Statistician Jörn Schulz has developed new methods for analysis of medical images in his PhD research. They could enable more precise diagnosis and treatment of diseases including prostate cancer.
Written by Vibeke Os 3 January 2014.
Various kinds of imaging can be used when illness or injury is suspected, depending on what needs to be examined: bones, tendons or soft tissue. Manual interpretation of medical images can be time consuming and can leave room for inaccuracies or subjective influences. Statistical shape analysis has therefore become an important tool for diagnosis, planning of surgical procedures and fine-tuning of radiotherapy for sites with cancer cells.
Jörn Schulz is an applied mathematician, originally from Berlin, who has specialized in statistical analysis of spatial medical data. On 18 December 2013, he completed a doctoral degree at Tromsø Telemedicine Laboratory and the Department of Mathematics and Statistics. The title of his thesis was “Statistical Analysis of Medical Shapes and Directional Data”. His research involved developing methods for statistical analysis of shapes that could be applied to medical problems. All objects can be described by their three-dimensional shape and Schulz has been involved in an exciting interdisciplinary project that involves medical and statistical research. The work has resulted in new methods for analysis of medical images.
Imagine a salmon caught in a fishing net. The shape of the salmon could be described by the coordinates of the net – a skeleton of coordinates and vectors. Schulz's work has helped to specify the shape of medical objects using a finer mesh. He has used the analyses in three different medical contexts: 1) Description of the shape and position of the prostate gland, 2) analysis of knee joints, 3) testing differences in the brains of people with schizophrenia and healthy people.
|The shape of medical objects can be described as a grid of vectors. As an example, this can be used to make analysis of radiological images more accurate.|
A more precise description of the prostate gland's shape and position can help to save time in radiotherapy of patients with prostate cancer, both because it is easier to detect divergent shapes and because radiation can be directed more specifically. New ways to model mechanical movements of joints have also been studied. Defective or abnormal movement in joints such as the elbow and knee can be challenging to describe and analyse. Schulz has therefore tried to find effective methods to analyse the mechanisms of the knee joint during movement. This knowledge will be valuable during surgical procedures and treatment.
Medical imaging is widely used to detect changed volume and shape in human organs and structures such as tendons and bones, but it is also used to explore the structure of the human brain. Illnesses such as Alzheimer's disease and schizophrenia can be analysed using medical shape analysis, where one looks for differences in the brains of ill and healthy people. Medical imaging is also used to detect any changes in the brains of patients after treatment. But the differences may be small and difficult to detect with the naked eye. Mathematical models can then be used to enable analysis even of small differences in brain structure. In his work, Schulz studied a new method to test the mean difference in volume and shape of the hippocampus. The method was used to describe differences in the brains of people with schizophrenia and a control group.
Supervisors for his PhD research were Fred Godtliebsen at the University of Tromsø as well as Stephen M. Pizer and Steve Marron from the University of North Carolina at Chapel Hill.