A5061076941- Radiomics and Machine Learning in Medical Imaging
- Colorectal Cancer Surgical Treatments
- Colorectal Cancer Screening and Detection
- Aerogels and thermal insulation
- Radiation Therapy and Dosimetry
- Transition Metal Oxide Nanomaterials
- Advanced Radiotherapy Techniques
- Brain Tumor Detection and Classification
- Ga2O3 and related materials
- MRI in cancer diagnosis
- Advanced Chemical Physics Studies
- Diagnosis and treatment of tuberculosis
- Radiation Detection and Scintillator Technologies
- Radiation Effects in Electronics
- Prostate Cancer Diagnosis and Treatment
- Neural Networks and Applications
- Prostate Cancer Treatment and Research
- Protein Structure and Dynamics
- Pancreatic and Hepatic Oncology Research
- Energy, Environment, Agriculture Analysis
- Optical Imaging and Spectroscopy Techniques
- Advanced Neuroimaging Techniques and Applications
- Gas Sensing Nanomaterials and Sensors
- Image and Signal Denoising Methods
- Spectroscopy and Quantum Chemical Studies
Heidelberg University
2019-2024
University Medical Centre Mannheim
2024
University Hospital Heidelberg
2022-2024
Brigham and Women's Hospital
2019-2020
Harvard University
2019-2020
University of Massachusetts Lowell
2019-2020
Karlsruhe Institute of Technology
2016
Karlsruhe University of Education
2016
Abstract Objectives Achieving a consensus on definition for different aspects of radiomics workflows to support their translation into clinical usage. Furthermore, assess the perspective experts important challenges successful workflow implementation. Materials and methods The was achieved by multi-stage process. Stage 1 comprised screening, retrospective analysis with semantic mapping terms found in 22 definitions, compilation an initial baseline definition. Stages 2 3 consisted Delphi...
In this study, we evaluate the influence of normalization on performance deep learning networks for tumor segmentation and prediction pathological response locally advanced rectal cancer to neoadjuvant chemoradiotherapy. The techniques were applied a multicenter multimodal magnet resonance imaging data set consisting 201 patients recorded at six centers. We implemented investigated different methods (setting mean standard deviation, histogram matching, percentiles, combining percentiles...
Hydration free energy estimation of small molecules from all-atom simulations was widely investigated in recent years, as it provides an essential test molecular force fields and our understanding solvation effects.
Purpose We developed a new class of aerogel‐based thin‐film self‐powered radiation sensors employing high‐energy electron current (HEC) in periodic multilayer (high‐Z | polyimide aerogel (PA) low‐Z) electrode microstructures. Materials Low‐Z (Al) and high‐Z (Ta) electrodes were deposited on 50 μm‐thick PA films to obtain with Al‐PA‐Ta‐PA‐Al structures. Sensors tested x rays the 40–120 kVp range 2.5 MV, 6 MV‐FFF linac beams (TrueBeam, Varian). Performance PA‐HEC was compared commercial A12...
This retrospective study aims to evaluate the generalizability of a promising state-of-the-art multitask deep learning (DL) model for predicting response locally advanced rectal cancer (LARC) neoadjuvant chemoradiotherapy (nCRT) using multicenter dataset. To this end, we retrained and validated Siamese network with two U-Nets joined at multiple layers pre- post-therapeutic T2-weighted (T2w), diffusion-weighted (DW) images apparent diffusion coefficient (ADC) maps 83 LARC patients acquired...
Purpose To explore 3D printing for rapid development of prototype thin slab low‐Z/density ionization chamber arrays viable custom needs in radiotherapy dosimetry and quality assurance (QA). Materials methods We designed fabricated parallel plate chambers using an off‐the‐shelf equipment. Conductive components the detectors were made conductive polylactic acid (cPLA) insulating acrylonitrile butadiene styrene (ABS). characterized detector responses a Varian TrueBeam linac at 95 cm SSD solid...
Linear Multi-scale Modeling (LMM) is an advanced diffusion-weighted imaging(DWI) technique that uses multi-shell, multi-diffusion-time DWI data to estimate tissue microstructure parameters, including volume fractions of restricted and hindered water compartments over a range length scales orientation distribution information. Here,we apply the LMM framework characterize prostate cancer(PCa) lesions correlate our results with histology. Within histopathologically proven cancerous we observed...
Filter-decomposition-based group equivariant convolutional neural networks show promising stability and data efficiency for 3D image feature extraction. However, the existing filter-decomposition-based rely on parameter-sharing designs are mostly limited to rotation transform groups, where chosen spherical harmonic filter bases consider only angular orthogonality. These limitations hamper its application deep network architectures medical segmentation. To address these issues, this paper...
The purpose of the present development is to employ 3D printing prototype an ion chamber array with a scalable design potentially allowing increased spatial resolution and larger active area. An additional goal fabricate custom size thin-panel detector low-Z components. As proof principle demonstration, medium 30 × air-vented chambers was 3D-printed using PLA as frame for electrodes. active-area 122 mm 120 4 mm2 resolution. External electrodes are cylindrical made from conductive PLA....
Group-equivariant convolutional neural networks (G-CNN) heavily rely on parameter sharing to increase CNN's data efficiency and performance. However, the parameter-sharing strategy greatly increases computational burden for each added parameter, which hampers its application deep network models. In this paper, we address these problems by proposing a non-parameter-sharing approach group equivariant networks. The proposed methods adaptively aggregate diverse range of filters weighted sum...
We evaluated the influence of normalization (setting mean and standard deviation, histogram matching percentiles) on segmentation rectal cancer multimodal images when operating multicenter data as part a Radiomics pipeline. used two different networks for segmentation. When training evaluating all or from single center, did not play significant role. In contrast, one center others, it major Best results are obtained by using percentiles. Fixing deviation work well.
Radiomics enables the extraction of quantitative features from medical images, potentially augmenting characterization healthy and diseased tissue. Before these can be routinely used as biomarkers in clinical practice, however, their repeatability reproducibility must ensured. This study seeks to investigate feature an in-vivo, test-retest dataset prostate cancer patients. Our results show that majority (71.8%) radiomic extracted T2-weighted images was not repeatable, emphasizing need for studies.