Nils Gessert

 

 

 

 

 

 

 Research Assistent

 +49 (0)40 42878 3389

  E.3088

  nils.gesserttuhh.de

2017 - today MTEC Institute / TUHH
2014 - 2016 Northern Institute of Technology Management
2011 - 2017 Hamburg University of Technology

Interests

  • Deep Learning for Medical Image Analysis
  • 3D CNN Architectures for Optical Coherence Tomography (OCT) Data
  • OCT-based Tracking, Force Estimation and Disease Classification
  • Calibration Strategies for OCT with Robotic Systems

Roles

  • Research assistant
  • PhD student

Projects

OCT_principle.png

The principle of Optical Coherence Tomography (OCT). The infrared light-based imaging modality allows for volumetric depth scans with micrometer level resolution.

CNN_principle.png

Deep learning example for OCT. B-Scans can be treated as 2D images and used for training with a CNN.

tracking_example.png

Application example: C-Scans are volumetric data which requires 3D CNN architectures, e.g. for tracking. Thousands of OCT volumes are acquired and labeld with robot poses. The trained CNN can infer the marker's pose from only an acuqired volume.

force_estimation.png

Application example: Visual deformation of tissue in C-Scans can be used for force estimation. This allows for sensorless haptic feedback in teleoperation.

Member of programm committees and editorial boards