This project is in collaboration with a pharmaceutical company AstraZeneca based in Cambridge. Digital Phase Contrast (DPC) microscopic image is a new form of imageries which uses optical microscopy technology to enhance the contrast of medical images. DPC is featured as a label free technology and is able to potentially capture undistorted cell/tissue morphology without straining (avoiding artefacts associated with fixation and labelling). Compared to the traditional phase contrast images, DPC images produce more homogeneous image background. Hence the use of DPC images could potentially lead to much reliable image segmentation results for medical research.
At AstraZeneca, two forms of DPC images can now be obtained from its Yokogawa CV7000 microscopy system. These two forms of DPC images are called the phase type DPC images and fluor type DPC images. As a novel and untested imaging modality, it is clearly required i) to test the safety of using it – e.g. if DPC images reflect the true tissue/cell morphology, and ii) to develop novel image segmentation methods for the processing of such imageries.
As a starting point, cell line image samples – without complex tissue structures, will be used for this study, where individual cells are required to be segmented. Foreseeable image analytical difficulties include heterogeneous image background and faint (non-continuous) cellular boundaries. The robustness of the developed segmentation algorithm can be tested by comparing the two forms of DPC images (phase type and fluor type). It can also be compared with the cell segmentation results from traditional phase contract images acquired from a same machine. The safety of using this novel form of images can be further validated by comparing with fluorescent labelled images with typical cell nuclear, membrane and/or cytoplasmic staining.