Fabian Hörst

Fabian Hörst

Auto-generated profile, based on publicly available information

Fabian Hörst

Institute for AI in Medicine (IKIM) University Hospital Essen (AöR) Essen Germany

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
16 March 2024 in Medical Image Analysis
Verified
Journal Article
Open access
FHFabian Hörst
Fabian Hörst
MRMoritz Rempe
Moritz Rempe
LHLukas Heine
Lukas Heine
CSConstantin Seibold
Constantin Seibold
JKJulius Keyl
Julius Keyl
JKJens Kleesiek
Jens Kleesiek
Nuclei detection and segmentation in hematoxylin and eosin-stained (H&E) tissue images are important clinical tasks and crucial for a wide range of applications. However, it is a challenging task due to nuclei variances in staining and size, overlapping boundaries, and nuclei clustering. While convolutional neural networks have been extensively used for this task, we explore the potential of Transformer-based networks in combination with large scale pre-training in this domain. Therefore, we introduce a new method for automated instance segmentation of cell nuclei in digitized tissue samples using a deep learning architecture based on Vision Transformer called CellViT. CellViT is trained and evaluated on the PanNuke dataset, which is one of the most challenging nuclei instance segmentation datasets, consisting of nearly 200,000 annotated nuclei into 5 clinically important classes in 19 tissue types. We demonstrate the superiority of large-scale in-domain and out-of-domain pre-trained Vision Transformers by leveraging the recently published Segment Anything Model and a ViT-encoder pre-trained on 104 million histological image patches - achieving state-of-the-art nuclei detection and instance segmentation performance on the PanNuke dataset with a mean panoptic quality of 0.50 and an F1-detection score of 0.83. The code is publicly available at https://github.com/TIO-IKIM/CellViT.

9 January 2026 in Computer Methods and Programs in Biomedicine
Verified
Journal Article
Open access
FHFabian Hörst
Fabian Hörst
MRMoritz Rempe
Moritz Rempe
HBHelmut Becker
Helmut Becker
LHLukas Heine
Lukas Heine
JKJulius Keyl
Julius Keyl
JKJens Kleesiek
Jens Kleesiek
Deep learning-based cell segmentation and classification methods in digital pathology are critical for diagnostics but are hampered by models that require extensive annotated datasets, are computationally expensive, and lack adaptability to new cell types. This creates a significant bottleneck in research and clinical workflows. This study introduces CellViT++, a data-efficient and lightweight framework for generalized cell segmentation that allows for rapid adaptation to novel cell taxonomies with minimal data. CellViT++ leverages a Vision Transformer with a frozen pretrained foundation model for segmentation. It simultaneously extracts deep cell embeddings from the transformer tokens during the forward pass at no extra computational cost. To adapt to new cell types, only a lightweight classifier is trained on these embeddings, bypassing the need to retrain the segmentation model. We also demonstrate an automated workflow to generate training data from registered H&E and immunofluorescence (IF) slides. The framework was validated on seven public datasets. The framework achieves remarkable zero-shot segmentation results and data efficiency. On the CoNSeP dataset for colon cancer, we achieved superior results with only 10% of the training data. On all other datasets, we outperformed competing methods or at least approached their performance, all in one model. The classifier approach, based on zero-shot segmentation models, drastically reduces computational costs, with training times of minutes versus hours for baseline models, decreasing CO2 emission by 96.93%. Models trained on automatically generated labels from IF-staining achieved performance comparable to (lymphocytes, ΔF1:-0.042) or even exceeding (plasma cells, ΔF1:+0.108) those trained on expert-annotated datasets. CellViT++ provides a robust and efficient open-source framework that addresses key limitations in computational pathology by decoupling segmentation from classification. Its ability to adapt to new cell types with minimal data and its support for automated dataset generation from IF slides significantly reduce the reliance on time-consuming expert annotation. This work provides a foundational tool to accelerate research, enhance diagnostic workflows, and enable deeper cohort analysis. The code is available under https://github.com/TIO-IKIM/CellViT-plus-plus and as a PyPI package.

11 May 2023 in Computerized Medical Imaging and Graphics
Verified
Journal Article
Open access
OEOliver Ester
Oliver Ester
FHFabian Hörst
Fabian Hörst
CSConstantin Seibold
Constantin Seibold
JKJulius Keyl
Julius Keyl
STSaskia Ting
Saskia Ting
JKJens Kleesiek
Jens Kleesiek
The segmentation of histopathological whole slide images into tumourous and non-tumourous types of tissue is a challenging task that requires the consideration of both local and global spatial contexts to classify tumourous regions precisely. The identification of subtypes of tumour tissue complicates the issue as the sharpness of separation decreases and the pathologist's reasoning is even more guided by spatial context. However, the identification of detailed tissue types is crucial for providing personalized cancer therapies. Due to the high resolution of whole slide images, existing semantic segmentation methods, restricted to isolated image sections, are incapable of processing context information beyond. To take a step towards better context comprehension, we propose a patch neighbour attention mechanism to query the neighbouring tissue context from a patch embedding memory bank and infuse context embeddings into bottleneck hidden feature maps. Our memory attention framework (MAF) mimics a pathologist's annotation procedure - zooming out and considering surrounding tissue context. The framework can be integrated into any encoder-decoder segmentation method. We evaluate the MAF on two public breast cancer and liver cancer data sets and an internal kidney cancer data set using famous segmentation models (U-Net, DeeplabV3) and demonstrate the superiority over other context-integrating algorithms - achieving a substantial improvement of up to 17% on Dice score. The code is publicly available at https://github.com/tio-ikim/valuing-vicinity.

23 August 2024 in European Journal of Cancer
Verified
Journal Article
Open access
GLGeorg Lodde
Georg Lodde
FZFang Zhao
Fang Zhao
RHRudolf Herbst
Rudolf Herbst
PTPatrick Terheyden
Patrick Terheyden
Jochen UtikalJochen Utikal
Jochen Utikal
SUSelma Ugurel
Selma Ugurel
Immune checkpoint inhibition (ICI) currently is the most effective treatment to induce durable responses in metastatic melanoma. The aims of this study are the characterization of patients with early, late and non-response to ICI and analysis of survival outcomes in a real-world patient cohort. Patients who received PD-1-based immunotherapy for non-resectable stage-IV melanoma in any therapy line were selected from the prospective multicenter real-world DeCOG study ADOREG-TRIM (NCT05750511). Patients showing complete (CR) or partial (PR) response already during the first 3 months of treatment (Early Responders, EarlyR) were compared to patients showing CR/PR at a later time (Late Responders, LateR), a stable disease (SD) and to patients showing progressive disease (Non-Responders, NonR). Of 522 patients, 8.2 % were EarlyR (n = 43), 19.0 % were LateR (n = 99), 37.0 % had a SD (n = 193) and 35.8 % were NonR (n = 187). EarlyR, LateR and SD patients had comparable baseline characteristics. Multivariate logbinomial regression analyses adjusted for age and sex revealed positive tumor PD-L1 (RR=1.99, 95 %-CI=1.14-3.46, p = 0.015), and normal serum CRP (RR=1.59, 95 %-CI=0.93-2.70, p = 0.036) as independently associated with the achievement of an early response compared to NonR. The median progression-free and overall survival was 46.0 months (95 % CI 19.1; NR) and 47.8 months (95 %-CI 36.9; NR) for EarlyR, NR (95 %-CI NR; NR) for LateR, 8.1 months (7.0; 10.4) and 35.4 months (29.2; NR) for SD, and 2.0 months (95 %-CI 1.9; 2.1) and 6.1 months (95 %-CI 4.6; 8.8) for NonR patients. Less than 10 % of metastatic melanoma patients achieved an early response during the first 3 months of PD-1-based immunotherapy. Early responders were not superior to late responders in terms of response durability and survival.

1 August 2023 in JCO Clinical Cancer Informatics
Verified
Journal Article
Closed access
FHFabian Hörst
Fabian Hörst
STSaskia Ting
Saskia Ting
SLSven-Thorsten Liffers
Sven-Thorsten Liffers
KPKelsey L. Pomykala
Kelsey L. Pomykala
Katja SteigerKatja Steiger
Katja Steiger
JKJens Kleesiek
Jens Kleesiek
PURPOSE: Quantifying treatment response to gastroesophageal junction (GEJ) adenocarcinomas is crucial to provide an optimal therapeutic strategy. Routinely taken tissue samples provide an opportunity to enhance existing positron emission tomography-computed tomography (PET/CT)–based therapy response evaluation. Our objective was to investigate if deep learning (DL) algorithms are capable of predicting the therapy response of patients with GEJ adenocarcinoma to neoadjuvant chemotherapy on the basis of histologic tissue samples. METHODS: This diagnostic study recruited 67 patients with I-III GEJ adenocarcinoma from the multicentric nonrandomized MEMORI trial including three German university hospitals TUM (University Hospital Rechts der Isar, Munich), LMU (Hospital of the Ludwig-Maximilians-University, Munich), and UME (University Hospital Essen, Essen). All patients underwent baseline PET/CT scans and esophageal biopsy before and 14-21 days after treatment initiation. Treatment response was defined as a ≥35% decrease in SUVmax from baseline. Several DL algorithms were developed to predict PET/CT-based responders and nonresponders to neoadjuvant chemotherapy using digitized histopathologic whole slide images (WSIs). RESULTS: The resulting models were trained on TUM (n = 25 pretherapy, n = 47 on-therapy) patients and evaluated on our internal validation cohort from LMU and UME (n = 17 pretherapy, n = 15 on-therapy). Compared with multiple architectures, the best pretherapy network achieves an area under the receiver operating characteristic curve (AUROC) of 0.81 (95% CI, 0.61 to 1.00), an area under the precision-recall curve (AUPRC) of 0.82 (95% CI, 0.61 to 1.00), a balanced accuracy of 0.78 (95% CI, 0.60 to 0.94), and a Matthews correlation coefficient (MCC) of 0.55 (95% CI, 0.18 to 0.88). The best on-therapy network achieves an AUROC of 0.84 (95% CI, 0.64 to 1.00), an AUPRC of 0.82 (95% CI, 0.56 to 1.00), a balanced accuracy of 0.80 (95% CI, 0.65 to 1.00), and a MCC of 0.71 (95% CI, 0.38 to 1.00). CONCLUSION: Our results show that DL algorithms can predict treatment response to neoadjuvant chemotherapy using WSI with high accuracy even before therapy initiation, suggesting the presence of predictive morphologic tissue biomarkers.

11 May 2024 in Computer Methods and Programs in Biomedicine
Verified
Journal Article
Open access
JRJohannes Raufeisen
Johannes Raufeisen
KXKunpeng Xie
Kunpeng Xie
FHFabian Hörst
Fabian Hörst
TBTill Braunschweig
Till Braunschweig
JLJianning Li
Jianning Li
BPBehrus Puladi
Behrus Puladi
Cell segmentation in bright-field histological slides is a crucial topic in medical image analysis. Having access to accurate segmentation allows researchers to examine the relationship between cellular morphology and clinical observations. Unfortunately, most segmentation methods known today are limited to nuclei and cannot segment the cytoplasm. We present a new network architecture Cyto R-CNN that is able to accurately segment whole cells (with both the nucleus and the cytoplasm) in bright-field images. We also present a new dataset CytoNuke, consisting of multiple thousand manual annotations of head and neck squamous cell carcinoma cells. Utilizing this dataset, we compared the performance of Cyto R-CNN to other popular cell segmentation algorithms, including QuPath's built-in algorithm, StarDist, Cellpose and a multi-scale Attention Deeplabv3+. To evaluate segmentation performance, we calculated AP50, AP75 and measured 17 morphological and staining-related features for all detected cells. We compared these measurements to the gold standard of manual segmentation using the Kolmogorov-Smirnov test. Cyto R-CNN achieved an AP50 of 58.65% and an AP75 of 11.56% in whole-cell segmentation, outperforming all other methods (QuPath 19.46/0.91%; StarDist 45.33/2.32%; Cellpose 31.85/5.61%, Deeplabv3+ 3.97/1.01%). Cell features derived from Cyto R-CNN showed the best agreement to the gold standard (D¯=0.15) outperforming QuPath (D¯=0.22), StarDist (D¯=0.25), Cellpose (D¯=0.23) and Deeplabv3+ (D¯=0.33). Our newly proposed Cyto R-CNN architecture outperforms current algorithms in whole-cell segmentation while providing more reliable cell measurements than any other model. This could improve digital pathology workflows, potentially leading to improved diagnosis. Moreover, our published dataset can be used to develop further models in the future.