Fabian Hörst

Fabian Hörst

PhD Student in Medical Artificial Intelligence

Institute for Artificial Intelligence in Medicine

Biography

I’m a Ph.D. student in the field of medical artificial intelligence, with a primary focus on computer vision and computational pathology. My passion lies in harnessing the power of AI to revolutionize healthcare. Through my research, I strive to develop innovative solutions for diagnosing and understanding diseases, contributing to the advancement of medical science. Join me on this journey to bridge the gap between data science and pathology, and help shape the future of medical AI.

Interests
  • Artificial Intelligence
  • Computer Vision
  • Medical Machine Learning
  • Computational Pathology
  • Digitizing Health Care
Education
  • PhD in Medical Physics, 2022 - ongoing

    Institute for Artificial Intelligence in Medicine and Technical University of Dortmund

  • MSc in Electrical Engineering (Information Technology), 2022

  • BSc in Electrical Engineering (Information Technology), 2019

Skills

Technical
Programming
Data Science
Computer Vision
Languages/Frameworks/Toolstack
Python
PyTorch
NumPy, Numba, Pandas
OpenCV, Scikit-learn
Docker
Kubeflow
Javascript
Softskills
Academic Writing

Projects

*
PathoPatcher
PathoPatcher is a Python project designed for accelerating Whole Slide Image Preprocessing, employing AI-based preprocessing techniques with features like annotation handling, color normalization, and configurable parameters - Accelerating Artificial Intelligence Based Whole Slide Image Analysis with an Optimized Preprocessing Pipeline
PathoPatcher
PathoViewer
Our Web-based Whole Slide Imaging (WSI) Viewer offers efficient access to gigapixel digital pathology slides from various vendor formats, including DICOM. It supports handling millions of cell annotations for detailed analysis. Built on a scalable FastAPI backend with a responsive JavaScript frontend, it ensures quick loading and smooth navigation.
PathoViewer
CellViT
CellViT sets the benchmark for nuclei instance segmentation, surpassing existing methods on the PanNuke dataset with remarkable performance improvements. This cutting-edge project integrates the power of Vision Transformer (ViT) encoders, enhancing segmentation accuracy, and leverages a U-Net architecture for efficient feature extraction. With fast inference capabilities on gigapixel Whole Slide Images (WSI), CellViT promises to revolutionize the field of cell analysis in computational pathology.
CellViT

Recent & Upcoming Talks

Recent Works

Quickly discover relevant content by filtering publications.
(2024). Tumor likelihood estimation on MRI prostate data by utilizing k-Space information.

PDF Cite DOI

(2024). Accelerating Artificial Intelligence-based Whole Slide Image Analysis with an Optimized Preprocessing Pipeline.

PDF Cite DOI

(2024). Cyto R-CNN and CytoNuke Dataset: Towards reliable whole-cell segmentation in bright-field histological images.

PDF Cite DOI

(2024). Lean Study Host: Towards an Automated Pipeline for Multi-Center Study Hosting.

PDF Cite DOI

(2023). MedShapeNet--A Large-Scale Dataset of 3D Medical Shapes for Computer Vision.

PDF Cite Dataset

Contact

If you’re interested in discussing potential business collaborations or research partnerships, please don’t hesitate to get in touch.