A5046896448- Robotics and Sensor-Based Localization
- Medical Image Segmentation Techniques
- Augmented Reality Applications
- Surgical Simulation and Training
- Advanced Vision and Imaging
- Radiomics and Machine Learning in Medical Imaging
- Medical Imaging Techniques and Applications
- Anatomy and Medical Technology
- Advanced Image and Video Retrieval Techniques
- Advanced Neural Network Applications
- AI in cancer detection
- Medical Imaging and Analysis
- 3D Shape Modeling and Analysis
- Soft Robotics and Applications
- Human Pose and Action Recognition
- 3D Surveying and Cultural Heritage
- Domain Adaptation and Few-Shot Learning
- Advanced X-ray and CT Imaging
- Advanced Radiotherapy Techniques
- Generative Adversarial Networks and Image Synthesis
- Computer Graphics and Visualization Techniques
- Multimodal Machine Learning Applications
- Optical measurement and interference techniques
- Advanced MRI Techniques and Applications
- COVID-19 diagnosis using AI
Technical University of Munich
2016-2025
Munich Center for Machine Learning
2024-2025
Johns Hopkins University
2015-2024
Data:Lab Munich (Germany)
2024
University Hospital Bonn
2022
Klinikum rechts der Isar
2009-2019
Ludwig-Maximilians-Universität München
2004-2019
King Abdullah University of Science and Technology
2019
Metropolitan University
2018
Japan Science and Technology Agency
2017
Convolutional Neural Networks (CNNs) have been recently employed to solve problems from both the computer vision and medical image analysis fields. Despite their popularity, most approaches are only able process 2D images while data used in clinical practice consists of 3D volumes. In this work we propose an approach segmentation based on a volumetric, fully convolutional, neural network. Our CNN is trained end-to-end MRI volumes depicting prostate, learns predict for whole volume at once....
<h3>Importance</h3> Application of deep learning algorithms to whole-slide pathology images can potentially improve diagnostic accuracy and efficiency. <h3>Objective</h3> Assess the performance automated at detecting metastases in hematoxylin eosin–stained tissue sections lymph nodes women with breast cancer compare it pathologists' diagnoses a setting. <h3>Design, Setting, Participants</h3> Researcher challenge competition (CAMELYON16) develop solutions for node (November 2015-November...
This paper addresses the problem of estimating depth map a scene given single RGB image. We propose fully convolutional architecture, encompassing residual learning, to model ambiguous mapping between monocular images and maps. In order improve output resolution, we present novel way efficiently learn feature up-sampling within network. For optimization, introduce reverse Huber loss that is particularly suited for task at hand driven by value distributions commonly in Our composed...
We present a novel method for detecting 3D model instances and estimating their 6D poses from RGB data in single shot. To this end, we extend the popular SSD paradigm to cover full pose space train on synthetic only. Our approach competes or surpasses current state-of-the-art methods that leverage RGBD multiple challenging datasets. Furthermore, our produces these results at around 10Hz, which is many times faster than related methods. For sake of reproducibility, make trained networks...
This paper addresses the problem of recognizing free-form 3D objects in point clouds. Compared to traditional approaches based on descriptors, which depend local information around points, we propose a novel method that creates global model description oriented pair features and matches locally using fast voting scheme. The consists all represents mapping from feature space model, where similar are grouped together. Such representation allows much sparser object scene clouds, resulting very...
Attenuation correction (AC) of whole-body PET data in combined PET/MRI tomographs is expected to be a technical challenge. In this study, potential solution based on segmented attenuation map proposed and evaluated clinical PET/CT cases. <b>Methods:</b> Segmentation the into 4 classes (background, lungs, fat, soft tissue) was hypothesized sufficient for AC purposes. The segmentation applied CT-based maps from <sup>18</sup>F-FDG oncologic examinations 35 patients with 52...
Given the recent advances in depth prediction from Convolutional Neural Networks (CNNs), this paper investigates how predicted maps a deep neural network can be deployed for goal of accurate and dense monocular reconstruction. We propose method where CNN-predicted are naturally fused together with measurements obtained direct SLAM, based on scheme that privileges image locations SLAM approaches tend to fail, e.g. along low-textured regions, vice-versa. demonstrate use estimate absolute scale...
Staining and scanning of tissue samples for microscopic examination is fraught with undesirable color variations arising from differences in raw materials manufacturing techniques stain vendors, staining protocols labs, responses digital scanners. When comparing samples, normalization separation the images can be helpful both pathologists software. Techniques that are used natural fail to utilize structural properties stained produce distortions. The concentration cannot negative. Tissue...
We present a method for detecting 3D objects using multi-modalities. While it is generic, we demonstrate on the combination of an image and dense depth map which give complementary object information. It works in real-time, under heavy clutter, does not require time consuming training stage, can handle untextured objects. based efficient representation templates that capture different modalities, show many experiments commodity hardware our approach significantly outperforms state-of-the-art...
The lack of publicly available ground-truth data has been identified as the major challenge for transferring recent developments in deep learning to biomedical imaging domain. Though crowdsourcing enabled annotation large scale databases real world images, its application purposes requires a deeper understanding and hence, more precise definition actual task. fact that expert tasks are being outsourced non-expert users may lead noisy annotations introducing disagreement between users....
We present a method for real-time 3D object instance detection that does not require time-consuming training stage, and can handle untextured objects. At its core, our approach is novel image representation template matching designed to be robust small transformations. This robustness based on spread gradient orientations allows us test only subset of all possible pixel locations when parsing the image, represent with limited set templates. In addition, we demonstrate if dense depth sensor...
Optical coherence tomography (OCT) is used for non-invasive diagnosis of diabetic macular edema assessing the retinal layers. In this paper, we propose a new fully convolutional deep architecture, termed ReLayNet, end-to-end segmentation layers and fluid masses in eye OCT scans. ReLayNet uses contracting path blocks (encoders) to learn hierarchy contextual features, followed by an expansive (decoders) semantic segmentation. trained optimize joint loss function comprising weighted logistic...
EMPIRE10 (Evaluation of Methods for Pulmonary Image REgistration 2010) is a public platform fair and meaningful comparison registration algorithms which are applied to database intrapatient thoracic CT image pairs. Evaluation nonrigid techniques nontrivial task. This compounded by the fact that researchers typically test only on their own data, varies widely. For this reason, reliable assessment different has been virtually impossible in past. In work we present results launch phase...
In a wide range of semantic segmentation tasks, fully convolutional neural networks (F-CNNs) have been successfully leveraged to achieve the state-of-the-art performance. Architectural innovations F-CNNs mainly on improving spatial encoding or network connectivity aid gradient flow. this paper, we aim toward an alternate direction recalibrating learned feature maps adaptively, boosting meaningful features while suppressing weak ones. The recalibration is achieved by simple computational...