Index

Deep Learning for low-dose Computed Tomography CAD systems

Thesis Description

Every year, more than two million people are diagnosed with lung cancer, demonstrating a survival rate of only 20% after 5 years after being diagnosed [1]. This leads to 1.8 million people dying from lung cancer every year, making it the leading cause of cancer-related deaths worldwide [2].

Lung cancer screening (LCS) programs have demonstrated success in decreasing the mortality when diagnosing the cancer in an early more treatable stage. Wide-scale LCS programs, such as the NLST (national lung screening trial), showed the reduction in 20% mortality in high-risk populations [3]. In direct comparison to X-ray radiography, the use of low-dose computed tomography (LDCT) has proven to be more effective for LCS, which is why upcoming nation-wide screening initiatives suggest the use of LDCT [4]. To support manual findings, experts suggest the additional use of computer-aided detection (CADe) as a secondary read to find, measure, and classify pulmonary nodules [4].

Even among experienced radiologists, there is still a moderate to high amount of inter-reader variability, depending on the size and density of individual pulmonary nodules, which affect patient management [5], but can also carry over to the both the truthing and performance of supervised learning methods. The first part of this thesis includes the assessment of the influence of the dose level (and thus the image noise) of different thoracic CT acquisitions onto the performance of different CAD systems (detection and segmentation) w.r.t the characteristics of individual pulmonary findings. Images with different dose levels will be provided. In the next step, based on the simulated dose levels, a neural network to denoise the thoracic LDCT images should be implemented. During the last years, various approaches have been introduced to apply denoising in the spatial domain, a transformed domain, and by utilizing convolutional neural networks (CNNs) and generative adversarial networks (GANs) [6] [7]. Within this work, the conventional dose level of the acquisition (ideally a full-dose scan) serves as the ground truth for the supervised learning.

Sources

[1]	K. C. Thandra, A. Barsouk, K. Saginala, J. S. Aluru and A. Barsouk, “Epidemiology of lung cancer,” Contemporary Oncology, vol. 25, no. 1, pp. 45–52, 2021.
[2]	H. Sung, J. Ferlay, R. L. Siegel, M. Laversanne, I. Soerjomataram, A. Jemal and F. Bray, “Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries,” CA: a cancer journal for clinicians, vol. 71, no. 3, pp. 209–249, 2021.
[3]	D. R. Aberle, A. M. Adams, C. D. Berg, W. C. Black, J. D. Clapp, R. M. Fagerstrom, I. F. Gareen, C. Gatsonis, P. M. Marcus and J. D. Sicks, “Reduced lung-cancer mortality with low-dose computed tomographic screening.,” The New England journal of medicine, vol. 365, no. 5, pp. 395–409, 2011.
[4]	T. G. Blum, J. Vogel-Claussen, S. Andreas, T. T. Bauer, J. Barkhausen, V. Harth, H.-U. Kauczor, W. Pankow, K. Welcker, R. Kaaks and H. Hoffmann, “Positionspapier zur Implementierung eines nationalen organisierten Programms in Deutschland zur Früherkennung von Lungenkrebs in Risikopopulationen mittels Low-dose-CT-Screening inklusive Management von abklärungsbedürftigen Screeningbefunden,” Pneumologie, vol. 78, no. 01, pp. 15–34, 2024.
[5]	S. J. Van Riel, C. I. Sánchez, A. A. Bankier, D. P. Naidich, J. Verschakelen, E. T. Scholten, P. A. de Jong, C. Jacobs, E. van Rikxoort, L. Peters-Bax, M. Snoeren, M. Prokop, B. van Ginneken and e. al., “Observer variability for classification of pulmonary nodules on low-dose CT images and its effect on nodule management,” Radiology, vol. 277, no. 3, pp. 863–871, 2015.
[6]	E. Eulig, B. Ommer and M. Kachelriess, “Benchmarking deep learning-based low-dose CT image denoising algorithms.,” Medical Physics, 2024.
[7]	R. T. SADIA, J. CHEN and J. ZHANG, “CT image denoising methods for image quality improvement and radiation dose reduction.,” Journal of Applied Clinical Medical Physics, vol. 25, no. 2, p. e14270, 2024.

Improving Deep Learning Target Volume Autosegmentation in Head and Neck Cancer for MR-guided Radiotherapy Treatment Planning

Automated Body Composition Analysis in Pancreatic Cancer CT Datasets using Deep Learning Auto-Segmentation

Development of a 3D CNN for Classifying Distortion Correction Status in Brain MRI Images for Radiotherapy Treatment Planning

Identifying Killer Whale Vocalizations using Graph Neural Networks

Spectral Plaque Analysis from Photon Counting CT

Generation of Artificial Vessel Trees for X-ray Image Analysis

Description

For the training of modern deep-learning-based image analysis algorithms, large scale datasets are required. For example, the segment anything model was trained 11 million images [1]. In the field of interventional X-ray imaging, this scale of data is unattainable, since not that many interventional procedures are performed, the images are not routinely stored, and are acquired as little as possible due to dose concerns. Furthermore, for some image analysis problems manual annotation of ground truth is infeasible even for optical computer vision. In such cases, synthetic data is used frequently, like the flying chairs dataset for optical flow [2].

For vessels synthesis, simple procedural approaches exist like the Sapling Tree Gen sampling algorithm [3] and space-filling [8]. For liver vessel analysis applications, an algorithm for generation of hepatic arterial trees is available [4]. In [5], hepatic vascular structure is generated using a physiological model of vessel growth. Generative learning-based methods have also been developed [6,9].

In this work, a pipeline for training and evaluating 2d X-ray image analysis methods is developed. The starting point is a procedural 3d vessel tree synthesis method [4,8]. As an exemplary organ, liver is used. The next step in the pipeline is forward projection to a 2d image according to the geometry of a real C-arm. As potential labels for the 2d images, depth, vessel segment IDs, vessel centerline IDs and overlap information are created. Finally, to demonstrate the applicability of the vessel tree for image analysis tasks, a vessel registration approach is evaluated on the resulting 2d vessel maps [2,7,10].

Work Packages

Literature research on simulation of vessel tree structures
Implementation in prototype environment (Python, Matlab, C++, …)
Quantitative and qualitative evaluation of registration using synthetic ground truth
Discussions of results & comparison to state of the art

Public software libraries should be used where possible. The complete source code including documentation must be provided.

Literature

[1] A. Kirillov et al., “Segment Anything,” in IEEE International Conference on Computer Vision (ICCV), 2023

[2] A. Dosovitskiy et al., “FlowNet: Learning Optical Flow with Convolutional Networks,” in IEEE International Conference on Computer Vision (ICCV), 2015

[3] N. Maul et al., “Transient Hemodynamics Prediction Using an Efficient Octree-Based Deep Learning Model,” in International Conference on Information Processing in Medical Imaging, 2023

[4] J. F. Whitehead, P. F. Laeseke, S. Periyasamy, M. A. Speidel, and M. G. Wagner, “In silico simulation of hepatic arteries: An open-source algorithm for efficient synthetic data generation,” Medical Physics, 2023

[5] M. Kretowski, Y. Rolland, J. Bézy-Wendling, and J.-L. Coatrieux, “Physiologically based modeling of 3-D vascular networks and CT scan angiography,” IEEE Transactions on Medical Imaging, 2003

[6] T. P. Kuipers, P. R. Konduri, H. Marquering, and E. J. Bekkers, “Generating Cerebral Vessel Trees of Acute Ischemic Stroke Patients using Conditional Set-Diffusion,” in Medical Imaging with Deep Learning, 2024

[7] G. Balakrishnan, A. Zhao, M. R. Sabuncu, J. Guttag, und A. V. Dalca, „Voxelmorph: a learning framework for deformable medical image registration“, IEEE Transactions on Medical Imaging, 2019

[8] N. Rauch and M. Harders, „Interactive Synthesis of 3D Geometries of Blood Vessels.“, in Eurographics (Short Papers), 2021

[9] C. Prabhakar et al., „3D Vessel Graph Generation Using Denoising Diffusion“, in International Conference on Medical Image Computing and Computer-Assisted Intervention, 2024

[10] S. Klein, M. Staring, K. Murphy, M. A. Viergever, and J. P. Pluim, “Elastix: a toolbox for intensity-based medical image registration,” IEEE Transactions on Medical Imaging, 2009

Alzheimer’s Classification from MRI

Frequency Domain Hierarchical Vision Transformer-based Perceptual Loss

This project focuses on improving image processing tasks, such as super-resolution or image restoration, by employing a novel feature comparison method. It leverages a Hierarchical Vision Transformer to extract multi-scale feature representations from images. These features capture both local and global information at various levels of abstraction. Crucially, these extracted features are then transformed into the frequency domain, likely via a Fast Fourier Transform (FFT) or similar method. The comparison between the generated image and the target image occurs in this frequency space. By analyzing differences in magnitude and/or phase across different frequency bands, the model can better understand and rectify discrepancies in texture, detail, and overall structure. This approach aims to produce perceptually superior results by guiding the model to reconstruct images that are more aligned with the frequency characteristics of the target, leading to improved visual quality, especially in terms of sharpness and fine-grained details.