Aktuelle Publikationen

This work shows the application of explainable AI (XIA) methods to resolve common challenges encountered in deep learning-based digital histopathology analyses. We investigate three types of biases and show that XAI techniques are helpful and highly relevant tool for the development and the deployment phases of real-world applications in digital pathology.

Coherent comb lasers may serve as a source for multiwavelength modulators in short reach transmission, or for phase controlled OFDM channels in long reach. We explore and compare quantum dot (QD) and quantum well (QW) lasers with more than 33 channels in the DWDM 50 GHz grid, thus enabling > 1 Tb/s optical transmission. In addition, the mode-locked devices can be applied as pulse sources with < 500 fs pulses by using a simple SMF.

Deep models are also being increasingly applied in distributed settings, where the data are separated by limited communication channels and privacy constraints. To address the challenges, a wide range of training and evaluation schemes have been developed, which require the communication of neural network parametrizations. This paper gives an overview over the recent advancements and challenges.

We determine the absorption and scattering coefficients of cholesteatoma and bone. In the near-UV and visual spectrum, clear differences exist between both tissues. These differences reveal the future possibility to detect and identify, automatically or semi-automatically, cholesteatoma tissue for active treatment decisions during image-guided surgery leading to a better surgical outcome.

Photo-realistic modeling and rendering of humans is extremely important for VR environments. While purely computer graphics modeling can achieve highly realistic human models, achieving real photo-realism with these models is computationally extremely expensive. Therefore, we enrich volumetric video with semantics and animation properties to make photo-realistic volumetric video animatable.

This paper presents DeepCABAC, a universal compression algorithm for deep neural networks (DNNs) that through its adaptive, context-based rate modeling, allows an optimal quantization and coding of parameters. It compresses DNNs up to 5% of their original size with no accuracy loss and has been selected as basic compression technology for the emerging MPEG-7 part 17 standard on DNN compression.

Inferring the properties of protein from its amino acid sequence is a key problem in bioinformatics. We put forward UDSMProt, a universal deep sequence model that is pretrained on a language modeling task and finetuned on protein classification tasks. For enzyme class prediction, gene ontology prediction and remote homology and fold detection, we reach sofa performance from the sequence alone.

We enable dynamic projection mapping on 3d objects and present a model-based tracking strategy for projector camera-systems, which directly takes advantage from the projected content for pose estimation. Our method establishes a distortion free projection by first analysing and then correcting the distortion of the projection in a closed loop. Therefor an optical flow-based model is extended to the geometry of a projector-camera unit. We evaluate our procedure with real and synthetic images and obtain very precise registration results.

We analyze five healthy soft tissue types as well as tumor tissue in terms of absorption and scatter coefficients in the spectral range of 250 to 800 nm. The spectral properties of the analyzed tissue types show relevant differences in this specific spectral range. This knowledge can be used for different medical and biomedical application areas as computer-aided diagnostics, intraoperative image guidance and histological analysis.

This paper introduces the current running project MultiARC, which has the aim to combine stereoscopic measurements and scene reconstruction with hyperspectral tissue differentiation in a surgical microscope and 3D endoscope for image-guided surgery.