Now showing1 - 4 of 4
  • Some of the metrics are blocked by your 
    Item type:Publication, Access status: with files ,
    Parameter identification for a two-compartment contrast flow field model
    This thesis applies a two-compartment model to reconstruct perfusion parameters from 3D dynamic contrast-enhanced ultrasound data, with the goal of providing information about tumor perfusion and treatment response. The method is demonstrated on 3D ultrasound data, which requires projecting the data onto a 2D-plane, resulting in reconstructed parameters that include physiologically plausible flow velocities and a conversion functions, which relates to the perfusion of the area.
    Publicationtype: Doctoral Thesis
      59  33
  • Some of the metrics are blocked by your 
    Item type:Publication, Access status: with files ,
    Soft tissue characterization from miniaturized imaging probes
    Characterizing soft tissue properties is essential for diagnosing and treating diseases. Optical coherence tomography offers high spatial and temporal resolution for analyzing small lesions. However, when applied in the body, image acquisition and interpretation are complicated by relative movements and deformations. This thesis investigates three miniaturized probe designs and image analysis methods, as well as the influence of motion on the assessment of tissue properties, underlying the potential of these probes for improving in-vivo tissue characterization.
    Publicationtype: Doctoral Thesis
      104  71
  • Some of the metrics are blocked by your 
    Item type:Publication, Access status: with files ,
    Sensorfusion für das robuste und kontaktlose Monitoring der Vitalparameter von Frühgeborenen auf der neonatologischen Intensivstation
    The current monitoring of vital signs of preterm neonates bears the risk to injure the little patients. Therefor, the question arises if it is possible to determine the vital parameters of neonates and preterm neonates precisely and robustly via non-contact methods. A non-contact system was developed for monitoring the respiratory and heart rates as well as the core temperature of the neonates. Different sensors such as an RGB camera, a thermal camera, several time-of-flight cameras and a radar sensor are combined or their data are fused. The region of interest detected by each sensor is automatically determined by a neural network (YOLOv3) which was trained with fused images. The necessary algorithms to calculate the vital parameters were developed or existing algorithms improved. The best approach found during the validation phase is the evaluation of ToF camera data for determining the respiratory rate and the temperature detection via thermal camera with reference temperature sensor. The evaluation of the heart rate using an RGB camera does not deliver the clinical necessary precision. It was shown that under optimal conditions a non-contact detection of the respiratory rate and the body temperature is possible with good results.
    Publicationtype: Doctoral Thesis
      54  40
  • Some of the metrics are blocked by your 
    Item type:Publication, Access status: with files ,
    Deep learning for paranasal anomaly classification
    The susceptibility of the paranasal sinuses to allergic and non-allergic infections complicates diagnosis as symptoms vary from mucosal thickening to polypoid masses. Previous manual analysis is resource intensive and leads to clinician fatigue. Computer-aided diagnosis (CAD) can optimize this process. While deep learning has made great strides in medical image analysis, the study of paranasal abnormalities has remained limited. This work proposes different deep learning approaches to classify maxillary sinus opacities, including unsupervised learning, self-supervised learning, and hybrid architectures of CNNs and transformers.
    Publicationtype: Doctoral Thesis
      46  85