A5075582704- Medical Imaging Techniques and Applications
- Advanced Radiotherapy Techniques
- Advanced X-ray and CT Imaging
- Radiomics and Machine Learning in Medical Imaging
- Acute Ischemic Stroke Management
- AI in cancer detection
- Cerebrovascular and Carotid Artery Diseases
- Colorectal Cancer Screening and Detection
- Cutaneous Melanoma Detection and Management
- Stroke Rehabilitation and Recovery
- Radiation Dose and Imaging
- Digital Radiography and Breast Imaging
- Advanced Neural Network Applications
- Brain Tumor Detection and Classification
- Digital Media Forensic Detection
- Neurological Disease Mechanisms and Treatments
- Digital Imaging for Blood Diseases
- Enhanced Recovery After Surgery
- Transplantation: Methods and Outcomes
- Lung Cancer Diagnosis and Treatment
- Generative Adversarial Networks and Image Synthesis
- Medical Imaging and Analysis
- Surgical Simulation and Training
- Radiation Therapy and Dosimetry
- Nonmelanoma Skin Cancer Studies
Universität Hamburg
2018-2025
University Medical Center Hamburg-Eppendorf
2018-2025
Infektionsmedizinisches Centrum Hamburg
2019
Objective: This work addresses two key problems of skin lesion classification. The first problem is the effective use high-resolution images with pretrained standard architectures for image second high class imbalance encountered in real-world multi-class datasets. Methods: To images, we propose a novel patch-based attention architecture that provides global context between small, patches. We modify three and study performance attention. counter problems, compare oversampling, balanced batch...
In this paper we present the methods of our submission to ISIC 2018 challenge for skin lesion diagnosis (Task 3). The dataset consists 10000 images with seven image-level classes be distinguished by an automated algorithm. We employ ensemble convolutional neural networks task. particular, fine-tune pretrained state-of-the-art deep learning models such as Densenet, SENet and ResNeXt. identify heavy class imbalance a key problem consider multiple balancing approaches loss weighting balanced...
Four-dimensional (4D) CT imaging is a central part of current treatment planning workflows in 4D radiotherapy (RT). However, clinical image data often suffer from severe artifacts caused by insufficient projection coverage due to the inability commercial protocols adapt breathing irregularity. We propose an intelligent sequence mode protocol (i4DCT) that builds on online curve analysis and respiratory signal-guided selection beam on/off periods during scan time order fulfill requirements....
Purpose Four‐dimensional cone‐beam computed tomography (4D CBCT) imaging has been suggested as a solution to account for interfraction motion variability of moving targets like lung and liver during radiotherapy (RT) targets. However, due severe sparse view sampling artifacts, current 4D CBCT data lack sufficient image quality accurate quantification. In the present paper, we introduce deep learning‐based framework boosting that can be combined with any reconstruction approach clinical...
Is self-supervised deep learning (DL) for medical image analysis already a serious alternative to the de facto standard of end-to-end trained supervised DL? We tackle this question classification, with particular focus on one currently most limiting factor field: (non-)availability
Abstract Background 4D CT imaging is an essential component of radiotherapy thoracic and abdominal tumors. images are, however, often affected by artifacts that compromise treatment planning quality image information reliability. Purpose In this work, deep learning (DL)‐based conditional inpainting proposed to restore anatomically correct artifact‐affected areas. Methods The restoration approach consists a two‐stage process: DL‐based detection common interpolation (INT) double structure (DS)...
4D CT imaging is an integral part of radiotherapy workflows. However, data often contain motion artifacts that mitigate treatment planning. Recently, breathing-adapted (i4DCT) was introduced into clinical practice, promising artifact reduction in in-silico and phantom studies. Here, we present image quality comparison study, pooling patient from two centers: a new i4DCT conventional spiral cohort.
Purpose Four‐dimensional (4D) computed tomography (CT) imaging is an essential part of current 4D radiotherapy treatment planning workflows, but clinical CT images are often affected by artifacts. The artifacts mainly caused breathing irregularity during data acquisition, which leads to projection coverage issues for currently available commercial protocols. It was proposed improve online respiratory signal analysis and signal‐guided tube control, related work always theoretical presented as...
Abstract 4D CT imaging is a cornerstone of radiotherapy treatment. Clinical data are, however, often affected by severe artifacts. The artifacts are mainly caused breathing irregularity and retrospective correlation phase information acquired projection data, which leads to insufficient coverage allow for proper reconstruction images. recently introduced approach i4DCT (intelligent sequence scanning) aims overcome this problem signal-driven tube control. present motion phantom study...
Background and Purpose: Mechanical thrombectomy is an established procedure for treatment of acute ischemic stroke. success commonly assessed by the Thrombolysis in Cerebral Infarction (TICI) score, assigned visual inspection X-ray digital subtraction angiography data. However, expert-based TICI scoring highly observer-dependent. This represents a major obstacle mechanical outcome comparison in, instance, multicentric clinical studies. Focusing on occlusions M1 segment middle cerebral...
The ability to automatically detect and track surgical instruments in endoscopic videos can enable transformational interventions. Assessing performance efficiency, identifying skilled tool use choreography, planning operational logistical aspects of OR resources are just a few the applications that could benefit. Unfortunately, obtaining annotations needed train machine learning models identify localize tools is difficult task. Annotating bounding boxes frame-by-frame tedious...
Inter-fractional magnitude and trajectory changes are of great importance for radiotherapy (RT) moving targets. In order to verify the amount characteristics patient-specific respiratory motion prior each RT treatment session, a time-resolved cone-beam computed tomography (4D CBCT) is necessary. However, due sparse view artifacts, resulting image quality limited when applying current 4D CBCT reconstruction approaches. this study, new deep learning-based boosting approach presented that does...
Breathing signal-guided 4D CT sequence scanning such as the intelligent (i4DCT) approach reduces imaging artifacts compared to conventional CT. By design, i4DCT captures entire breathing cycles during beam-on periods, leading redundant projection data and increased radiation exposure patients exhibiting prolonged exhalation phases. A recently proposed breathing-guided dose modulation (DM) algorithm promises lower by temporarily reducing tube current, but impact on image reconstruction...
In radiotherapy, precise comparison of fan-beam computed tomography (CT) and cone-beam CT (CBCT) arises as a commonplace, yet intricate task. This paper proposes publicly available end-to-end pipeline featuring an intrinsic deep-learning-based speedup technique for generating virtual 3D 4D CBCT from images.
In lung radiotherapy, the primary objective is to optimize treatment outcomes by minimizing exposure healthy tissues while delivering prescribed dose target volume. The challenge lies in accounting for tissue motion due breathing, which impacts precise alignment. To address this, paper proposes a prospective approach that relies solely on pre-treatment information, such as planning CT scans and derived data like vector fields from deformable image registration. This compared analogous...
4D CT imaging is an essential component of radiotherapy thoracic/abdominal tumors. images are, however, often affected by artifacts that compromise treatment planning quality. In this work, deep learning (DL)-based conditional inpainting proposed to restore anatomically correct image information artifact-affected areas. The restoration approach consists a two-stage process: DL-based detection common interpolation (INT) and double structure (DS) artifacts, followed applied the artifact...
Is self-supervised deep learning (DL) for medical image analysis already a serious alternative to the de facto standard of end-to-end trained supervised DL? We tackle this question classification, with particular focus on one currently most limiting factors field: (non-)availability labeled data. Based three common imaging modalities (bone marrow microscopy, gastrointestinal endoscopy, dermoscopy) and publicly available data sets, we analyze performance DL within self-distillation no labels...
Quality assurance in current 4D radiotherapy workflows is of great importance to assure a positive treatment outcome, i.e. total tumor eradication. Especially for the lung and liver tumors, which are subject high motion magnitudes due patient breathing, it crucial verify applied dose target volume. In this study, we present new Monte Carlo accumulation approach that accounts internal during therefore able predict actual 3D distribution delivered quality purposes. simulations conducted using...