Phase-contrast magnetic resonance imaging (PC-MRI) is an imaging technique that allows to measure time-resolved volumetric flow patterns (4D flow). In particular, the ability to measure velocity and turbulence in the aorta, makes 4D flow MRI very promising for future assessment of aortic pathologies in a clinical framework. However, automatic processing tools are necessary to extract the valuable information contained in 4D flow images. Calcific aortic valve disease (CAVD) is the most common valve disease in developed countries and is defined by a narrowing on the aortic valve orifice called aortic stenosis (AS). The presence of AS severely affects the flow in the aorta and quantifying its severity (eg. area reduction) is fundamental to improve diagnostic and treatment.
Convolutional Neural Networks (CNN) are known for their capabilities in image analysis and pixelwise prediction. Recent developments in deep learning [1] show that it is possible to predict AS from echocardiography data.
Reference:
[1] Related Master Thesis by Xiaolu Guo: https://dspace.mit.edu/bitstream/handle/1721.1/139153/Guo-xiaolu20-meng-eecs-2021-thesis.pdf?sequence=1&isAllowed=y
Phase-contrast magnetic resonance imaging (PC-MRI) is an imaging technique that allows to measure time-resolved volumetric flow patterns (4D flow). In particular, the ability to measure velocity and turbulence in the aorta, makes 4D flow MRI very promising for future assessment of aortic pathologies in a clinical framework. However, automatic processing tools are necessary to extract the valuable information contained in 4D flow images. Calcific aortic valve disease (CAVD) is the most common valve disease in developed countries and is defined by a narrowing on the aortic valve orifice called aortic stenosis (AS). The presence of AS severely affects the flow in the aorta and quantifying its severity (eg. area reduction) is fundamental to improve diagnostic and treatment.
Convolutional Neural Networks (CNN) are known for their capabilities in image analysis and pixelwise prediction. Recent developments in deep learning [1] show that it is possible to predict AS from echocardiography data.
Reference:
[1] Related Master Thesis by Xiaolu Guo: https://dspace.mit.edu/bitstream/handle/1721.1/139153/Guo-xiaolu20-meng-eecs-2021-thesis.pdf?sequence=1&isAllowed=y
In this project, the aim is to model an automatic way of classifying the degree of stenosis on synthetic and real 4D flow MRI data. Therefore, a CNN will be trained using synthetic data to predict the degree of severity of stenosis under known circumstances. After successful implementation of the synthetic cases, the model will be evaluated on real 4D flow data and further analysis of Aortic properties can be performed.
**We offer**
- A warm start of the project with the state-of-the-art knowledge of the group in this field
- A chance to connect state of the art research in MRI with modern ML tools
**We expect you to have**
- Ongoing bachelor in computer science, Biomedical Engineering, Physics, or related fields
- Strong interest in Deep Learning, programming experience in Python, additional experience with Tensorflow or PyTorch is a plus
- Basic Knowledge of imaging principles in MRI
- Passions in Research and state of the art NN applications (which is the most important thing)
In this project, the aim is to model an automatic way of classifying the degree of stenosis on synthetic and real 4D flow MRI data. Therefore, a CNN will be trained using synthetic data to predict the degree of severity of stenosis under known circumstances. After successful implementation of the synthetic cases, the model will be evaluated on real 4D flow data and further analysis of Aortic properties can be performed.
**We offer**
- A warm start of the project with the state-of-the-art knowledge of the group in this field - A chance to connect state of the art research in MRI with modern ML tools
**We expect you to have** - Ongoing bachelor in computer science, Biomedical Engineering, Physics, or related fields - Strong interest in Deep Learning, programming experience in Python, additional experience with Tensorflow or PyTorch is a plus - Basic Knowledge of imaging principles in MRI - Passions in Research and state of the art NN applications (which is the most important thing)
If you are interested in this work and ready for a new challenge, please feel free to contact Gloria Wolkerstorfer wolkerstorfer@biomed.ee.ethz.ch or Pietro Dirix dirix@biomed.ee.ethz.ch. The project focus can be adjusted to the student’s interest and experience. Interested students are asked to send a CV and a transcript of records of the Masters. Supervising professor: Sebastian Kozerke (kozerke@biomed.ee.ethz.ch)
If you are interested in this work and ready for a new challenge, please feel free to contact Gloria Wolkerstorfer wolkerstorfer@biomed.ee.ethz.ch or Pietro Dirix dirix@biomed.ee.ethz.ch. The project focus can be adjusted to the student’s interest and experience. Interested students are asked to send a CV and a transcript of records of the Masters. Supervising professor: Sebastian Kozerke (kozerke@biomed.ee.ethz.ch)