Student Opportunities

We are constantly looking for students with an interest in medical image analysis, as well as the use of machine learning and computer vision in novel and established clinical and forensic applications. Please contact Darko Stern if you are interested in joining our group. This page lists specific open student projects and also gives an overview of currently running and already finished projects on a master and bachelor level. Students coming with their own research ideas are also welcome to get in contact!


Open Student Projects

OPEN: Instance Segmentation in Medical Image Applications

Suitable as: Master's Project / Thesis

Detailed Description

OPEN: Detection of Infected Teeth in 3D CBCT Images

Suitable as: Master's Project / Thesis

Detailed Description

OPEN: Deep Reinforcement Learning in Medical Image Applications

Suitable as: Master's Project / Thesis

Detailed Description

OPEN: Rotation Invariant Deep Neural Networks

Suitable as: Master's Project / Thesis

Detailed Description

OPEN: Segmentation of the beating heart from dynamic MRI

Suitable as: Master's Thesis

Aim: We have recently developed a method for segmenting different structures in the heart (ventricles, atria, myocardium, ...) from 3D MRI and CT images [1], which was well received by the scientific community after winning the 2017 MMWHS challenge at a MICCAI workshop in Quebec City, and winning the best paper award of this workshop. We are now interested in extending this approach to dynamic MRI data, initially with 2D slices of the beating heart, but on the long run on the 3D beating heart. There is interest in this work by the Institute of Biophysics of Medical University Graz, who could contribute data.

The goal of this student project is to transfer our deep learning based approach from [1] to dynamic MRI. The student will not start from scratch, but from a very recent, unpublished work that we just submitted to a conference. This work will be performed in close collaboration with the Ludwig Boltzmann Institute for Clinical Forensic Imaging, Graz. The student needs good knowledge in Python programming and should be familiar with deep learning frameworks like e.g. TensorFlow. Further, the student will work in a small team of researchers of different experience levels, joined by their interest in forensic applications of image analysis and machine learning.

[1] C. Payer, D. Stern, H. Bischof, M. Urschler. Multi-Label Whole Heart Segmentation Using Anatomical Label Configurations. In: Proceedings of the STACOM 2017 MMWHS Workshop, Quebec City, Canada, 2017 Link to paper

 

OPEN: A novel approach for airway tree segmentation

Suitable as: Master's Thesis

Aim: According to a recent paper [1], implement a novel approach for airway tree segmentation from vascular computed tomography (CT) images.

The goal of this student project is to implement an algorithm for automatic airway tree extraction and segmentation from high-resolution CT images. Together with our collaboration partners at the Ludwig Boltzmann Institute for Lung Vascular Research, who are interested in lung diseases like pulmonary hypertension, we will use the airway tree to improve our existing artery/vein separation and lung lobe segmentation algorithms. The student needs good knowledge in C++ programming and will work in a small team of researchers interested in lung image analysis.

[1] St├╝hmer J, Cremers D. A Fast Projection Method for Connectivity Constraints in Image Segmentation. Proc EMMCVPR 2015.

 


Currently Ongoing Student Projects

TAKEN: Age estimation and growth prediction from MRI data of the knee

Suitable as: Master's Thesis

Student currently working on this project: Stefan Eggenreich

Aim: Together with our collaboration partners at the Medical University Graz, we are interested in extending our work on age estimation from different bony structures (hand, clavicle) to the knee [1]. The advantage of investigating the knee for age estimation is that epiphyseal gaps from three bones can be studied simultaneously. Studying the knee is also interesting from the point of view of predicting growth of children and adolescents, since tibia and femur bones have a large impact on one's final height as an adult.

Image taken from [1].

The goal of this student project is to implement a deep learning based approach for chronological age estimation from knee MR images provided by our medical collaboration partner. A subset of these images are from the same person, taken at different points in time. With this data we hope to be able to additionally study feasibility of growth prediction in a very preliminary stage. This work will be performed in close collaboration with the Ludwig Boltzmann Institute for Clinical Forensic Imaging, Graz. The student needs good knowledge in Python programming and should be familiar with deep learning frameworks like e.g. TensorFlow. Further, the student will work in a small team of researchers of different experience levels, joined by their interest in forensic applications of image analysis and machine learning.

This project is supported by funding from the City of Graz, the prospective student will be employed for six months with 8 hours a week.

[1] F. Dedouit,J. Auriol,H. Rousseau, D. Rouge, E. Crubezy, N. Telmon.Age assessment by magnetic resonance imaging of the knee: A preliminary study. Forensic Science International 217, 232.e1-232.e7. doi: 10.1016/j.forsciint.2011.11.013

 


Finished Student Projects

FINISHED (2018/03): Data Augmentation in Deep Learning using Generative Adversarial Networks

Student who worked on the topic: Thomas Neff

Link to Thomas' master's thesis.

Aim: Evaluate different strategies for synthetically augmenting image datasets for deep learning using recent generative models, especially generative adversarial networks.

The work in this student project has led to a publication at OAGM 2017, which received the best paper award, as well as a publication at OAGM 2018.

FINISHED (2017/12): Lung Lobe Segmentation from Thorax CT

Student who worked on the topic: Nicola Giuliani

Links to Nicola's master's thesis and VISAPP 2018 publication.

Aim: Investigate a lung lobe segmentation strategy for dual energy CT images provided by the Ludwig Boltzmann Institute for Lung Vascular Research (LBI-LVR).

Over the last years we have developed a GPU based vascular tree extraction scheme, which can be used for segmenting the pulmonary vessel tree from dual energy thorax CT data. The LBI-LVR is interested in determining early signs of pulmonary hypertension by performing a CT scan of the lung vessels and analyzing the vessel structures of disease and healthy patients.

When analyzing measures that try to predict pulmonary hypertension (PH) from the vascular tree, an interesting question is if a global measure for the whole tree is sufficient, or if individual measures of the different lung lobes give a more sensitive and specific prediction result. Therefore, in this master project we aim at developing a lung lobe segmentation algorithm, based on [1], to apply the quantitative measures predicting PH on the individual lobes. Students interested in this project should have C++ programming skills. Our libraries make extensive use of CUDA based image processing, therefore interest or even knowledge in this direction would be beneficial.

[1] Lassen, B.; van Rikxoort, E.M. ; Schmidt, M. ; Kerkstra, S. ; van Ginneken, B. ; Kuhnigk, J.-M. Automatic Segmentation of the Pulmonary Lobes From Chest CT Scans Based on Fissures, Vessels, and Bronchi. IEEE TMI 2013