PhD Student in Deep Learning for Image Segmentation

Segmenting pathology in medical images is crucial for clinical studies and applications, therefore it plays an important role in our research at the Department of Biomedical Engineering (DBE): Affected organs need to be delineated, or lesions, such as tumors, must be traced for monitoring disease progress and assessing treatment effects.

If done manually or semi-automatically, image segmentation is a tedious and error-prone process. Even if supervised learning-based automated approaches are used, still a substantial amount of training samples, and thus human interaction, is required. In this project, we therefore aim at automating pathology segmentation even further.

Your position

You will explore techniques from weakly supervised learning to make sample segmentations dispensable or obsolete for learning the task of pathology segmentation. You will investigate in possibilities to set up a disease-agnostic approach that learns to identify, find, and outline the visual manifestations of a certain pathology, simply from two sets of given exemplary images, one showing patients and one showing healthy subjects for comparison. You will try, evaluate, and refine your approach on different diseases and imaging modalities in public as well as internal datasets.

Your profile

• Master's or comparable degree in computer science, machine learning, or related field
• Profound knowledge in deep learning, image processing, and computer vision
• Previous experience with a common deep learning framework (PyTorch, TensorFlow)
• Strong English language skills

 

We offer you

• The opportunity to work, study, and do research in a quickly emerging field
• Enrollment in an established three-year Biomedical Engineering PhD program (in English)
• Competitive salary by national and international standards (salary and social benefits according to University of Basel rules)
• Interdisciplinary work environment: we are a team of computer scientists, engineers, mathematicians, and physicists that closely collaborate with medical and clinical partners
• Work and life in the Basel area, the sunny and cosmopolitan region in the Northwest of Switzerland right next to France and Germany

 

Application / Contact

Please submit your application no later than September 16, 2018 as a single PDF document via e-mail (subject: "Application: PhD Student in Weakly Supervised Pathology Segmentation", contact: see below). The document should include your letter of motivation, your CV, transcripts of your bachelor's and master's diploma, your TOEFL/IELTS scores if available, as well as contact information (address, phone, e-mail) of up to three references.

For applications and questions, please contact Dr. Simon Pezold (simon.pezold@unibas.ch).