Florian Thamm

Dr.-Ing. Florian Thamm


Department of Computer Science
Chair of Computer Science 5 (Pattern Recognition)

Room: Room 04.133
Martensstr. 3
91058 Erlangen

Office hours

Each week Mo, Tu,

Please do not call me in the office. Write me an E-Mail instead 🙂


Wanna’ see how we synthesized 57 300 000 000 (!) patients out of 151?


Check out our most recent work “Building Brains” [arXiv] accepted at MICCAI to see how we pushed the detection of LVOs even further! Open the Projects tab below ; )


Building Brains aka The Sandwich Method

Building Brains: Subvolume Recombination for Data Augmentation in Large Vessel Occlusion Detection

Ischemic strokes are often caused by large vessel occlusions (LVOs), which can be visualized and diagnosed with Computed Tomography Angiography scans. As time is brain, a fast, accurate and automated diagnosis of these scans is desirable. Human readers compare the left and right hemispheres in their assessment of strokes. A large training data set is required for a standard deep learning-based model to learn this strategy from data. As labeled medical data in this field is rare, other approaches need to be developed. To both include the prior knowledge of side comparison and increase the amount of training data, we propose an augmentation method that generates artificial training samples by recombining vessel tree segmentations of the hemispheres or hemisphere subregions from different patients. The subregions cover vessels commonly affected by LVOs, namely the internal carotid artery (ICA) and middle cerebral artery (MCA). In line with the augmentation scheme, we use a 3D-DenseNet fed with task-specific input, fostering a side-by-side comparison between the hemispheres. Furthermore, we propose an extension of that architecture to process the individual hemisphere subregions. All configurations predict the presence of an LVO, its side, and the affected subregion. We show the effect of recombination as an augmentation strategy in a 5-fold cross validated ablation study. We enhanced the AUC for patient-wise classification regarding the presence of an LVO of all investigated architectures. For one variant, the proposed method improved the AUC from 0.73 without augmentation to 0.89. The best configuration detects LVOs with an AUC of 0.91, LVOs in the ICA with an AUC of 0.96, and in the MCA with 0.91 while accurately predicting the affected side.



Can you tell which one is fake?

No? Don’t worry*, radiologists neither can 😉 Check out our most recent work “SyNCCT” published at MICCAI to see how we synthesized CT images from angiography scans.

SyNCCT: Synthetic Non-contrast Images of the Brain from Single-Energy Computed Tomography Angiography

By injecting contrast agent during a CT acquisition, the vascular system can be enhanced. This acquisition type is known as CT Angiography (CTA). However, due to typically lower dose levels of CTA scans compared to non-contrast CT acquisitions (NCCT) and the employed reconstruction designed specifically for vessel reconstruction, soft tissue contrast in the brain parenchyma is usually subpar. Hence, an NCCT scan is preferred for the visualization of such tissue. We propose SyNCCT, an approach which synthesizes NCCT images from the CTA domain by removing enhanced vessel structures and improving soft tissue contrast. Contrary to virtual non-contrast (VNC) images based on dual energy scans, which target the physically accurate removal of iodine rather than generating a realistic NCCT with improved gray/white matter separation, our approach only requires a conventional single-energy acquisition. By design, our method integrates prior domain knowledge and employs residual learning as well as a discriminator to achieve perceptual realism. In our data set of patients with ischemic stroke, the absolute differences in automatic ASPECT scoring, which rates early signs of an occlusion in the anterior circulation on a scale from 0 (most severe) to 10 (no signs), was 0.78 ± 0.75 (median of 1) when comparing our SyNCCT to the real NCCT images. Qualitatively, realistic appearance of the images was confirmed by means of a Turing test with a radiologist, who classified 64% of 64 (32 real, 32 generated) images correctly. Two other physicians classified 65% correctly, on average.


*And by the way, the left one was fake 😉


VirtualDSA++: Automated Segmentation, Vessel Labeling, Occlusion Detection and Graph Search on CT-Angiography Data

Computed Tomography Angiography (CTA) is one of the most commonly used modalities in the diagnosis of cerebrovascular diseases like ischemic strokes. Usually, the anatomy of interest in ischemic stroke cases is the Circle of Willis and its peripherals, the cerebral arteries, as these vessels are the most prominent candidates for occlusions. The diagnosis of occlusions in these vessels remains challenging, not only because of the large amount of surrounding vessels but also due to the large number of anatomical variants. We propose a fully automated image processing and visualization pipeline, which provides a full segmentation and modelling of the cerebral arterial tree for CTA data. The model itself enables the interactive masking of unimportant vessel structures e.g. veins like the Sinus Sagittalis, and the interactive planning of shortest paths meant to be used to prepare further treatments like a mechanical thrombectomy. Additionally, the algorithm automatically labels the cerebral arteries (Middle Cerebral Artery left and right, Anterior Cerebral Artery, Posterior Cerebral Artery left and right) and detects occlusions or interruptions in these vessels. The proposed pipeline does not require a prior non-contrast CT scan and achieves a comparable segmentation appearance as in a Digital Subtraction Angiography (DSA).


A 1minute Teaser 😉

And here is more Material

Watch the playlist with 3 videos on youtube https://youtu.be/b_IshmarO7E?list=PLJgm15h1F7z5pvrb6qHB9Z0QHbcALefQj

Academic CV

  • 11/2019 – 10/2023
    Dr.-Ing Computer Science at Pattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg
  • 04/2017 – 10/2019
    M.Sc. Computer Science, 1.1, Friedrich-Alexander-Universität Erlangen-Nürnberg
  • 10/2013 – 03/2017
    B.Eng. Medical Engineering, 1.4, Technische Hochschule Nürnberg

Working Experience

Siemens AG, Erlangen

  • Since 10/2022
    Machine Learning Engineer for Computer Vision

Siemens Healthcare GmbH, Forchheim 

  • 11/2019 – 09/2022
    Ph.D. Candidate
  • 03/2019 – 09/2019
    Master’s Thesis Student
  • 03/2017 – 03/2019
    Working Student
  • 10/2016 – 04/2017
    Bachelor’s Thesis Student
  • 03/2016 – 10/2016
    Working Student
  • 10/2015 – 03/2016


  • Since 01/2022
  • 01/2020 – 01/2022
    Research Assistant for CTA Image Analysis
    Exercise Instructor of the Deep Learning Course
  • 10/2018 – 12/2020
    Teaching Assistant, Exercise Instructor of the Deep Learning Course

Technische Hochschule Nürnberg

  • 03/2015 – 10/2015
    E-Learning Tutor
  • 10/2014 – 10/2015
    Teaching Assistant









No matching records found.

Students Theses

Student Title Type Status
Lukas Rosteck AI-based Anomaly Detection in Process Signals for Condition Monitoring of Industrial Machines MA thesis running
Peter Herbst Unsupervised Contextual Anomaly Detection in Frequency Converter Data MA thesis finished
Sebastian Müller Detection of localized necking in Hydraulic Bulge Tests using Deep Learning Methods MA thesis finished
Eduard Reger Modeling of Randomized Cerebrovascular Trees for Artifical Data Generation using Blender MA thesis finished
Ankitha Rama Krishna Reddy GAN-based Synthetic Chest X-ray Generation for Training Lung Disease Classification Systems MA thesis finished
Benjamin Geißler Interpolation of ARAMIS Grids and Analysis of Numerical Stability on Deep Learning Methods Project finished
Dmitrij Vinokour Reinforcement Learning in Finance – Add and adapt the DDQN to an existing Reinforcement Learning Framework Project finished
Julian Jendryka Advanced Model Architectures for Interactive Segmentation and Segmentation Enhancement in CT Images MA thesis finished
Jonas Schauer Cerebral Vessel Tree Estimation from Non-Contrast CT using Deep Learning Methods MA thesis finished
Linda Vorberg Detection of Pulmonary Embolisms in NCCT Data using Deep Learning Methods MA thesis finished
Laurin Schuster Graph Augmentation using Cond.-GANs Project finished
Tasnim Nova Post-Processing of DTF-Skeletonizations Project running
Jad Kassam Detection of Large Vessel Occlusions using Graph Deep Learning MA thesis finished
Amit Kumar Sharma Prediction of Steam Turbine Blade Vibration Amplitudes using Machine Learning Methods MA thesis finished
Annette Schwarz Real-Time Prospective Respiratory Triggering for Free-Breathing Lung Computed Tomography MA thesis finished
Melanie Kienberger Predictive Maintenance for SINAMICs Frequency Converter MA thesis finished
Dmitrij Vinokour Detection of Hand Drawn Electrical Circuit Diagrams and their Components using Deep Learning Methods and Conversion into LTspice Format MA thesis finished
Stephanie Mehltretter Augmentation of CT Images by Variation of Non-Rigid Deformation Vector Field Amplitudes Project running
Antonia Popp Thrombus Detection in Non-Contrast Head CT using Graph Deep Learning MA thesis finished
Leonhard Rist Geometric Deep Learning for Multifocal Diseases MA thesis finished
Jiayue Zhao Optimization of the Input Resolution for Dermoscopy Image Classification Tasks MA thesis finished
Lukas Folle Classification of Rotator Cuff Tears in MRI using Neural Networks MA thesis finished