A5019457098- Radiomics and Machine Learning in Medical Imaging
- Advanced X-ray and CT Imaging
- Medical Imaging Techniques and Applications
- Lung Cancer Diagnosis and Treatment
- COVID-19 diagnosis using AI
- Radiation Dose and Imaging
- Brain Tumor Detection and Classification
- Atomic and Subatomic Physics Research
- Phonocardiography and Auscultation Techniques
- Acute Ischemic Stroke Management
- Medical Imaging and Analysis
- Medical Image Segmentation Techniques
- Infrared Thermography in Medicine
- Chronic Obstructive Pulmonary Disease (COPD) Research
- Advanced X-ray Imaging Techniques
- Seismic Imaging and Inversion Techniques
Technical University of Munich
2020-2024
Klinikum rechts der Isar
2020-2024
Abstract Lung cancer is a major cause of death worldwide. As early detection can improve outcome, regular screening great interest, especially for certain risk groups. Besides low-dose computed tomography, chest X-ray potential option screening. Convolutional network (CNN) based computer aided diagnosis systems have proven their ability identifying nodules in radiographies and thus may assist radiologists clinical practice. Based on segmented pulmonary nodules, we trained CNN one-stage...
Abstract Background Currently, alternative medical imaging methods for the assessment of pulmonary involvement in patients infected with COVID-19 are sought that combine a higher sensitivity than conventional (attenuation-based) chest radiography lower radiation dose CT imaging. Methods Sixty COVID-19-associated lung changes scan and 40 subjects without pathologic visible were included (in total, 100, 59 male, mean age 58 ± 14 years). All gave written informed consent. We employed clinical...
Abstract We present a method to generate synthetic thorax radiographs with realistic nodules from CT scans, and perfect ground truth knowledge. evaluated the detection performance of nine radiologists two convolutional neural networks in reader study. Nodules were artificially inserted into lung volume obtained by forward-projecting volume. Hence, our framework allowed for detailed evaluation CAD systems’ radiologists’ due availability accurate ground-truth labels data. Radiographs network...
Abstract Purpose : To evaluate the benefit of additional available information present in spectral CT datasets, as compared to conventional when utilizing convolutional neural networks for fully automatic localisation and classification liver lesions images. Materials Methods Conventional images (iodine maps, virtual monochromatic (VMI)) were obtained from a dual-layer system. Patient diagnosis known clinical reports classified into healthy, cyst hypodense metastasis. In order compare value...
Purpose To explore the potential benefits of deep learning-based artifact reduction in sparse-view cranial CT scans and its impact on automated hemorrhage detection. Materials Methods In this retrospective study, a U-Net was trained for simulated 3000 patients, obtained from public dataset reconstructed with varying levels. Additionally, EfficientNet-B2 full-view data 17 545 patients Detection performance evaluated using area under receiver operating characteristic curve (AUC), differences...
Zielsetzung Chronic Obstructive Pulmonary Disease (COPD) is a leading cause of death worldwide, yet early detection and treatment can prevent the progression disease. X-ray Computed Tomography (CT) scans chest provide quantitative measure morphological changes in lung. The potential incorporating optimal Window Setting (WS) selection, typically carried out during examination CT for COPD, generally overlooked deep learning models. We aim to optimize COPD with densely connected convolutional...
Automatic evaluation of 3D volumes is a topic importance in order to speed up clinical decision making. We describe method classify computed tomography scans on volume level for the presence non-acute cerebral infarction. This not trivial task, as lesions are often similar other areas brain regarding shape and intensity. A three stage architecture used classification: 1) cranial cavity segmentation network developed, trained applied. 2) Region proposals generated 3) Connected regions...
Assessing tumor response to systemic therapies is one of the main applications PET/CT. Routinely, only a small subset index lesions out multiple analyzed. However, this operator dependent selection may bias results due possible significant inter-metastatic heterogeneity therapy. Automated, AI based approaches for lesion tracking hold promise in enabling analysis many more and thus providing better assessment response. This work introduces Siamese CNN approach between PET/CT scans. Our...
Dual Energy CT is a modern imaging technique that utilized in clinical practice to acquire spectral information for various diagnostic purposes including the identification, classification, and characterization of different liver lesions. It provides additional that, when compared available from conventional datasets, has potential benefit existing computer vision techniques by improving their accuracy reliability. In order evaluate value versus datasets being used as input machine learning...
We aim to optimize the binary detection of Chronic Obstructive Pulmonary Disease (COPD) based on emphysema presence in lung with convolutional neural networks (CNN) by exploring manually adjusted versus automated window-setting optimization (WSO) computed tomography (CT) images. 7,194 CT images (3,597 COPD; 3,597 healthy controls) from 78 subjects (43 35 were selected retrospectively (10.2018-12.2019) and preprocessed. For each image, intensity values clipped window setting a baseline...
This is a preprint. The latest version has been published here: https://pubs.rsna.org/doi/10.1148/ryai.230275 Purpose: Sparse-view computed tomography (CT) an effective way to reduce dose by lowering the total number of views acquired, albeit at expense image quality, which, in turn, can impact ability detect diseases. We explore deep learning-based artifact reduction sparse-view cranial CT scans and its on automated hemorrhage detection. Methods: trained U-Net for artefact simulated from...
Zielsetzung Untersuchung des Potenzials von Machine Learning Algorithmen zur Erkennung Lungenrundherden in Röntgenaufnahmen Thorax.
Zielsetzung Mithilfe von Convolutional Neural Networks (CNN) soll die Erkennung des hyperdensen Mediazeichens in der klinischen Diagnostik verbessert werden. Das hyperdense Mediazeichen ist ein Frühzeichen ischämischen Schlaganfalles und somit hoher klinischer Bedeutung. Da dieses Zeichen Radiologen leicht übersehen werden kann, wird hier vorgestellten Arbeit CNN-basiertes computerassistiertes Diagnosesystem (CAD) entwickelt.
Zielsetzung Es sollen die Positionen in der Lunge ermittelt werden, an denen Wahrscheinlichkeit für nicht detektierte Lungenrundherde erhöht ist.
Abstract Medical imaging is performed in daily clinical routine for the assessment of pulmonary involvement patients infected with COVID-19. As conventional (attenuation-based) chest radiography provides only a low sensitivity COVID-19-pneumonia, CT gold standard lung COVID-19 patients. However, exposes patient to considerable amount radiation, and not as widely available plain X-rays. Therefore, alternative low-dose X-ray techniques are highly desirable. Here we present first results...
Deep learning based solutions are being succesfully implemented for a wide variety of applications. Most notably, clinical use-cases have gained an increased interest and been the main driver behind some cutting-edge data-driven algorithms proposed in last years. For applications like sparse-view tomographic reconstructions, where amount measurement data is small order to keep acquisition time short radiation dose low, reduction streaking artifacts has prompted development denoising with...
Estimating the lung depth on x-ray images could provide both an accurate opportunistic volume estimation during clinical routine and improve image contrast in modern structural chest imaging techniques like dark-field imaging. We present a method based convolutional neural network that allows per-pixel thickness subsequent total capacity estimation. The was trained validated using 5250 simulated radiographs generated from 525 real CT scans. evaluated test set of 131 synthetic retrospective...