While automated blood cell counters have made great progress in detecting abnormalities in blood, the lack of specificity for a particular disease, limited information on single cell morphology and intrinsic uncertainly due to high throughput in these instruments often necessitates detailed inspection in the form of a peripheral blood smear. Such tests are relatively time consuming and frequently rely on medical professionals tally counting specific cell types. These assays rely on the contrast generated by chemical stains, with the signal intensity strongly related to staining and preparation techniques, frustrating machine learning algorithms that require consistent quantities to denote the features in question. Instead we opt to use quantitative phase imaging, understanding that the resulting image is entirely due to the structure (intrinsic contrast) rather than the complex interplay of stain and sample. We present here our first steps to automate peripheral blood smear scanning, in particular a method to generate the quantitative phase image of an entire blood smear at high throughput using white light diffraction phase microscopy (wDPM), a single shot and common path interferometric imaging technique.

Date Published: 11 March 2015
PDF: 5 pages
Proc. SPIE 9336, Quantitative Phase Imaging, 93361Y (11 March 2015); doi: 10.1117/12.2080200

Published in SPIE Proceedings Vol. 9336:
Quantitative Phase Imaging
Gabriel Popescu; YongKeun Park, Editor(s)