colorectal cancer (CRC) was the most common type of cancer studied, followed by endometrial, gastric, and ovarian cancers. The AI models have shown the potential to predict MSI with the highest AUC of 0.93 in the case of CRC. The relatively limited scale of datasets and lack of external validation were the limitations of most studies. Future studies with larger datasets are required to implicate AI models in routine diagnostic practice for MSI prediction.

Molecular Pathology of Lung Cancer

This overview of the molecular pathology of lung cancer includes a review of the most salient molecular alterations of the genome, transcriptome, and the epigenome. The insights provided by the growing use of next-generation sequencing (NGS) in lung cancer will be discussed, and interrelated concepts such as intertumor heterogeneity, intratumor heterogeneity, tumor mutational burden, and the advent of liquid biopsy will be explored. Moreover, this work describes how the evolving field of molecular pathology refines the understanding of different histologic phenotypes of non-small-cell lung cancer (NSCLC) and the underlying biology of small-cell lung cancer. This review will provide an appreciation for how ongoing scientific findings and technologic advances in molecular pathology are crucial for development of biomarkers, therapeutic agents, clinical trials, and ultimately improved patient care.

Artificial Intelligence for Predicting Microsatellite Instability Based on Tumor Histomorphology: A Systematic Review

Microsatellite instability (MSI)/defective DNA mismatch repair (dMMR) is receiving more attention as a biomarker for eligibility for immune checkpoint inhibitors in advanced diseases. However, due to high costs and resource limitations, MSI/dMMR testing is not widely performed. Some attempts are in progress to predict MSI/dMMR status through histomorphological features on H&E slides using artificial intelligence (AI) technology. In this study, the potential predictive role of this new methodology was reviewed through a systematic review. Studies up to September 2021 were searched through PubMed and Embase database searches. The design and results of each study were summarized, and the risk of bias for each study was evaluated. For colorectal cancer, AI-based systems showed excellent performance with the highest standard of 0.972; for gastric and endometrial cancers they showed a relatively low but satisfactory performance, with the highest standard of 0.81 and 0.82, respectively. However, analyzing the risk of bias, most studies were evaluated at high-risk. AI-based systems showed a high potential in predicting the MSI/dMMR status of different cancer types, and particularly of colorectal cancers. Therefore, a confirmation test should be required only for the results that are positive in the AI test.

The Digital Brain Tumour Atlas, an open histopathology resource

Currently, approximately 150 different brain tumour types are defined by the WHO. Recent endeavours to exploit machine learning and deep learning methods for supporting more precise diagnostics based on the histological tumour appearance have been hampered by the relative paucity of accessible digital histopathological datasets. While freely available datasets are relatively common in many medical specialties such as radiology and genomic medicine, there is still an unmet need regarding histopathological data. Thus, we digitized a significant portion of a large dedicated brain tumour bank based at the Division of Neuropathology and Neurochemistry of the Medical University of Vienna, covering brain tumour cases from 1995-2019. A total of 3,115 slides of 126 brain tumour types (including 47 control tissue slides) have been scanned. Additionally, complementary clinical annotations have been collected for each case. In the present manuscript, we thoroughly discuss this unique dataset and make it publicly available for potential use cases in machine learning and digital image analysis, teaching and as a reference for external validation.

Development of Planar Illumination Strategies for Solving Mysteries in the Sub-Cellular Realm

Optical microscopy has vastly expanded the frontiers of structural and functional biology, due to the non-invasive probing of dynamic volumes in vivo. However, traditional widefield microscopy illuminating the entire field of view (FOV) is adversely affected by out-of-focus light scatter. Consequently, standard upright or inverted microscopes are inept in sampling diffraction-limited volumes smaller than the optical system's point spread function (PSF). Over the last few decades, several planar and structured (sinusoidal) illumination modalities have offered unprecedented access to sub-cellular organelles and 4D (3D + time) image acquisition. Furthermore, these optical sectioning systems remain unaffected by the size of biological samples, providing high signal-to-noise (SNR) ratios for objective lenses (OLs) with long working distances (WDs). This review aims to guide biologists regarding planar illumination strategies, capable of harnessing sub-micron spatial resolution with a millimeter depth of penetration.

Code-Free Development and Deployment of Deep Segmentation Models for Digital Pathology

Application of deep learning on histopathological whole slide images (WSIs) holds promise of improving diagnostic efficiency and reproducibility but is largely dependent on the ability to write computer code or purchase commercial solutions. We present a code-free pipeline utilizing free-to-use, open-source software (QuPath, DeepMIB, and FastPathology) for creating and deploying deep learning-based segmentation models for computational pathology. We demonstrate the pipeline on a use case of separating epithelium from stroma in colonic mucosa. A dataset of 251 annotated WSIs, comprising 140 hematoxylin-eosin (HE)-stained and 111 CD3 immunostained colon biopsy WSIs, were developed through active learning using the pipeline. On a hold-out test set of 36 HE and 21 CD3-stained WSIs a mean intersection over union score of 95.5 and 95.3% was achieved on epithelium segmentation. We demonstrate pathologist-level segmentation accuracy and clinical acceptable runtime performance and show that pathologists without programming experience can create near state-of-the-art segmentation solutions for histopathological WSIs using only free-to-use software. The study further demonstrates the strength of open-source solutions in its ability to create generalizable, open pipelines, of which trained models and predictions can seamlessly be exported in open formats and thereby used in external solutions. All scripts, trained models, a video tutorial, and the full dataset of 251 WSIs with ~31 k epithelium annotations are made openly available at https://github.com/andreped/NoCodeSeg to accelerate research in the field.

Harnessing artificial intelligence to infer novel spatial biomarkers for the diagnosis of eosinophilic esophagitis

Eosinophilic esophagitis (EoE) is a chronic allergic inflammatory condition of the esophagus associated with elevated esophageal eosinophils. Second only to gastroesophageal reflux disease, EoE is one of the leading causes of chronic refractory dysphagia in adults and children. EoE is a clinicopathologic disorder and the histological portion of the diagnosis requires enumerating the density of esophageal eosinophils in esophageal biopsies, and evaluating additional features such as basal zone hyperplasia is helpful. However, this task requires time-consuming, somewhat subjective manual analysis, thus reducing the ability to process the complex tissue structure and infer its relationship with the patient's clinical status. Previous artificial intelligence (AI) approaches that aimed to improve histology-based diagnosis focused on recapitulating identification and quantification of the area of maximal eosinophil density, the gold standard manual metric for determining EoE disease activity. However, this metric does not account for the distribution of eosinophils or other histological features, over the whole slide image. Here, we developed an artificial intelligence platform that infers local and spatial biomarkers based on semantic segmentation of intact eosinophils and basal zone distributions. Besides the maximal density of eosinophils [referred to as Peak Eosinophil Count (PEC)] and a maximal basal zone fraction, we identify the value of two additional metrics that reflect the distribution of eosinophils and basal zone fractions. This approach enables a decision support system that predicts EoE activity and potentially classifies the histological severity of EoE patients. We utilized a cohort that includes 1,066 biopsy slides from 400 subjects to validate the system's performance and achieved a histological severity classification accuracy of 86.70%, sensitivity of 84.50%, and specificity of 90.09%. Our approach highlights the importance of systematically analyzing the distribution of biopsy features over the entire slide and paves the way toward a personalized decision support system that will assist not only in counting cells but can also potentially improve diagnosis and provide treatment prediction.

Deep Learning Approaches and Applications in Toxicologic Histopathology: Current Status and Future Perspectives

Whole slide imaging enables the use of a wide array of digital image analysis tools that are revolutionizing pathology. Recent advances in digital pathology and deep convolutional neural networks have created an enormous opportunity to improve workflow efficiency, provide more quantitative, objective, and consistent assessments of pathology datasets, and develop decision support systems. Such innovations are already making their way into clinical practice. However, the progress of machine learning - in particular, deep learning (DL) - has been rather slower in nonclinical toxicology studies. Histopathology data from toxicology studies are critical during the drug development process that is required by regulatory bodies to assess drug-related toxicity in laboratory animals and its impact on human safety in clinical trials. Due to the high volume of slides routinely evaluated, low-throughput, or narrowly performing DL methods that may work well in small-scale diagnostic studies or for the identification of a single abnormality are tedious and impractical for toxicologic pathology. Furthermore, regulatory requirements around good laboratory practice are a major hurdle for the adoption of DL in toxicologic pathology. This paper reviews the major DL concepts, emerging applications, and examples of DL in toxicologic pathology image analysis. We end with a discussion of specific challenges and directions for future research.

Foundations of Machine Learning-Based Clinical Prediction Modeling: Part V-A Practical Approach to Regression Problems

This chapter goes through the steps required to train and validate a simple, machine learning-based clinical prediction model for any continuous outcome. We supply fully structured code for the readers to download and execute in parallel to this section, as well as a simulated database of 10,000 glioblastoma patients who underwent microsurgery, and predict survival from diagnosis in months. We walk the reader through each step, including import, checking, splitting of data. In terms of pre-processing, we focus on how to practically implement imputation using a k-nearest neighbor algorithm. We also illustrate how to select features based on recursive feature elimination and how to use k-fold cross validation. We demonstrate a generalized linear model, a generalized additive model, a random forest, a ridge regressor, and a Least Absolute Shrinkage and Selection Operator (LASSO) regressor. Specifically for regression, we discuss how to evaluate root mean square error (RMSE), mean average error (MAE), and the R² statistic, as well as how a quantile-quantile plot can be used to assess the performance of the regressor along the spectrum of the outcome variable, similarly to calibration when dealing with binary outcomes. Finally, we explain how to arrive at a measure of variable importance using a universal, nonparametric method.

Foundations of Machine Learning-Based Clinical Prediction Modeling: Part IV-A Practical Approach to Binary Classification Problems

We illustrate the steps required to train and validate a simple, machine learning-based clinical prediction model for any binary outcome, such as, for example, the occurrence of a complication, in the statistical programming language R. To illustrate the methods applied, we supply a simulated database of 10,000 glioblastoma patients who underwent microsurgery, and predict the occurrence of 12-month survival. We walk the reader through each step, including import, checking, and splitting of datasets. In terms of pre-processing, we focus on how to practically implement imputation using a k-nearest neighbor algorithm, and how to perform feature selection using recursive feature elimination. When it comes to training models, we apply the theory discussed in Parts I-III. We show how to implement bootstrapping and to evaluate and select models based on out-of-sample error. Specifically for classification, we discuss how to counteract class imbalance by using upsampling techniques. We discuss how the reporting of a minimum of accuracy, area under the curve (AUC), sensitivity, and specificity for discrimination, as well as slope and intercept for calibration-if possible alongside a calibration plot-is paramount. Finally, we explain how to arrive at a measure of variable importance using a universal, AUC-based method. We provide the full, structured code, as well as the complete glioblastoma survival database for the readers to download and execute in parallel to this section.

The Artificial Intelligence Doctor: Considerations for the Clinical Implementation of Ethical AI

The applications of artificial intelligence (AI) and machine learning (ML) in modern medicine are growing exponentially, and new developments are fast-paced. However, the lack of trust and appropriate legislation hinder its clinical implementation. Recently, there is a clear increase of directives and considerations on Ethical AI. However, most literature broadly deals with ethical tensions on a meta-level without offering hands-on advice in practice. In this article, we non-exhaustively cover basic practical guidelines regarding AI-specific ethical aspects, including transparency and explicability, equity and mitigation of biases, and lastly, liability.

Machine Learning Algorithms in Neuroimaging: An Overview

Machine learning (ML) and artificial intelligence (AI) applications in the field of neuroimaging have been on the rise in recent years, and their clinical adoption is increasing worldwide. Deep learning (DL) is a field of ML that can be defined as a set of algorithms enabling a computer to be fed with raw data and progressively discover-through multiple layers of representation-more complex and abstract patterns in large data sets. The combination of ML and radiomics, namely the extraction of features from medical images, has proven valuable, too: Radiomic information can be used for enhanced image characterization and prognosis or outcome prediction. This chapter summarizes the basic concepts underlying ML application for neuroimaging and discusses technical aspects of the most promising algorithms, with a specific focus on Convolutional Neural Networks (CNNs) and Generative Adversarial Networks (GANs), in order to provide the readership with the fundamental theoretical tools to better understand ML in neuroimaging. Applications are highlighted from a practical standpoint in the last section of the chapter, including: image reconstruction and restoration, image synthesis and super-resolution, registration, segmentation, classification, and outcome prediction.

Prediction of genetic alterations from gastric cancer histopathology images using a fully automated deep learning approach

BACKGROUND

Studies correlating specific genetic mutations and treatment response are ongoing to establish an effective treatment strategy for gastric cancer (GC). To facilitate this research, a cost- and time-effective method to analyze the mutational status is necessary. Deep learning (DL) has been successfully applied to analyze hematoxylin and eosin (H and E)-stained tissue slide images.

AIM

To test the feasibility of DL-based classifiers for the frequently occurring mutations from the H and E-stained GC tissue whole slide images (WSIs).

METHODS

From the GC dataset of The Cancer Genome Atlas (TCGA-STAD), wild-type/mutation classifiers for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes were trained on 360 × 360-pixel patches of tissue images.

RESULTS

The area under the curve (AUC) for the receiver operating characteristic (ROC) curves ranged from 0.727 to 0.862 for the TCGA frozen WSIs and 0.661 to 0.858 for the TCGA formalin-fixed paraffin-embedded (FFPE) WSIs. The performance of the classifier can be improved by adding new FFPE WSI training dataset from our institute. The classifiers trained for mutation prediction in colorectal cancer completely failed to predict the mutational status in GC, indicating that DL-based mutation classifiers are incompatible between different cancers.

CONCLUSION

This study concluded that DL could predict genetic mutations in H and E-stained tissue slides when they are trained with appropriate tissue data.

Neural network fast-classifies biological images through features selecting to power automated microscopy

Artificial intelligence is nowadays used for cell detection and classification in optical microscopy during post-acquisition analysis. The microscopes are now fully automated and next expected to be smart by making acquisition decisions based on the images. It calls for analysing them on the fly. Biology further imposes training on a reduced data set due to cost and time to prepare the samples and have the data sets annotated by experts. We propose a real-time image processing compliant with these specifications by balancing accurate detection and execution performance. We characterised the images using a generic, high-dimensional feature extractor. We then classified the images using machine learning to understand the contribution of each feature in decision and execution time. We found that the non-linear-classifier random forests outperformed Fisher's linear discriminant. More importantly, the most discriminant and time-consuming features could be excluded without significant accuracy loss, offering a substantial gain in execution time. It suggests a feature-group redundancy likely related to the biology of the observed cells. We offer a method to select fast and discriminant features. In our assay, a 79.6 ± 2.4% accurate classification of a cell took 68.7 ± 3.5 ms (mean ± SD, 5-fold cross-validation nested in 10 bootstrap repeats), corresponding to 14 cells per second, dispatched into eight phases of the cell cycle, using 12 feature groups and operating a consumer market ARM-based embedded system. A simple neural network offered similar performances paving the way to faster training and classification, using parallel execution on a general-purpose graphic processing unit. Finally, this strategy is also usable for deep neural networks paving the way to optimizing these algorithms for smart microscopy.

Ethics of AI in Pathology: Current Paradigms and Emerging Issues

Deep learning has rapidly advanced artificial intelligence (AI) and algorithmic decision-making (ADM) paradigms, affecting many traditional fields of medicine, including pathology, which is a heavily data-centric specialty of medicine. The structured nature of pathology data repositories makes it highly attractive to AI researchers to train deep learning models to improve health care delivery. Additionally, there are enormous financial incentives driving adoption of AI and ADM due to promise of increased efficiency of the health care delivery process. AI, if used unethically, may exacerbate existing inequities of health care, especially if not implemented correctly. There is an urgent need to harness the vast power of AI in an ethically and morally justifiable manner. This review explores the key issues involving AI ethics in pathology. Issues related to ethical design of pathology AI studies and the potential risks associated with implementation of AI and ADM within the pathology workflow are discussed. Three key foundational principles of ethical AI: transparency, accountability, and governance, are described in the context of pathology. The future practice of pathology must be guided by these principles. Pathologists should be aware of the potential of AI to deliver superlative health care and the ethical pitfalls associated with it. Finally, pathologists must have a seat at the table to drive future implementation of ethical AI in the practice of pathology.

Visual Analytics for Hypothesis-Driven Exploration in Computational Pathology

Recent advances in computational and algorithmic power are evolving the field of medical imaging rapidly. In cancer research, many new directions are sought to characterize patients with additional imaging features derived from radiology and pathology images. The emerging field of Computational Pathology targets the high-throughput extraction and analysis of the spatial distribution of cells from digital histopathology images. The associated morphological and architectural features allow researchers to quantify and characterize new imaging biomarkers for cancer diagnosis, prognosis, and treatment decisions. However, while the image feature space grows, exploration and analysis become more difficult and ineffective. There is a need for dedicated interfaces for interactive data manipulation and visual analysis of computational pathology and clinical data. For this purpose, we present IIComPath, a visual analytics approach that enables clinical researchers to formulate hypotheses and create computational pathology pipelines involving cohort construction, spatial analysis of image-derived features, and cohort analysis. We demonstrate our approach through use cases that investigate the prognostic value of current diagnostic features and new computational pathology biomarkers.

Machine Learning Approach for Biopsy-Based Identification of Eosinophilic Esophagitis Reveals Importance of Global features

Goal

Eosinophilic esophagitis (EoE) is an allergic inflammatory condition characterized by eosinophil accumulation in the esophageal mucosa. EoE diagnosis includes a manual assessment of eosinophil levels in mucosal biopsies-a time-consuming, laborious task that is difficult to standardize. One of the main challenges in automating this process, like many other biopsy-based diagnostics, is detecting features that are small relative to the size of the biopsy.

Results

In this work, we utilized hematoxylin- and eosin-stained slides from esophageal biopsies from patients with active EoE and control subjects to develop a platform based on a deep convolutional neural network (DCNN) that can classify esophageal biopsies with an accuracy of 85%, sensitivity of 82.5%, and specificity of 87%. Moreover, by combining several downscaling and cropping strategies, we show that some of the features contributing to the correct classification are global rather than specific, local features.

Conclusions

We report the ability of artificial intelligence to identify EoE using computer vision analysis of esophageal biopsy slides. Further, the DCNN features associated with EoE are based on not only local eosinophils but also global histologic changes. Our approach can be used for other conditions that rely on biopsy-based histologic diagnostics.

Deep Learning for Automatic Subclassification of Gastric Carcinoma Using Whole-Slide Histopathology Images

Histomorphologic types of gastric cancer (GC) have significant prognostic values that should be considered during treatment planning. Because the thorough quantitative review of a tissue slide is a laborious task for pathologists, deep learning (DL) can be a useful tool to support pathologic workflow. In the present study, a fully automated approach was applied to distinguish differentiated/undifferentiated and non-mucinous/mucinous tumor types in GC tissue whole-slide images from The Cancer Genome Atlas (TCGA) stomach adenocarcinoma dataset (TCGA-STAD). By classifying small patches of tissue images into differentiated/undifferentiated and non-mucinous/mucinous tumor tissues, the relative proportion of GC tissue subtypes can be easily quantified. Furthermore, the distribution of different tissue subtypes can be clearly visualized. The patch-level areas under the curves for the receiver operating characteristic curves for the differentiated/undifferentiated and non-mucinous/mucinous classifiers were 0.932 and 0.979, respectively. We also validated the classifiers on our own GC datasets and confirmed that the generalizability of the classifiers is excellent. The results indicate that the DL-based tissue classifier could be a useful tool for the quantitative analysis of cancer tissue slides. By combining DL-based classifiers for various molecular and morphologic variations in tissue slides, the heterogeneity of tumor tissues can be unveiled more efficiently.

Artificial Intelligence for solid tumor diagnosis in digital pathology

Tumor diagnosis relies on the visual examination of histological slides by pathologists through a microscope eyepiece. Digital pathology, the digitalization of histological slides at high magnification with slides scanners, has raised the opportunity to extract quantitative information thanks to image analysis. In the last decade, medical image analysis has made exceptional progress due to the development of artificial intelligence (AI) algorithms. AI has been successfully used in the field of medical imaging and more recently in digital pathology. The feasibility and usefulness of AI assisted pathology tasks have been demonstrated in the very last years and we can expect those developments to be applied on routine histopathology in the future. In this review, we will describe and illustrate this technique and present the most recent applications in the field of tumor histopathology.

Classifying Retinal Degeneration in Histological Sections Using Deep Learning

Purpose

Artificial intelligence (AI) techniques are increasingly being used to classify retinal diseases. In this study we investigated the ability of a convolutional neural network (CNN) in categorizing histological images into different classes of retinal degeneration.

Methods

Images were obtained from a chemically induced feline model of monocular retinal dystrophy and split into training and testing sets. The training set was graded for the level of retinal degeneration and used to train various CNN architectures. The testing set was evaluated through the best architecture and graded by six observers. Comparisons between model and observer classifications, and interobserver variability were measured. Finally, the effects of using less training images or images containing half the presentable context were investigated.

Results

The best model gave weighted-F1 scores in the range 85% to 90%. Cohen kappa scores reached up to 0.86, indicating high agreement between the model and observers. Interobserver variability was consistent with the model-observer variability in the model's ability to match predictions with the observers. Image context restriction resulted in model performance reduction by up to 6% and at least one training set size resulted in a model performance reduction of 10% compared to the original size.

Conclusions

Detecting the presence and severity of up to three classes of retinal degeneration in histological data can be reliably achieved with a deep learning classifier.

Translational Relevance

This work lays the foundations for future AI models which could aid in the evaluation of more intricate changes occurring in retinal degeneration, particularly in other types of clinically derived image data.

Feasibility of deep learning-based fully automated classification of microsatellite instability in tissue slides of colorectal cancer

High levels of microsatellite instability (MSI-H) occurs in about 15% of sporadic colorectal cancer (CRC) and is an important predictive marker for response to immune checkpoint inhibitors. To test the feasibility of a deep learning (DL)-based classifier as a screening tool for MSI status, we built a fully automated DL-based MSI classifier using pathology whole-slide images (WSIs) of CRCs. On small image patches of The Cancer Genome Atlas (TCGA) CRC WSI dataset, tissue/non-tissue, normal/tumor and MSS/MSI-H classifiers were applied sequentially for the fully automated prediction of the MSI status. The classifiers were also tested on an independent cohort. Furthermore, to test how the expansion of the training data affects the performance of the DL-based classifier, additional classifier trained on both TCGA and external datasets was tested. The areas under the receiver operating characteristic curves were 0.892 and 0.972 for the TCGA and external datasets, respectively, by a classifier trained on both datasets. The performance of the DL-based classifier was much better than that of previously reported histomorphology-based methods. We speculated that about 40% of CRC slides could be screened for MSI status without molecular testing by the DL-based classifier. These results demonstrated that the DL-based method has potential as a screening tool to discriminate molecular alteration in tissue slides.

Prognostic Modeling of Patients Undergoing Surgery Alone for Esophageal Squamous Cell Carcinoma: A Histopathological and Computed Tomography Based Quantitative Analysis

Objective

To evaluate the effectiveness of a novel computerized quantitative analysis based on histopathological and computed tomography (CT) images for predicting the postoperative prognosis of esophageal squamous cell carcinoma (ESCC) patients.

Methods

We retrospectively reviewed the medical records of 153 ESCC patients who underwent esophagectomy alone and quantitatively analyzed digital histological specimens and diagnostic CT images. We cut pathological images (6000 × 6000) into 50 × 50 patches; each patient had 14,400 patches. Cluster analysis was used to process these patches. We used the pathological clusters to all patches ratio (PCPR) of each case for pathological features and we obtained 20 PCPR quantitative features. Totally, 125 computerized quantitative (20 PCPR and 105 CT) features were extracted. We used a recursive feature elimination approach to select features. A Cox hazard model with L1 penalization was used for prognostic indexing. We compared the following prognostic models: Model A: clinical features; Model B: quantitative CT and clinical features; Model C: quantitative histopathological and clinical features; and Model D: combined information of clinical, CT, and histopathology. Indices of concordance (C-index) and leave-one-out cross-validation (LOOCV) were used to assess prognostic model accuracy.

Results

Five PCPR and eight CT features were treated as significant indicators in ESCC prognosis. C-indices adjusted for LOOCV were comparable among four models, 0.596 (Model A) vs. 0.658 (Model B) vs. 0.651 (Model C), and improved to 0.711with Model D combining information of clinical, CT, and histopathology (all p<0.05). Using Model D, we stratified patients into low- and high-risk groups. The 3-year overall survival rates of low- and high-risk patients were 38.0% and 25.0%, respectively (p<0.001).

Conclusion

Quantitative prognostic modeling using a combination of clinical data, histopathological, and CT images can stratify ESCC patients with surgery alone into high-risk and low-risk groups.

Clinical applications of artificial intelligence in urologic oncology

PURPOSE OF REVIEW

This review aims to shed light on recent applications of artificial intelligence in urologic oncology.

RECENT FINDINGS

Artificial intelligence algorithms harness the wealth of patient data to assist in diagnosing, staging, treating, and monitoring genitourinary malignancies. Successful applications of artificial intelligence in urologic oncology include interpreting diagnostic imaging, pathology, and genomic annotations. Many of these algorithms, however, lack external validity and can only provide predictions based on one type of dataset.

SUMMARY

Future applications of artificial intelligence will need to incorporate several forms of data in order to truly make headway in urologic oncology. Researchers must actively ensure future artificial intelligence developments encompass the entire prospective patient population.

Deep learning classification of lung cancer histology using CT images

Tumor histology is an important predictor of therapeutic response and outcomes in lung cancer. Tissue sampling for pathologist review is the most reliable method for histology classification, however, recent advances in deep learning for medical image analysis allude to the utility of radiologic data in further describing disease characteristics and for risk stratification. In this study, we propose a radiomics approach to predicting non-small cell lung cancer (NSCLC) tumor histology from non-invasive standard-of-care computed tomography (CT) data. We trained and validated convolutional neural networks (CNNs) on a dataset comprising 311 early-stage NSCLC patients receiving surgical treatment at Massachusetts General Hospital (MGH), with a focus on the two most common histological types: adenocarcinoma (ADC) and Squamous Cell Carcinoma (SCC). The CNNs were able to predict tumor histology with an AUC of 0.71(p = 0.018). We also found that using machine learning classifiers such as k-nearest neighbors (kNN) and support vector machine (SVM) on CNN-derived quantitative radiomics features yielded comparable discriminative performance, with AUC of up to 0.71 (p = 0.017). Our best performing CNN functioned as a robust probabilistic classifier in heterogeneous test sets, with qualitatively interpretable visual explanations to its predictions. Deep learning based radiomics can identify histological phenotypes in lung cancer. It has the potential to augment existing approaches and serve as a corrective aid for diagnosticians.

Searching Images for Consensus: Can AI Remove Observer Variability in Pathology?

One of the major obstacles in reaching diagnostic consensus is observer variability. With the recent success of artificial intelligence, particularly the deep networks, the question emerges as to whether the fundamental challenge of diagnostic imaging can now be resolved. This article briefly reviews the problem and how eventually both supervised and unsupervised AI technologies could help to overcome it.

High throughput assessment of biomarkers in tissue microarrays using artificial intelligence: PTEN loss as a proof-of-principle in multi-center prostate cancer cohorts

Phosphatase and tensin homolog (PTEN) loss is associated with adverse outcomes in prostate cancer and has clinical potential as a prognostic biomarker. The objective of this work was to develop an artificial intelligence (AI) system for automated detection and localization of PTEN loss on immunohistochemically (IHC) stained sections. PTEN loss was assessed using IHC in two prostate tissue microarrays (TMA) (internal cohort, n = 272 and external cohort, n = 129 patients). TMA cores were visually scored for PTEN loss by pathologists and, if present, spatially annotated. Cores from each patient within the internal TMA cohort were split into 90% cross-validation (N = 2048) and 10% hold-out testing (N = 224) sets. ResNet-101 architecture was used to train core-based classification using a multi-resolution ensemble approach (×5, ×10, and ×20). For spatial annotations, single resolution pixel-based classification was trained from patches extracted at ×20 resolution, interpolated to ×40 resolution, and applied in a sliding-window fashion. A final AI-based prediction model was created from combining multi-resolution and pixel-based models. Performance was evaluated in 428 cores of external cohort. From both cohorts, a total of 2700 cores were studied, with a frequency of PTEN loss of 14.5% in internal (180/1239) and external 13.5% (43/319) cancer cores. The final AI-based prediction of PTEN status demonstrated 98.1% accuracy (95.0% sensitivity, 98.4% specificity; median dice score = 0.811) in internal cohort cross-validation set and 99.1% accuracy (100% sensitivity, 99.0% specificity; median dice score = 0.804) in internal cohort test set. Overall core-based classification in the external cohort was significantly improved in the external cohort (area under the curve = 0.964, 90.6% sensitivity, 95.7% specificity) when further trained (fine-tuned) using 15% of cohort data (19/124 patients). These results demonstrate a robust and fully automated method for detection and localization of PTEN loss in prostate cancer tissue samples. AI-based algorithms have potential to streamline sample assessment in research and clinical laboratories.

Generalizability of Deep Learning System for the Pathologic Diagnosis of Various Cancers

The deep learning (DL)-based approaches in tumor pathology help to overcome the limitations of subjective visual examination from pathologists and improve diagnostic accuracy and objectivity. However, it is unclear how a DL system trained to discriminate normal/tumor tissues in a specific cancer could perform on other tumor types. Herein, we cross-validated the DL-based normal/tumor classifiers separately trained on the tissue slides of cancers from bladder, lung, colon and rectum, stomach, bile duct, and liver. Furthermore, we compared the differences between the classifiers trained on the frozen or formalin-fixed paraffin-embedded (FFPE) tissues. The Area under the curve (AUC) for the receiver operating characteristic (ROC) curve ranged from 0.982 to 0.999 when the tissues were analyzed by the classifiers trained on the same tissue preparation modalities and cancer types. However, the AUCs could drop to 0.476 and 0.439 when the classifiers trained for different tissue modalities and cancer types were applied. Overall, the optimal performance could be achieved only when the tissue slides were analyzed by the classifiers trained on the same preparation modalities and cancer types.

Development and Validation of a Deep Learning Model to Quantify Glomerulosclerosis in Kidney Biopsy Specimens

IMPORTANCE

A chronic shortage of donor kidneys is compounded by a high discard rate, and this rate is directly associated with biopsy specimen evaluation, which shows poor reproducibility among pathologists. A deep learning algorithm for measuring percent global glomerulosclerosis (an important predictor of outcome) on images of kidney biopsy specimens could enable pathologists to more reproducibly and accurately quantify percent global glomerulosclerosis, potentially saving organs that would have been discarded.

OBJECTIVE

To compare the performances of pathologists with a deep learning model on quantification of percent global glomerulosclerosis in whole-slide images of donor kidney biopsy specimens, and to determine the potential benefit of a deep learning model on organ discard rates.

DESIGN, SETTING, AND PARTICIPANTS

This prognostic study used whole-slide images acquired from 98 hematoxylin-eosin-stained frozen and 51 permanent donor biopsy specimen sections retrieved from 83 kidneys. Serial annotation by 3 board-certified pathologists served as ground truth for model training and for evaluation. Images of kidney biopsy specimens were obtained from the Washington University database (retrieved between June 2015 and June 2017). Cases were selected randomly from a database of more than 1000 cases to include biopsy specimens representing an equitable distribution within 0% to 5%, 6% to 10%, 11% to 15%, 16% to 20%, and more than 20% global glomerulosclerosis.

MAIN OUTCOMES AND MEASURES

Correlation coefficient (r) and root-mean-square error (RMSE) with respect to annotations were computed for cross-validated model predictions and on-call pathologists' estimates of percent global glomerulosclerosis when using individual and pooled slide results. Data were analyzed from March 2018 to August 2020.

RESULTS

The cross-validated model results of section images retrieved from 83 donor kidneys showed higher correlation with annotations (r = 0.916; 95% CI, 0.886-0.939) than on-call pathologists (r = 0.884; 95% CI, 0.825-0.923) that was enhanced when pooling glomeruli counts from multiple levels (r = 0.933; 95% CI, 0.898-0.956). Model prediction error for single levels (RMSE, 5.631; 95% CI, 4.735-6.517) was 14% lower than on-call pathologists (RMSE, 6.523; 95% CI, 5.191-7.783), improving to 22% with multiple levels (RMSE, 5.094; 95% CI, 3.972-6.301). The model decreased the likelihood of unnecessary organ discard by 37% compared with pathologists.

CONCLUSIONS AND RELEVANCE

The findings of this prognostic study suggest that this deep learning model provided a scalable and robust method to quantify percent global glomerulosclerosis in whole-slide images of donor kidneys. The model performance improved by analyzing multiple levels of a section, surpassing the capacity of pathologists in the time-sensitive setting of examining donor biopsy specimens. The results indicate the potential of a deep learning model to prevent erroneous donor organ discard.

Predicting single-cell gene expression profiles of imaging flow cytometry data with machine learning

High-content imaging and single-cell genomics are two of the most prominent high-throughput technologies for studying cellular properties and functions at scale. Recent studies have demonstrated that information in large imaging datasets can be used to estimate gene mutations and to predict the cell-cycle state and the cellular decision making directly from cellular morphology. Thus, high-throughput imaging methodologies, such as imaging flow cytometry can potentially aim beyond simple sorting of cell-populations. We introduce IFC-seq, a machine learning methodology for predicting the expression profile of every cell in an imaging flow cytometry experiment. Since it is to-date unfeasible to observe single-cell gene expression and morphology in flow, we integrate uncoupled imaging data with an independent transcriptomics dataset by leveraging common surface markers. We demonstrate that IFC-seq successfully models gene expression of a moderate number of key gene-markers for two independent imaging flow cytometry datasets: (i) human blood mononuclear cells and (ii) mouse myeloid progenitor cells. In the case of mouse myeloid progenitor cells IFC-seq can predict gene expression directly from brightfield images in a label-free manner, using a convolutional neural network. The proposed method promises to add gene expression information to existing and new imaging flow cytometry datasets, at no additional cost.

Feasibility of fully automated classification of whole slide images based on deep learning

Although microscopic analysis of tissue slides has been the basis for disease diagnosis for decades, intra- and inter-observer variabilities remain issues to be resolved. The recent introduction of digital scanners has allowed for using deep learning in the analysis of tissue images because many whole slide images (WSIs) are accessible to researchers. In the present study, we investigated the possibility of a deep learning-based, fully automated, computer-aided diagnosis system with WSIs from a stomach adenocarcinoma dataset. Three different convolutional neural network architectures were tested to determine the better architecture for tissue classifier. Each network was trained to classify small tissue patches into normal or tumor. Based on the patch-level classification, tumor probability heatmaps can be overlaid on tissue images. We observed three different tissue patterns, including clear normal, clear tumor and ambiguous cases. We suggest that longer inspection time can be assigned to ambiguous cases compared to clear normal cases, increasing the accuracy and efficiency of histopathologic diagnosis by pre-evaluating the status of the WSIs. When the classifier was tested with completely different WSI dataset, the performance was not optimal because of the different tissue preparation quality. By including a small amount of data from the new dataset for training, the performance for the new dataset was much enhanced. These results indicated that WSI dataset should include tissues prepared from many different preparation conditions to construct a generalized tissue classifier. Thus, multi-national/multi-center dataset should be built for the application of deep learning in the real world medical practice.

Prediction of clinically actionable genetic alterations from colorectal cancer histopathology images using deep learning

BACKGROUND

Identifying genetic mutations in cancer patients have been increasingly important because distinctive mutational patterns can be very informative to determine the optimal therapeutic strategy. Recent studies have shown that deep learning-based molecular cancer subtyping can be performed directly from the standard hematoxylin and eosin (H&E) sections in diverse tumors including colorectal cancers (CRCs). Since H&E-stained tissue slides are ubiquitously available, mutation prediction with the pathology images from cancers can be a time- and cost-effective complementary method for personalized treatment.

AIM

To predict the frequently occurring actionable mutations from the H&E-stained CRC whole-slide images (WSIs) with deep learning-based classifiers.

METHODS

A total of 629 CRC patients from The Cancer Genome Atlas (TCGA-COAD and TCGA-READ) and 142 CRC patients from Seoul St. Mary Hospital (SMH) were included. Based on the mutation frequency in TCGA and SMH datasets, we chose APC, KRAS, PIK3CA, SMAD4, and TP53 genes for the study. The classifiers were trained with 360 × 360 pixel patches of tissue images. The receiver operating characteristic (ROC) curves and area under the curves (AUCs) for all the classifiers were presented.

RESULTS

The AUCs for ROC curves ranged from 0.693 to 0.809 for the TCGA frozen WSIs and from 0.645 to 0.783 for the TCGA formalin-fixed paraffin-embedded WSIs. The prediction performance can be enhanced with the expansion of datasets. When the classifiers were trained with both TCGA and SMH data, the prediction performance was improved.

CONCLUSION

APC, KRAS, PIK3CA, SMAD4, and TP53 mutations can be predicted from H&E pathology images using deep learning-based classifiers, demonstrating the potential for deep learning-based mutation prediction in the CRC tissue slides.

Closing the translation gap: AI applications in digital pathology

Recent advances in artificial intelligence show tremendous promise to improve the accuracy, reproducibility, and availability of medical diagnostics across a number of medical subspecialities. This is especially true in the field of digital pathology, which has recently witnessed a surge in publications describing state-of-the-art performance for machine learning models across a wide range of diagnostic applications. Nonetheless, despite this promise, there remain significant gaps in translating applications for any of these technologies into actual clinical practice. In this review, we will first give a brief overview of the recent progress in applying AI to digitized pathology images, focusing on how these tools might be applied in clinical workflows in the near term to improve the accuracy and efficiency of pathologists. Then we define and describe in detail the various factors that need to be addressed in order to successfully close the "translation gap" for AI applications in digital pathology.

Weakly supervised instance learning for thyroid malignancy prediction from whole slide cytopathology images

We consider machine-learning-based thyroid-malignancy prediction from cytopathology whole-slide images (WSI). Multiple instance learning (MIL) approaches, typically used for the analysis of WSIs, divide the image (bag) into patches (instances), which are used to predict a single bag-level label. These approaches perform poorly in cytopathology slides due to a unique bag structure: sparsely located informative instances with varying characteristics of abnormality. We address these challenges by considering multiple types of labels: bag-level malignancy and ordered diagnostic scores, as well as instance-level informativeness and abnormality labels. We study their contribution beyond the MIL setting by proposing a maximum likelihood estimation (MLE) framework, from which we derive a two-stage deep-learning-based algorithm. The algorithm identifies informative instances and assigns them local malignancy scores that are incorporated into a global malignancy prediction. We derive a lower bound of the MLE, leading to an improved training strategy based on weak supervision, that we motivate through statistical analysis. The lower bound further allows us to extend the proposed algorithm to simultaneously predict multiple bag and instance-level labels from a single output of a neural network. Experimental results demonstrate that the proposed algorithm provides competitive performance compared to several competing methods, achieves (expert) human-level performance, and allows augmentation of human decisions.