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.

Oesophageal adenocarcinoma and adenocarcinoma of the oesophagogastric junction are among the most common malignant epithelial tumours. Most patients receive neoadjuvant therapy before complete tumour resection. Histological assessment after resection includes identification of residual tumour tissue and areas of regressive tumour, data which are used to calculate a clinically relevant regression score. We developed an artificial intelligence (AI) algorithm for tumour tissue detection and tumour regression grading in surgical specimens from patients with oesophageal adenocarcinoma or adenocarcinoma of the oesophagogastric junction. We used one training cohort and four independent test cohorts to develop, train, and validate a deep learning tool. The material consisted of histological slides from surgically resected specimens from patients with oesophageal adenocarcinoma and adenocarcinoma of the oesophagogastric junction from three pathology institutes (two in Germany, one in Austria) and oesophageal cancer cohort of The Cancer Genome Atlas (TCGA). All slides were from neoadjuvantly treated patients except for those from the TCGA cohort, who were neoadjuvant-therapy naive. Data from training cohort and test cohort cases were extensively manually annotated for 11 tissue classes. A convolutional neural network was trained on the data using a supervised principle. First, the tool was formally validated using manually annotated test datasets. Next, tumour regression grading was assessed in a retrospective cohort of post-neoadjuvant therapy surgical specimens. The grading of the algorithm was compared with that of a group of 12 board-certified pathologists from one department. To further validate the tool, three pathologists processed whole resection cases with and without AI assistance. Of the four test cohorts, one included 22 manually annotated histological slides (n=20 patients), one included 62 sides (n=15), one included 214 slides (n=69), and the final one included 22 manually annotated histological slides (n=22). In the independent test cohorts the AI tool had high patch-level accuracy for identifying both tumour and regression tissue. When we validated the concordance of the AI tool against analyses by a group of pathologists (n=12), agreement was 63·6% (quadratic kappa 0·749; p<0·0001) at case level. The AI-based regression grading triggered true reclassification of resected tumour slides in seven cases (including six cases who had small tumour regions that were initially missed by pathologists). Use of the AI tool by three pathologists increased interobserver agreement and substantially reduced diagnostic time per case compared with working without AI assistance. Use of our AI tool in the diagnostics of oesophageal adenocarcinoma resection specimens by pathologists increased diagnostic accuracy, interobserver concordance, and significantly reduced assessment time. Prospective validation of the tool is required. North Rhine-Westphalia state, Federal Ministry of Education and Research of Germany, and the Wilhelm Sander Foundation.

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.

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.

Aims: Determining the origin of metastases is an important task of pathologists to allow for the initiation of a tumor-specific therapy. Recently, homeobox protein Hox-B13 (HOXB13) has been suggested as a new marker for the detection of prostatic origin. The aim of this study was to evaluate the diagnostic sensitivity of HOXB13 in comparison to commonly used immunohistochemical markers for prostate cancer. Materials and methods: Histologically confirmed prostate cancer lymph node metastases from 64 cases were used to test the diagnostic value of immunohistochemical markers: prostate specific antigen (PSA), Prostatic acid phosphatase (PSAP), prostate specific membrane antigen (PSMA), homeobox gene NKX3.1, prostein, androgen receptor (AR), HOXB13, and ETS-related gene (ERG). All markers were evaluated semi-quantitatively using Remmele's immune reactive score. Results: The detection rate of prostate origin of metastasis for single markers was 100% for NKX3.1, 98.1% for AR, 84.3% for PSMA, 80.8% for PSA, 66% for PSAP, 60.4% for HOXB13, 59.6% for prostein, and 50.0% for ERG. Conclusions: Our data suggest that HOXB13 on its own lacks sensitivity for the detection of prostatic origin. Therefore, this marker should be only used in conjunction with other markers, preferably the highly specific PSA. The combination of PSA with NKX3.1 shows a higher sensitivity and thus appears preferable in this setting.