A5034261670- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- Neural Networks and Applications
- Neural Networks and Reservoir Computing
- Time Series Analysis and Forecasting
- Cancer Cells and Metastasis
- Fault Detection and Control Systems
- Advanced Memory and Neural Computing
- Lung Cancer Diagnosis and Treatment
- Digital Imaging for Blood Diseases
- Machine Learning in Materials Science
- Cell Image Analysis Techniques
- Anomaly Detection Techniques and Applications
- Gaussian Processes and Bayesian Inference
- Software Reliability and Analysis Research
- Metallurgy and Material Forming
- Cancer Immunotherapy and Biomarkers
- Advanced machining processes and optimization
- Software System Performance and Reliability
- Nonmelanoma Skin Cancer Studies
- Neural dynamics and brain function
- Colorectal Cancer Screening and Detection
- EEG and Brain-Computer Interfaces
- Stock Market Forecasting Methods
- Control Systems in Engineering
Radboud University Nijmegen
2021-2025
Radboud University Medical Center
2022-2025
University Medical Center
2022-2024
Bielefeld University
2015-2018
Osnabrück University
2015
The digitalization of clinical workflows and the increasing performance deep learning algorithms are paving way towards new methods for tackling cancer diagnosis. However, availability medical specialists to annotate digitized images free-text diagnostic reports does not scale with need large datasets required train robust computer-aided diagnosis that can target high variability cases data produced. This work proposes evaluates an approach eliminate manual annotations tools in digital...
Abstract Background Invasive breast cancer patients are increasingly being treated with neoadjuvant chemotherapy; however, only a fraction of the respond to it completely. To prevent overtreatment, there is an urgent need for biomarkers predict treatment response before administering therapy. Methods In this retrospective study, we developed hypothesis-driven interpretable based on deep learning, pathological complete (pCR, i.e., absence tumor cells in surgical resection specimens)...
Abstract The prognostic significance of tumor-infiltrating lymphocytes (TILs) in breast cancer has been recognized for over a decade. Although histology-based scoring recommendations exist to standardize visual TILs assessment, interobserver agreement and reproducibility are hampered by heterogeneous infiltration patterns, highlighting the importance computational approaches. Despite advances automate quantification, adoption models hindered lack consensus on methods large-scale benchmarks....
Artificial Intelligence can mitigate the global shortage of medical diagnostic personnel but requires large-scale annotated datasets to train clinical algorithms. Natural Language Processing (NLP), including Large Models (LLMs), shows great potential for annotating data facilitate algorithm development remains underexplored due a lack public benchmarks. This study introduces DRAGON challenge, benchmark NLP with 28 tasks and 28,824 reports from five Dutch care centers. It facilitates...
The frequency of basal cell carcinoma (BCC) cases is putting an increasing strain on dermatopathologists. BCC the most common type skin cancer, and its incidence rapidly worldwide. AI can play a significant role in reducing time effort required for diagnostics thus improve overall efficiency process. To train such system fully-supervised fashion however, would require large amount pixel-level annotation by already strained Therefore, this study, our primary objective was to develop...
Current hardware limitations make it impossible to train convolutional neural networks on gigapixel image inputs directly. Recent developments in weakly supervised learning, such as attention-gated multiple instance have shown promising results, but often use multi-stage or patch-wise training strategies risking suboptimal feature extraction, which can negatively impact performance. In this paper, we propose a ResNet-34 encoder with an classification head end-to-end fashion, call...
Neural plasticity plays an important role in learning and memory. Reward-modulation of offers explanation for the ability brain to adapt its neural activity achieve a rewarded goal. Here, we define network model that learns through interaction Intrinsic Plasticity (IP) reward-modulated Spike-Timing-Dependent (STDP). IP enables explore possible output sequences STDP, modulated by reward, reinforces creation sequences. The is tested on tasks prediction, recall, non-linear computation, pattern...
Classification of non-small-cell lung cancer (NSCLC) into adenocarcinoma (LUAD) and squamous cell carcinoma (LUSC) via histopathology is a vital prerequisite to select the appropriate treatment for patients. Most machine learning approaches rely on manually annotating large numbers whole slide images (WSI) training. However, delineating areas or even single cells hundreds thousands slides tedious, subjective requires highly trained pathologists. We propose use Neural Image Compression (NIC),...
Training neural networks with high-quality pixel-level annotation in histopathology whole-slide images (WSI) is an expensive process due to gigapixel resolution of WSIs. However, recent advances self-supervised learning have shown that highly descriptive image representations can be learned without the need for annotations. We investigate application Hierarchical Image Pyramid Transformer (HIPT) model specific task classification colorectal biopsies and polyps. After evaluating effectiveness...
Abstract Background Histopathological growth patterns are one of the strongest prognostic factors in patients with resected colorectal liver metastases. Development an efficient, objective and ideally automated histopathological pattern scoring method can substantially help implementation assessment daily practice research. This study aimed to develop validate a deep-learning algorithm, namely neural image compression, distinguish desmoplastic from non-desmoplastic metastases based on...
Screening programs for early detection of cancer such as colorectal and cervical have led to an increased demand histopathological analysis biopsies. Advanced image with Deep Learning has shown the potential automate in digital pathology whole-slide images. Particularly, techniques weakly supervised learning can achieve classification without need tedious, manual annotations, using only slide-level labels. Here, we used data from n=12,580 images n=9,141 tissue blocks train validate a deep...
Abstract Purpose Invasive breast cancer patients are increasingly being treated with neoadjuvant chemotherapy, however, only a fraction of the respond to it completely. To prevent over-treating toxic drug, there is an urgent need for biomarkers capable predicting treatment response before administering therapy. In this retrospective study, we developed interpretable, deep-learning based predict pathological complete (pCR, i.e. absence tumor cells in surgical resection specimens) chemotherapy...