Image processing program detects tumor cells

A researcher at the University of Twente’s MIRA Research Institute (Enschede, The Netherlands) has developed image processing software that can count the circulating tumor cells in a blood sample and identify them by type.

Previous research has shown that the higher the number of circulating tumor cells (CTCs), the shorter a patient’s life expectancy. However, it is not easy to trace CTCs in the blood because there is no common set of characteristics shared by all of these cells. Also, there are very few of them in the blood - just one CTC to every billion healthy blood cells.

The blood of patients is currently assessed by detecting CTCs with iron particles that bind to potentially harmful cells in the blood. A magnetic field then separates these cells from other components of the blood. The remaining cells are then stained to improve their contrast, and photographed using a camera mounted on a microscope. In current medical practice, specialists assess the recorded images by eye and count the number of CTCs found.

Sjoerd Ligthart’s image analysis software, which was developed using a combination of Mathworks’ (Natick, MA, USA) Matlab and the DIPimage toolbox from the Delft University of Technology (Delft, The Netherlands) has eliminated the need for this procedure by automatically counting, then characterizing the CTCs, enabling specific cancer types to be identified.

Ligthart carried out the PhD research at the Department of Medical Cell Biophysics (MCBP), part of the University of Twente’s MIRA research institute. The research was conducted in collaboration with Veridex (Raritan, NJ, USA) a subsidiary of Johnson and Johnson.

Ligthart’s PhD thesis describing the work in detail is available here.

-- by Dave Wilson, Senior Editor, Vision Systems Design