Figure 1.
A grating is placed at the image plane of a commercial microscope (the dashed box) to create different diffraction orders. Filter masks are projected onto a spatial light modulator (SLM) placed at the back focal plane of a Fourier lens L1 to low pass filter the 0th order beam (reference) and allow the entire 1st order beam (imaging field) to pass through. Another Fourier lens L2 recombines the two beams to create an interferogram on a CCD. A Hilbert transform is then used to reconstruct the phase information from the interferogram.
Figure 2.
The imaging system wDPM is communicated to the user through the camera interface. User can control several parameters like exposure time, image size, etc. Several CUDA based modules are implemented to speed up the processing: The phase reconstruction module recovers the phase information induced by the objects in the field of view; the segmentation module isolates and labels individual objects in the field of view; and the blood analysis module was designed specifically for blood smear analysis application to calculate different morphological parameters of red blood cells. All the results are computed and displayed in real time at the speed of up to 40 frames/s. The computed parameters can be saved in a small file and sent anywhere for remote diagnosis.
Figure 3.
Phase image reconstruction and processing procedure.
(A) An interferogram acquired from the wDPM system; (B) Reconstructed phase map; (C) The output of the segmentation module; (D) A snapshot of the screen visualizing different parameters of each red blood cell including volume (V), surface area (SA), projected area (PA), sphericity index (SI), minimum cylindrical diameter (MCD) and the mean height.
Figure 4.
Morphological parameter distributions of red blood cells of different patients.
(A). Red blood cell volume distribution of a healthy patient (N = 6,181 cells) and patients with microcytic (N = 8,442 cells) and macrocytic (N = 4,535 cells) anemias. The inset compares the measured results with the results obtained from a Beckman Coulter counter on the same samples with vertical and horizontal error bars are standard deviations from the measurement data and the Beckman counter, respectively. The red line illustrates the line y = x, of perfect agreement; (B). Red blood cell surface area distribution of the three patients. The inset elaborates the statistical information of the distributions with the range within one standard deviation, the boxes indicating the interquartile range (IQR) and the square symbol showing the median.
Figure 5.
Examples of other morphological parameters of red blood cells.
A. Sphericity index distribution; B. Minimum cylindrical diameter distribution.