A self-supervised framework for cross-modal search in histopathology archives using scale harmonization.

The exponential growth of data across various medical domains has generated a substantial demand for techniques to analyze multimodal big data. This demand is particularly pronounced in fields such as computational pathology due to the diverse nature of the tissue. Cross-modal retrieval aims to identify a common latent space where different modalities, such as image-text pairs, exhibit close alignment. The primary challenge, however, often lies in the representation of tissue features. While language models can be trained relatively easily, visual models frequently struggle due to the scarcity of labeled data. To address this issue, the innovative concept of harmonization has been introduced, extending the learning scheme distillation without supervision, known as DINO. The harmonization of scale refines the DINO paradigm through a novel patching approach, overcoming the complexities posed by gigapixel whole slide images in digital pathology. Experiments conducted on diverse datasets have demonstrated that the proposed approach significantly enhances cross-modal retrieval in tissue imaging. Moreover, it exhibits vast potential for other fields that rely on gigapixel imaging.

On image search in histopathology.

Pathology images of histopathology can be acquired from camera-mounted microscopes or whole-slide scanners. Utilizing similarity calculations to match patients based on these images holds significant potential in research and clinical contexts. Recent advancements in search technologies allow for implicit quantification of tissue morphology across diverse primary sites, facilitating comparisons, and enabling inferences about diagnosis, and potentially prognosis, and predictions for new patients when compared against a curated database of diagnosed and treated cases. In this article, we comprehensively review the latest developments in image search technologies for histopathology, offering a concise overview tailored for computational pathology researchers seeking effective, fast, and efficient image search methods in their work.

Creating an atlas of normal tissue for pruning WSI patching through anomaly detection.

Patching whole slide images (WSIs) is an important task in computational pathology. While most of them are designed to classify or detect the presence of pathological lesions in a WSI, the confounding role and redundant nature of normal histology are generally overlooked. In this paper, we propose and validate the concept of an "atlas of normal tissue" solely using samples of WSIs obtained from normal biopsies. Such atlases can be employed to eliminate normal fragments of tissue samples and hence increase the representativeness of the remaining patches. We tested our proposed method by establishing a normal atlas using 107 normal skin WSIs and demonstrated how established search engines like Yottixel can be improved. We used 553 WSIs of cutaneous squamous cell carcinoma to demonstrate the advantage. We also validated our method applied to an external dataset of 451 breast WSIs. The number of selected WSI patches was reduced by 30% to 50% after utilizing the proposed normal atlas while maintaining the same indexing and search performance in leave-one-patient-out validation for both datasets. We show that the proposed concept of establishing and using a normal atlas shows promise for unsupervised selection of the most representative patches of the abnormal WSI patches.

Composite Biomarker Image for Advanced Visualization in Histopathology.

Immunohistochemistry (IHC) biomarkers are essential tools for reliable cancer diagnosis and subtyping. It requires cross-staining comparison among Whole Slide Images (WSIs) of IHCs and hematoxylin and eosin (H&E) slides. Currently, pathologists examine the visually co-localized areas across IHC and H&E glass slides for a final diagnosis, which is a tedious and challenging task. Moreover, visually inspecting different IHC slides back and forth to analyze local co-expressions is inherently subjective and prone to error, even when carried out by experienced pathologists. Relying on digital pathology, we propose "Composite Biomarker Image" (CBI) in this work. CBI is a single image that can be composed using different filtered IHC biomarker images for better visualization. We present a CBI image produced in two steps by the proposed solution for better visualization and hence more efficient clinical workflow. In the first step, IHC biomarker images are aligned with the H&E images using one coordinate system and orientation. In the second step, the positive or negative IHC regions from each biomarker image (based on the pathologists' recommendation) are filtered and combined into one image using a fuzzy inference system. For evaluation, the resulting CBI images, from the proposed system, were evaluated qualitatively by the expert pathologists. The CBI concept helps the pathologists to identify the suspected target tissues more easily, which could be further assessed by examining the actual WSIs at the same suspected regions.

Immunohistochemistry Biomarkers-Guided Image Search for Histopathology.

Medical practitioners use a number of diagnostic tests to make a reliable diagnosis. Traditionally, Haematoxylin and Eosin (H&E) stained glass slides have been used for cancer diagnosis and tumor detection. However, recently a variety of immunohistochemistry (IHC) stained slides can be requested by pathologists to examine and confirm diagnoses for determining the subtype of a tumor when this is difficult using H&E slides only. Deep learning (DL) has received a lot of interest recently for image search engines to extract features from tissue regions, which may or may not be the target region for diagnosis. This approach generally fails to capture high-level patterns corresponding to the malignant or abnormal content of histopathology images. In this work, we are proposing a targeted image search approach, inspired by the pathologists' workflow, which may use information from multiple IHC biomarker images when available. These IHC images could be aligned, filtered, and merged together to generate a composite biomarker image (CBI) that could eventually be used to generate an attention map to guide the search engine for localized search. In our experiments, we observed that an IHC-guided image search engine can retrieve relevant data more accurately than a conventional (i.e., H&E-only) search engine without IHC guidance. Moreover, such engines are also able to accurately conclude the subtypes through majority votes.

Class-imbalanced Unsupervised and Semi-Supervised Domain Adaptation for Histopathology Images.

In dealing with the lack of sufficient annotated data and in contrast to supervised learning, unsupervised, self-supervised, and semi-supervised domain adaptation methods are promising approaches, enabling us to transfer knowledge from rich labeled source domains to different (but related) unlabeled target domains, reducing distribution discrepancy between the source and target domains. However, most existing domain adaptation methods do not consider the imbalanced nature of the real-world data, affecting their performance in practice. We propose to overcome this limitation by proposing a novel domain adaptation approach that includes two modifications to the existing models. Firstly, we leverage the focal loss function in response to class-imbalanced labeled data in the source domain. Secondly, we introduce a novel co-training approach to involve pseudo-labeled target data points in the training process. Experiments show that the proposed model can be effective in transferring knowledge from source to target domain. As an example, we use the classification of prostate cancer images into low-cancerous and high-cancerous regions.

Whole slide image representation in bone marrow cytology.

One of the goals of AI-based computational pathology is to generate compact representations of whole slide images (WSIs) that capture the essential information needed for diagnosis. While such approaches have been applied to histopathology, few applications have been reported in cytology. Bone marrow aspirate cytology is the basis for key clinical decisions in hematology. However, visual inspection of aspirate specimens is a tedious and complex process subject to variation in interpretation, and hematopathology expertise is scarce. The ability to generate a compact representation of an aspirate specimen may form the basis for clinical decision-support tools in hematology. In this study, we leverage our previously published end-to-end AI-based system for counting and classifying cells from bone marrow aspirate WSIs, which enables the direct use of individual cells as inputs rather than WSI patches. We then construct bags of individual cell features from each WSI, and apply multiple instance learning to extract their vector representations. To evaluate the quality of our representations, we conducted WSI retrieval and classification tasks. Our results show that we achieved a mAP@10 of 0.58 ±0.02 in WSI-level image retrieval, surpassing the random-retrieval baseline of 0.39 ±0.1. Furthermore, we predicted five diagnostic labels for individual aspirate WSIs with a weighted-average F1 score of 0.57 ±0.03 using a k-nearest-neighbors (k-NN) model, outperforming guessing using empirical class prior probabilities (0.26 ±0.02). We present the first example of exploring trainable mechanisms to generate compact, slide-level representations in bone marrow cytology with deep learning. This method has the potential to summarize complex semantic information in WSIs toward improved diagnostics in hematology, and may eventually support AI-assisted computational pathology approaches.

Cell projection plots: A novel visualization of bone marrow aspirate cytology.

Deep models for cell detection have demonstrated utility in bone marrow cytology, showing impressive results in terms of accuracy and computational efficiency. However, these models have yet to be implemented in the clinical diagnostic workflow. Additionally, the metrics used to evaluate cell detection models are not necessarily aligned with clinical goals and targets. In order to address these issues, we introduce novel, automatically generated visual summaries of bone marrow aspirate specimens called cell projection plots (CPPs). Encompassing relevant biological patterns such as neutrophil maturation, CPPs provide a compact summary of bone marrow aspirate cytology. To gauge clinical relevance, CPPs were inspected by 3 hematopathologists, who decided whether corresponding diagnostic synopses matched with generated CPPs. Pathologists were able to match CPPs to the correct synopsis with a matching degree of 85%. Our finding suggests CPPs can represent clinically relevant information from bone marrow aspirate specimens and may be used to efficiently summarize bone marrow cytology to pathologists. CPPs could be a step toward human-centered implementation of artificial intelligence (AI) in hematopathology, and a basis for a diagnostic-support tool for digital pathology workflows.

Learning binary and sparse permutation-invariant representations for fast and memory efficient whole slide image search.

Considering their gigapixel sizes, the representation of whole slide images (WSIs) for classification and retrieval systems is a non-trivial task. Patch processing and multi-Instance Learning (MIL) are common approaches to analyze WSIs. However, in end-to-end training, these methods require high GPU memory consumption due to the simultaneous processing of multiple sets of patches. Furthermore, compact WSI representations through binary and/or sparse representations are urgently needed for real-time image retrieval within large medical archives. To address these challenges, we propose a novel framework for learning compact WSI representations utilizing deep conditional generative modeling and the Fisher Vector Theory. The training of our method is instance-based, achieving better memory and computational efficiency during the training. To achieve efficient large-scale WSI search, we introduce new loss functions, namely gradient sparsity and gradient quantization losses, for learning sparse and binary permutation-invariant WSI representations called Conditioned Sparse Fisher Vector (C-Deep-SFV), and Conditioned Binary Fisher Vector (C-Deep-BFV). The learned WSI representations are validated on the largest public WSI archive, The Cancer Genomic Atlas (TCGA) and also Liver-Kidney-Stomach (LKS) dataset. For WSI search, the proposed method outperforms Yottixel and Gaussian Mixture Model (GMM)-based Fisher Vector both in terms of retrieval accuracy and speed. For WSI classification, we achieve competitive performance against state-of-art on lung cancer data from TCGA and the public benchmark LKS dataset.

Biased data, biased AI: deep networks predict the acquisition site of TCGA images.

BACKGROUND: Deep learning models applied to healthcare applications including digital pathology have been increasing their scope and importance in recent years. Many of these models have been trained on The Cancer Genome Atlas (TCGA) atlas of digital images, or use it as a validation source. One crucial factor that seems to have been widely ignored is the internal bias that originates from the institutions that contributed WSIs to the TCGA dataset, and its effects on models trained on this dataset. METHODS: 8,579 paraffin-embedded, hematoxylin and eosin stained, digital slides were selected from the TCGA dataset. More than 140 medical institutions (acquisition sites) contributed to this dataset. Two deep neural networks (DenseNet121 and KimiaNet were used to extract deep features at 20× magnification. DenseNet was pre-trained on non-medical objects. KimiaNet has the same structure but trained for cancer type classification on TCGA images. The extracted deep features were later used to detect each slide's acquisition site, and also for slide representation in image search. RESULTS: DenseNet's deep features could distinguish acquisition sites with 70% accuracy whereas KimiaNet's deep features could reveal acquisition sites with more than 86% accuracy. These findings suggest that there are acquisition site specific patterns that could be picked up by deep neural networks. It has also been shown that these medically irrelevant patterns can interfere with other applications of deep learning in digital pathology, namely image search. This study shows that there are acquisition site specific patterns that can be used to identify tissue acquisition sites without any explicit training. Furthermore, it was observed that a model trained for cancer subtype classification has exploited such medically irrelevant patterns to classify cancer types. Digital scanner configuration and noise, tissue stain variation and artifacts, and source site patient demographics are among factors that likely account for the observed bias. Therefore, researchers should be cautious of such bias when using histopathology datasets for developing and training deep networks.

Decentralized federated learning through proxy model sharing.

Institutions in highly regulated domains such as finance and healthcare often have restrictive rules around data sharing. Federated learning is a distributed learning framework that enables multi-institutional collaborations on decentralized data with improved protection for each collaborator's data privacy. In this paper, we propose a communication-efficient scheme for decentralized federated learning called ProxyFL, or proxy-based federated learning. Each participant in ProxyFL maintains two models, a private model, and a publicly shared proxy model designed to protect the participant's privacy. Proxy models allow efficient information exchange among participants without the need of a centralized server. The proposed method eliminates a significant limitation of canonical federated learning by allowing model heterogeneity; each participant can have a private model with any architecture. Furthermore, our protocol for communication by proxy leads to stronger privacy guarantees using differential privacy analysis. Experiments on popular image datasets, and a cancer diagnostic problem using high-quality gigapixel histology whole slide images, show that ProxyFL can outperform existing alternatives with much less communication overhead and stronger privacy.

Generalization of vision pre-trained models for histopathology.

Out-of-distribution (OOD) generalization, especially for medical setups, is a key challenge in modern machine learning which has only recently received much attention. We investigate how different convolutional pre-trained models perform on OOD test data-that is data from domains that have not been seen during training-on histopathology repositories attributed to different trial sites. Different trial site repositories, pre-trained models, and image transformations are examined as specific aspects of pre-trained models. A comparison is also performed among models trained entirely from scratch (i.e., without pre-training) and models already pre-trained. The OOD performance of pre-trained models on natural images, i.e., (1) vanilla pre-trained ImageNet, (2) semi-supervised learning (SSL), and (3) semi-weakly-supervised learning (SWSL) models pre-trained on IG-1B-Targeted are examined in this study. In addition, the performance of a histopathology model (i.e., KimiaNet) trained on the most comprehensive histopathology dataset, i.e., TCGA, has also been studied. Although the performance of SSL and SWSL pre-trained models are conducive to better OOD performance in comparison to the vanilla ImageNet pre-trained model, the histopathology pre-trained model is still the best in overall. In terms of top-1 accuracy, we demonstrate that diversifying the images in the training using reasonable image transformations is effective to avoid learning shortcuts when the distribution shift is significant. In addition, XAI techniques-which aim to achieve high-quality human-understandable explanations of AI decisions-are leveraged for further investigations.

Proportionally Fair Hospital Collaborations in Federated Learning of Histopathology Images.

Medical centers and healthcare providers have concerns and hence restrictions around sharing data with external collaborators. Federated learning, as a privacy-preserving method, involves learning a site-independent model without having direct access to patient-sensitive data in a distributed collaborative fashion. The federated approach relies on decentralized data distribution from various hospitals and clinics. The collaboratively learned global model is supposed to have acceptable performance for the individual sites. However, existing methods focus on minimizing the average of the aggregated loss functions, leading to a biased model that performs perfectly for some hospitals while exhibiting undesirable performance for other sites. In this paper, we improve model "fairness" among participating hospitals by proposing a novel federated learning scheme called Proportionally Fair Federated Learning, short Prop-FFL. Prop-FFL is based on a novel optimization objective function to decrease the performance variations among participating hospitals. This function encourages a fair model, providing us with more uniform performance across participating hospitals. We validate the proposed Prop-FFL on two histopathology datasets as well as two general datasets to shed light on its inherent capabilities. The experimental results suggest promising performance in terms of learning speed, accuracy, and fairness.

Learning to predict RNA sequence expressions from whole slide images with applications for search and classification.

Deep learning methods are widely applied in digital pathology to address clinical challenges such as prognosis and diagnosis. As one of the most recent applications, deep models have also been used to extract molecular features from whole slide images. Although molecular tests carry rich information, they are often expensive, time-consuming, and require additional tissue to sample. In this paper, we propose tRNAsformer, an attention-based topology that can learn both to predict the bulk RNA-seq from an image and represent the whole slide image of a glass slide simultaneously. The tRNAsformer uses multiple instance learning to solve a weakly supervised problem while the pixel-level annotation is not available for an image. We conducted several experiments and achieved better performance and faster convergence in comparison to the state-of-the-art algorithms. The proposed tRNAsformer can assist as a computational pathology tool to facilitate a new generation of search and classification methods by combining the tissue morphology and the molecular fingerprint of the biopsy samples.

Bias reduction in representation of histopathology images using deep feature selection.

Appearing traces of bias in deep networks is a serious reliability issue which can play a significant role in ethics and generalization related concerns. Recent studies report that the deep features extracted from the histopathology images of The Cancer Genome Atlas (TCGA), the largest publicly available archive, are surprisingly able to accurately classify the whole slide images (WSIs) based on their acquisition site while these features are extracted to primarily discriminate cancer types. This is clear evidence that the utilized Deep Neural Networks (DNNs) unexpectedly detect the specific patterns of the source site, i.e, the hospital of origin, rather than histomorphologic patterns, a biased behavior resulting in degraded trust and generalization. This observation motivated us to propose a method to alleviate the destructive impact of hospital bias through a novel feature selection process. To this effect, we have proposed an evolutionary strategy to select a small set of optimal features to not only accurately represent the histological patterns of tissue samples but also to eliminate the features contributing to internal bias toward the institution. The defined objective function for an optimal subset selection of features is to minimize the accuracy of the model to classify the source institutions which is basically defined as a bias indicator. By the conducted experiments, the selected features extracted by the state-of-the-art network trained on TCGA images (i.e., the KimiaNet), considerably decreased the institutional bias, while improving the quality of features to discriminate the cancer types. In addition, the selected features could significantly improve the results of external validation compared to the entire set of features which has been negatively affected by bias. The proposed scheme is a model-independent approach which can be employed when it is possible to define a bias indicator as a participating objective in a feature selection process; even with unknown bias sources.

Evolutionary deep feature selection for compact representation of gigapixel images in digital pathology.

Despite the recent progress in Deep Neural Networks (DNNs) to characterize histopathology images, compactly representing a gigapixel whole-slide image (WSI) via salient features to enable computational pathology is still an urgent need and a significant challenge. In this paper, we propose a novel WSI characterization approach to represent, search and classify biopsy specimens using a compact feature vector (CFV) extracted from a multitude of deep feature vectors. Since the non-optimal design and training of deep networks may result in many irrelevant and redundant features and also cause computational bottlenecks, we proposed a low-cost stochastic method to optimize the output of pre-trained deep networks using evolutionary algorithms to generate a very small set of features to accurately represent each tissue/biopsy. The performance of the proposed method has been assessed using WSIs from the publicly available TCGA image data. In addition to acquiring a very compact representation (i.e., 11,000 times smaller than the initial set of features), the optimized features achieved 93% classification accuracy resulting in 11% improvement compared to the published benchmarks. The experimental results reveal that the proposed method can reliably select salient features of the biopsy sample. Furthermore, the proposed approach holds the potential to immensely facilitate the adoption of digital pathology by enabling a new generation of WSI representation for efficient storage and more user-friendly visualization.

Hospital-Agnostic Image Representation Learning in Digital Pathology.

Whole Slide Images (WSIs) in digital pathology are used to diagnose cancer subtypes. The difference in procedures to acquire WSIs at various trial sites gives rise to variability in the histopathology images, thus making consistent diagnosis challenging. These differences may stem from variability in image acquisition through multi-vendor scanners, variable acquisition parameters, and differences in staining procedure; as well, patient demographics may bias the glass slide batches before image acquisition. These variabilities are assumed to cause a domain shift in the images of different hospitals. It is crucial to overcome this domain shift because an ideal machine-learning model must be able to work on the diverse sources of images, independent of the acquisition center. A domain generalization technique is leveraged in this study to improve the generalization capability of a Deep Neural Network (DNN), to an unseen histopathology image set (i.e., from an unseen hospital/trial site) in the presence of domain shift. According to experimental results, the conventional supervisedlearning regime generalizes poorly to data collected from different hospitals. However, the proposed hospital-agnostic learning can improve the generalization considering the lowdimensional latent space representation visualization, and classification accuracy results.

A self-supervised contrastive learning approach for whole slide image representation in digital pathology.

Image analysis in digital pathology has proven to be one of the most challenging fields in medical imaging for AI-driven classification and search tasks. Due to their gigapixel dimensions, whole slide images (WSIs) are difficult to represent for computational pathology. Self-supervised learning (SSL) has recently demonstrated excellent performance in learning effective representations on pretext objectives, which may improve the generalizations of downstream tasks. Previous self-supervised representation methods rely on patch selection and classification such that the effect of SSL on end-to-end WSI representation is not investigated. In contrast to existing augmentation-based SSL methods, this paper proposes a novel self-supervised learning scheme based on the available primary site information. We also design a fully supervised contrastive learning setup to increase the robustness of the representations for WSI classification and search for both pretext and downstream tasks. We trained and evaluated the model on more than 6000 WSIs from The Cancer Genome Atlas (TCGA) repository provided by the National Cancer Institute. The proposed architecture achieved excellent results on most primary sites and cancer subtypes. We also achieved the best result on validation on a lung cancer classification task.

Analysis of Language Embeddings for Classification of Unstructured Pathology Reports.

A pathology report is one of the most significant medical documents providing interpretive insights into the visual appearance of the patient's biopsy sample. In digital pathology, high-resolution images of tissue samples are stored along with pathology reports. Despite the valuable information that pathology reports hold, they are not used in any systematic manner to promote computational pathology. In this work, we focus on analyzing the reports, which are generally unstructured documents written in English with sophisticated and highly specialized medical terminology. We provide a comparative analysis of various embedding models like BioBERT, Clinical BioBERT, BioMed-RoBERTa and Term Frequency-Inverse Document Frequency (TF-IDF), a traditional NLP technique, as well as the combination of embeddings from pre-trained models with TF-IDF. Our results demonstrate the effectiveness of various word embedding techniques for pathology reports.

A Similarity Measure of Histopathology Images by Deep Embeddings.

Histopathology digital scans are large-size images that contain valuable information at the pixel level. Contentbased comparison of these images is a challenging task. This study proposes a content-based similarity measure for highresolution gigapixel histopathology images. The proposed similarity measure is an expansion of cosine vector similarity to a matrix. Each image is divided into same-size patches with a meaningful amount of information (i.e., contained enough tissue). The similarity is measured by the extraction of patchlevel deep embeddings of the last pooling layer of a pre-trained deep model at four different magnification levels, namely, 1x, 2.5x, 5x, and 10x magnifications. In addition, for faster measurement, embedding reduction is investigated. Finally, to assess the proposed method, an image search method is implemented. Results show that the similarity measure represents the slide labels with a maximum accuracy of 93.18% for top-5 search at 5x magnification.