Computer vision and medical image analysis



Research group leader: Professor Fredrik Kahl

Our researchers are listed below.

About the research area Computer vision and medical image analysis

The aim of the field of image analysis and computer vision is to make computers understand images. To understand the width of applications one can consider what humans use their vision for. It can be finding a tumour in a three-dimensional magnetic resonance image, detecting a possibly dangerous traffic situation or recognizing a face. Some tasks which are simple for a human, such as recognizing an activity from an image, can still be very challenging for a computer. In other tasks, such as estimating an accurate metric 3D model from thousands of images, computers are already superior.

The problems arising in image analysis and computer vision are complex and require wide spectrum of tools from mathematics, computer science and statistics. To large extent, the research in our group is concerned with developing new methods and theory.  It can be the study of an optimization problem that is central in image analysis or the development of suitable mathematical models. There are also projects closer to the applications. Most importantly, several of the researchers within the group work medical image analysis and are connected to Sahlgrenska University hospital through MedTech West.

Research projects

> Anatomy segmentation in 3D medical images
Locating and segmenting anatomical structures such as the heart, vertebrae or different regions of the brain in an image is an important step for many clinical applications, including visualization, surgical planning and dose radiation therapy. It is also an important tool in order to obtain quantitative measures for diagnostic purposes, for example, the blood volume of the left ventricle.

> Robust Model Estimation in Computer Vision
This project is concerned with model estimation when the measurements are plagued by large quantities of outliers, i.e. measurements with very large errors. This problem arises frequently in computer vision applications. Model estimation with moderate measurement noise is a well-studied problem and even very large problems can be solved accurately. For robust model estimation we have a very different picture where heuristic methods without performance guarantees are often used. In this project we develop new efficient algorithms for these types of problems.

> DTI methods fo MRI brain image analysis

Lesions affecting the visual pathways in the human brain are common and may cause reduced visual acuity or visual field defects, either directly or as a result of surgery. These pathways can be visualised using tractography. The procedure is based on a combination of a magnetic resonance imaging technique known as diffusion tensor imaging (DTI) and computer-based image analysis...

> Patient specific brain segmentation with applications
The aim of this project is to develop state-of-the-art algorithms for automatically and accurately performing patient specific brain segmentation. The efficacy of each algorithm will be determined using real clinical data. Automatic segmentation will be quantitatively compared to manual segmentations performed by several radiologists...

Published: Wed 05 Sep 2012. Modified: Mon 13 Feb 2017