Case study: Martin Hailstone
Martin came to biomedical imaging by an indirect route. A biochemist by training, it became increasingly apparent during his time as a research assistant in Singapore and Copenhagen that what these labs really needed was someone to help automatically analyse images. Having acquired some Matlab programming skills, he was able to do some simple image analysis and became more and more interested in this aspect of his work. In order to gain further (and more formal) training, he applied to the ONBI programme, whose other attraction was the chance to try out possible research projects in the field of image analysis applied to a biological context.
The project he eventually chose, aimed at understanding the development of the brain at the cellular and molecular level, is highly interdisciplinary, combining traditional biology and microscopy with image analysis and computer vision. Imaging brain development is difficult to do in humans, so a simpler animal is used as a model. The fruit fly brain is sufficiently complex that it requires many of the same processes in order to develop, but is much smaller. The brain can be imaged using confocal microscopy as it develops, but this creates vast amounts of data and processing this to create useful information is a problem. In collaboration with Dr Dominic Waithe, Martin has been developing software called QBrain to automatically count cells using supervised machine learning; very different to traditional approaches that rely on image segmentation. The software allows researchers to study the growth of the brain, including measuring important parameters like the cell division rate. This information can then be applied to look at the changes associated with cancer and developmental defects.
Martin’s approach places emphasis on simple, easy user input. While the project has an immediate specific biological focus – brain development – the resulting software, QBrain, is a more general tool, aimed at biologists with little or no programming knowledge, which makes analysis of large live cell datasets practical and straightforward. He has already started to publish his research1,2 and his research group is looking to make the tool available as open source software in 2017 (a pre-release version is available at