Explore the vast range of titles available.

Digital pathology provides a possibility for computational analysis of histological slides and automatization of routine pathological tasks. Histological slides are very heterogeneous concerning staining, sections’ thickness, and artifacts arising during tissue processing, cutting, staining, and digitization. In this study, we digitally reproduce major types of artifacts. Using six datasets from four different institutions digitized by different scanner systems, we systematically explore artifacts’ influence on the accuracy of the pre-trained, validated, deep learning-based model for prostate cancer detection in histological slides. We provide evidence that any histological artifact dependent on severity can lead to a substantial loss in model performance. Strategies for the prevention of diagnostic model accuracy losses in the context of artifacts are warranted. Stress-testing of diagnostic models using synthetically generated artifacts might be an essential step during clinical validation of deep learning-based algorithms.

Deep learning (DL) is a powerful methodology for the recognition and classification of tissue structures in digital pathology. Its performance in prostate cancer pathology is still under intensive investigation. Here we develop DL-based models for the detection of prostate cancer tissue in whole-slide images based on a large high-quality annotated training dataset and a modern state-of-the-art convolutional network architecture (NASNetLarge). The overall accuracy of our model for tumour detection in two validation cohorts is comparable to that of pathologists and reaches 97.3% in a native version and more than 98% using the suggested DL-based augmentation strategies. As a second step, we suggest a new biologically meaningful DL-based algorithm for Gleason grading of prostatic adenocarcinomas with high, human-level performance in prognostic stratification of patients when tested in several well-characterized validation cohorts. Furthermore, we determine the optimal minimal tumour size (real size of approximately 560 × 560 µm) for robust Gleason grading representative of the whole tumour focus. Our approach is realized in the unified digital pathology pipeline, which delivers all the relevant tumour metrics for a pathology report. Deep learning methods can be a powerful part of digital pathology workflows, provided well-annotated training datasets are available. Tolkach and colleagues develop a deep learning model to recognize and grade prostate cancer, based on a convolution neural network and a dataset with high-quality labels at gland-level precision.

Federated and Continual Learning have emerged as promising paradigms for the privacy-aware use of Deep Learning in dynamic environments by addressing spatial and temporal constraints on data availability. However, Client Drift and Catastrophic Forgetting are fundamental obstacles to ensuring robust performance. Existing work only addresses these problems separately, neglecting the fact that the root cause behind them, namely an unexpected shift in the data distribution, is connected. We propose a unified analysis framework for building a controlled test environment where we can jointly model spatial and temporal shifts, more closely emulating real dynamic settings. By generating a 3D landscape of the combined performance impact, we show that a moderate combination of both shifts can even improve the performance of the resulting model (“Generalization Bump”). We apply a simple and commonly used method from continual learning in the federated setting and observe this reoccurring phenomenon.

Histological slides contain numerous artifacts that can significantly deteriorate the performance of image analysis algorithms. Here we develop the GrandQC tool for tissue and multi-class artifact segmentation. GrandQC allows for high-precision tissue segmentation (Dice score 0.957) and segmentation of tissue without artifacts (Dice score 0.919–0.938 dependent on magnification). Slides from 19 international pathology departments digitized with the most common scanning systems and from The Cancer Genome Atlas dataset were used to establish a QC benchmark, analyzing inter-institutional, intra-institutional, temporal, and inter-scanner slide quality variations. GrandQC improves the performance of downstream image analysis algorithms. We open-source the GrandQC tool, our large manually annotated test dataset, and all QC masks for the entire TCGA cohort to address the problem of QC in digital/computational pathology. GrandQC can be used as a tool to monitor sample preparation and scanning quality in pathology departments and help to track and eliminate major artifact sources. Histological slides often contain artifacts that affect the performance of downstream image analysis. Here, the authors present GrandQC, a tool that enables high-precision tissue and artifact segmentation in histological slides. This tool can be used to monitor sample preparation and scanning quality across pathology departments.

Pleomorphic dermal sarcomas are infrequent neoplastic skin tumors, manifesting in regions of the skin exposed to ultraviolet radiation. Diagnosing the entity can be challenging and therapeutic options are limited. We analyzed 20 samples of normal healthy skin tissue (SNT), 27 malignant melanomas (MM), 20 cutaneous squamous cell carcinomas (cSCC), and 24 pleomorphic dermal sarcomas (PDS) using mass spectrometry. We explored a potential cell of origin in PDS and validated our findings using publicly available single-cell sequencing data. By correlating tumor purity (TP), inferred by both RNA- and DNA-sequencing, to protein abundance, we found that fibroblasts shared most of the proteins correlating to TP. This observation could also be made using publicly available SNT single cell sequencing data. Moreover, we studied relevant pathways of receptor/ligand (R/L) interactions. Analysis of R/L interactions revealed distinct pathways in cSCC, MM and PDS, with a prominent role of PDGFRB-PDGFD R/L interactions and upregulation of PI3K/AKT signaling pathway. By studying differentially expressed proteins between cSCC and PDS, markers such as MAP1B could differentiate between these two entities. To this end, we studied proteins associated with immunosuppression in PDS, uncovering that immunologically cold PDS cases shared a “negative regulation of interferon-gamma signaling” according to overrepresentation analysis.

Prostate cancer pathology plays a crucial role in clinical management but is time-consuming. Artificial intelligence (AI) shows promise in detecting prostate cancer and grading patterns. We tested an AI-based digital twin of a pathologist, vPatho, on 2603 histological images of prostate tissue stained with hematoxylin and eosin. We analyzed various factors influencing tumor grade discordance between the vPatho system and six human pathologists. Our results demonstrated that vPatho achieved comparable performance in prostate cancer detection and tumor volume estimation, as reported in the literature. The concordance levels between vPatho and human pathologists were examined. Notably, moderate to substantial agreement was observed in identifying complementary histological features such as ductal, cribriform, nerve, blood vessel, and lymphocyte infiltration. However, concordance in tumor grading decreased when applied to prostatectomy specimens (κ = 0.44) compared to biopsy cores (κ = 0.70). Adjusting the decision threshold for the secondary Gleason pattern from 5 to 10% improved the concordance level between pathologists and vPatho for tumor grading on prostatectomy specimens (κ from 0.44 to 0.64). Potential causes of grade discordance included the vertical extent of tumors toward the prostate boundary and the proportions of slides with prostate cancer. Gleason pattern 4 was particularly associated with this population. Notably, the grade according to vPatho was not specific to any of the six pathologists involved in routine clinical grading. In conclusion, our study highlights the potential utility of AI in developing a digital twin for a pathologist. This approach can help uncover limitations in AI adoption and the practical application of the current grading system for prostate cancer pathology.

Claudin 18.2 (CLDN18.2) is a therapeutically relevant biomarker in esophageal (EAC) -and gastric adenocarcinoma (GAC). Little is known about its heterogeneity within the primary tumor and corresponding metastases and how many biopsies are needed to determine the true CLDN18.2 status of a tumor. CLDN18.2 was assessed in 1,283 patients (822 EAC, 461 GAC), using the antibody clone 43–14 A. Eight virtual endoscopic biopsies were taken from digitized whole tumor blocks. 204 of 822 EAC (24.8%) and 132 of 461 GAC (28,6%) were positive for CLDN18.2. In GAC, CLDN18.2 expression showed a trend towards less invasive growth ( p  = 0.02) and less common lymph node metastasis ( p  = 0.07) and was more often observed within the EBV-associated subtype ( p  = 0.01). When comparing the expression in one biopsy with the whole tissue section, the sensitivity for CLDN18.2 was low (58.8%). The sensitivity increased to a maximum of 76.5% with 6 and 8 biopsies (positive likelihood ratio = 17.8). Discordant expression between primary tumor and corresponding local lymph node metastasis was observed in 18.5%. This study highlights that CLDN18.2 is a heterogeneously expressed biomarker, within the tumor tissue as well as in primary tumor and corresponding lymph node metastasis. In case of CLN18.2-negative results the representativity of the biopsies must be critically assessed and a re-biopsy might be recommendable.

High-precision pixel-level annotation has been a major bottleneck in computational pathology due to its time-consuming nature and reliance on expert knowledge. Semi-supervised learning (SSL) provides a promising approach to alleviate this challenge by leveraging large amounts of unlabeled data. However, existing pseudo-labeling-based SSL methods often overlook intrinsic properties, such as inter-case similarities, which are critical for generating accurate pseudo-labels in complex tissue environments. In this study, we propose a Swarm-of-Models (S–o-M) SSL framework that dynamically selects “morphology expert” models (i.e., models specialized in recognizing specific tissue structures) for each unlabeled whole-slide image (WSI) based on similarity, thereby improving the reliability of pseudo-labeling for semantic segmentation tasks. In an evaluation on a large international dataset (multi-class tissue segmentation algorithm for colorectal domain), our approach outperforms traditional supervised and semi-supervised strategies by improving the Dice score by 3.6% for tumor segmentation and 2.1% for tumor/tumor stroma segmentation. Ablation studies performed with different numbers of annotated and unannotated WSIs, as well as training in a monocentric training scenario, further confirm the robustness and superior performance of the proposed S–o-M framework. These findings highlight the value of incorporating case-to-case similarities into SSL strategies to build more effective and general computational pathology models.

Background The APLNR (apelin receptor) has been shown to be an essential gene for cancer immunotherapy, with deficiency in APLNR leading to immunotherapy failure. The aim of this study is to investigate the expression of APLN (apelin) and APLNR in patients with renal cell carcinoma (RCC), and its association with clinicopathological parameters and survival. Methods Three well-characterised patient cohorts with RCC were used: Study cohort 1 (clear-cell RCC; APLN/APLNR mRNA expression; n  = 166); TCGA validation cohort (clear-cell RCC; APLN/APLNR mRNA expression; n  = 481); Study cohort 2 (all RCC subtypes; APLNR protein expression/immunohistochemistry; n  = 300). Associations between mRNA/protein expression and clinicopathological variables/patients’ survival were tested statistically. Results While APLN showed only very weak association with tumour histological grade (TCGA cohort), APLNR/mRNA protein expression correlate significantly with ccRCC aggressiveness. APLNR is expressed in tumour vasculature and tumour cells at different levels, and these expression levels associate with tumour aggressiveness in opposing directions. APLNR expression was negatively correlated with PD-L1 expression by tumour cells in a subset of patients with ccRCC. APLNR expression in either compartment is an independent prognostic factor for survival of patients with ccRCC. Conclusion The APLNR/APLN-system appears to play an important role in ccRCC, warranting further clinical investigation.

Background Pathology departments generate large volumes of unstructured data as free-text diagnostic reports. Converting these reports into structured formats for analytics or artificial intelligence projects requires substantial manual effort by specialized personnel. While recent studies show promise in using advanced language models for structuring pathology data, they primarily rely on proprietary models, raising cost and privacy concerns. Additionally, important aspects such as prompt engineering and model quantization for deployment on consumer-grade hardware remain unaddressed. Methods We created a dataset of 579 annotated pathology reports in German and English versions. Six language models (proprietary: GPT-4; open-source: Llama2 13B, Llama2 70B, Llama3 8B, Llama3 70B, and Qwen2.5 7B) were evaluated for their ability to extract eleven key parameters from these reports. Additionally, we investigated model performance across different prompt engineering strategies and model quantization techniques to assess practical deployment scenarios. Results Here we show that open-source language models extract structured data from pathology reports with high precision, matching the accuracy of proprietary GPT-4 model. The precision varies significantly across different models and configurations. These variations depend on specific prompt engineering strategies and quantization methods used during model deployment. Conclusions Open-source language models demonstrate comparable performance to proprietary solutions in structuring pathology report data. This finding has significant implications for healthcare institutions seeking cost-effective, privacy-preserving data structuring solutions. The variations in model performance across different configurations provide valuable insights for practical deployment in pathology departments. Our publicly available bilingual dataset serves as both a benchmark and a resource for future research. Plain language summary Pathology departments produce many diagnostic reports as free text, which is hard to analyze or use in research and computer projects. Converting this free text into more standard organized information like test results or diagnoses, makes it easier to use. This task often requires human experts and takes time. Large language models (LLMs), which are advanced computer systems designed to understand and generate human-like text, might simplify this process. Here, we tested six LLMs, including freely available models and the commercial GPT-4 model, using 579 pathology reports in English and German. Our results show that freely available models can perform as well as commercial, providing a cheaper solution while avoiding privacy concerns. The shared dataset will support future research in pathology data processing. Grothey et al. examine the performance of large language models in structuring pathology reports. Findings demonstrate similar accuracy between commercial and open-source models providing a cost-effective, privacy-conscious solution to extract structured data with high precision from bilingual datasets.