A5048208182- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- Medical Image Segmentation Techniques
- Artificial Intelligence in Healthcare
- Artificial Intelligence in Healthcare and Education
- Advanced X-ray and CT Imaging
- Generative Adversarial Networks and Image Synthesis
- Digital Imaging for Blood Diseases
- Cancer Immunotherapy and Biomarkers
- Brain Tumor Detection and Classification
- Image Processing Techniques and Applications
- Aortic aneurysm repair treatments
- Image Retrieval and Classification Techniques
- Liver Disease Diagnosis and Treatment
- Cell Image Analysis Techniques
- Prostate Cancer Diagnosis and Treatment
- Sparse and Compressive Sensing Techniques
- Colorectal Cancer Treatments and Studies
- Immune cells in cancer
- Hepatocellular Carcinoma Treatment and Prognosis
- Cancer Cells and Metastasis
- Renal Diseases and Glomerulopathies
- Multiple Sclerosis Research Studies
- Renal and Vascular Pathologies
- Immunotherapy and Immune Responses
Fraunhofer Institute for Digital Medicine
2015-2023
University of Lübeck
2013
Prostate cancer (PCa) is graded by pathologists examining the architectural pattern of cancerous epithelial tissue on hematoxylin and eosin (H&E) stained slides. Given importance gland morphology, automatically differentiating between glandular other tissues an important prerequisite for development automated methods detecting PCa. We propose a new method, using deep learning, segmenting in digitized prostatectomy employed immunohistochemistry (IHC) to render ground truth less subjective...
Automatic Non-rigid Histological Image Registration (ANHIR) challenge was organized to compare the performance of image registration algorithms on several kinds microscopy histology images in a fair and independent manner. We have assembled 8 datasets, containing 355 with 18 different stains, resulting 481 pairs be registered. accuracy evaluated using manually placed landmarks. In total, 256 teams registered for challenge, 10 submitted results, 6 participated workshop. Here, we present...
Deep convolutional neural networks have become a widespread tool for the detection of nuclei in histopathology images. Many implementations share basic approach that includes generation an intermediate map indicating presence nucleus center, which we refer to as PMap. Nevertheless, these often still differ several parameters, resulting different qualities. We identified essential parameters and configured PMap using combinations them. thoroughly evaluated compared various configurations on...
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....
We propose a virtual staining methodology based on Generative Adversarial Networks to map histopathology images of breast cancer tissue from H&E stain PHH3 and vice versa. use the resulting synthetic build Convolutional Neural (CNN) for automatic detection mitotic figures, strong prognostic biomarker used in routine diagnosis grading. several scenarios, which CNN trained with synthetically generated perform par or even better than same baseline model real images. discuss potential this...
The tumour microenvironment has been shown to be a valuable source of prognostic information for different cancer types. This holds in particular triple negative breast (TNBC), subtype which currently no biomarkers are established. Although methods assess infiltrating lymphocytes (TILs) have published, it remains unclear method (marker, region) yields the most optimal information. In addition, date, objective TILs assessment available. For this proof concept study, subset our previously...
SignificanceAlthough the registration of restained sections allows nucleus-level alignment that enables a direct analysis interacting biomarkers, consecutive only allow transfer region-level annotations. The latter can be achieved at low computational cost using coarser image resolutions.PurposeIn digital histopathology, virtual multistaining is important for diagnosis and biomarker research. Additionally, it provides accurate ground truth various deep-learning tasks. Virtual obtained...
Background: Features characterizing the immune contexture (IC) in tumor microenvironment can be prognostic and predictive biomarkers. Identifying novel biomarkers challenging due to complex interactions between cells abundance of possible features. Methods: We describe an approach for data-driven identification IC For this purpose, we provide mathematical definitions different feature classes, based on cell densities, cell-to-cell distances, spatial heterogeneity thereof. Candidate are...
Automated image analysis and artificial intelligence (AI) are becoming increasingly common in digital pathology software. While various proprietary systems exist, there no fully vendor-agnostic integration approaches for AI apps. This makes it difficult vendors of solutions to integrate their products into the multitude non-standard software pathology. The EMPAIA Consortium is developing an open decentralized platform allowing AI-based apps different be integrated with existing clinical IT...
Steatosis is routinely assessed histologically in clinical practice and research. Automated image analysis can reduce the effort of quantifying steatosis. Since reproducibility essential for practical use, we have evaluated different methods terms their agreement with stereological point counting (SPC) performed by a hepatologist. The evaluation was based on large representative data set 970 histological images from human patients liver diseases. Three were built previously published...
Artificial intelligence (AI) apps hold great potential to make pathological diagnoses more accurate and time efficient. Widespread use of AI in pathology is hampered by interface incompatibilities between software. We studied the existing interfaces order develop EMPAIA App Interface, an open standard for integration apps.The Interface relies on widely-used web communication protocols containerization. It consists three parts: A standardized format describe semantics app, a mechanism deploy...
We present a 3-step registration pipeline for differently stained histological serial sections that consists of 1) robust pre-alignment, 2) parametric computed on coarse resolution images, and 3) an accurate nonlinear registration. In all three steps the NGF distance measure is minimized with respect to increasingly flexible transformation. apply method in ANHIR image challenge evaluate its performance training data. The presented (error reduction 99.6% cases), fast (runtime 4 seconds)...
Automated image analysis and artificial intelligence (AI) are a growing market in digital pathology. While various proprietary pathology systems exist, there no fully vendor-agnostic integration approaches for AI apps. This makes it difficult vendors of solutions to integrate their products into the multitude non-standard software The EMPAIA Consortium (EcosysteM Pathology Diagnostics with Assistance) develops an open decentralized platform allowing AI-based apps different be integrated...
ABSTRACT In the glomerulus, Bowman's space is formed by a continuum of glomerular epithelial cells. focal segmental glomerulosclerosis (FSGS), glomeruli show scarring, result activated parietal cells (PECs) invading tuft. The scars interrupt continuum. However, non-sclerotic segments seem to be preserved even in with advanced lesions. We studied histology pattern Munich Wistar Frömter rats, model for secondary FSGS. Our results showed that matrix layers lined PECs cover sclerotic These...
Many physiological processes and pathological phenomena in the liver tissue are spatially heterogeneous. At a local scale, biomarkers can be quantified along axis of blood flow, from portal fields (PFs) to central veins (CVs), i.e., zonated form. This requires detecting PFs CVs. However, manually annotating these structures multiple whole-slide images is tedious task. We describe evaluate fully automated method, based on convolutional neural network, for simultaneously CVs single stained...
Purpose: In digital histopathology, virtual multi-staining is important for diagnosis and biomarker research. Additionally, it provides accurate ground-truth various deep-learning tasks. Virtual can be obtained using different stains consecutive sections or by re-staining the same section. Both approaches require image registration to compensate tissue deformations, but little attention has been devoted comparing their accuracy. Approach: We compare variational of re-stained analyze effect...