Automatic Segmentation of Medical Image Data Based on Deep Learning

Semester and Master's Work at the IT'IS Foundation - Automatic Segmentation of Medical Image Data Based on Deep Learning


Automatic segmentation of brain structures often is performed by means of image registration techniques that deform a template segmentation onto the target image, thereby generating a probable segmentation of the target patient data. The recent explosion of research and applications built on machine-earning approaches, such as deep neural networks, suggest promising alternatives.

The aim of this project is to develop a framework that allows deep learning models of segmented structures to be easily built, including systematic building of databases of annotated segmented image data and evaluation of different learning strategies and models. For this purpose, the student will be provided with a collection of multi-modal image data of the head region together with manually segmented tissue structures for training the models.

Depending on the interest and ability of the student, the resulting framework will be integrated in the commercial software iSEG. Good C++ or Python skills are prefered. As a prerequisite, the student will have attended at least one course on machine learning. The project will give the student the opportunity to gain hands-on experience with these exciting techniques.

The extent of the project will be defined and finalized according to the interests and background of the student.

The workplace will be at the IT’IS Foundation in Zurich. The workflow will include:

  • a survey of the approaches reported in recent literature
  • development of the mesh-generation framework in C++
  • investigation of efficient computational architectures (e.g., GPU processing)
  • assessment of the quality and robustness of the mesh generation scheme in realistic scenarios

 Please contact Dr. Bryn Lloyd for more information and further details.


Dr. Bryn Lloyd
Type of Work:



Niels Kuster

Please send applications to Mimi Sun at