- Biomedical Optics & Medical Imaging
- Defense & Security
- Electronic Imaging & Signal Processing
- Illumination & Displays
- Lasers & Sources
- Micro/Nano Lithography
- Optical Design & Engineering
- Optoelectronics & Communications
- Remote Sensing
- Sensing & Measurement
- Solar & Alternative Energy
- Sign up for Newsroom E-Alerts
- Information for:
Electronic Imaging & Signal Processing
Automatic identification of micro-organisms in real-time
A 3D optical imaging system may provide a fast and non-destructive technique to visualize and identify living specimens.
5 September 2006, SPIE Newsroom. DOI: 10.1117/2.1200609.0369
Many difficulties are associated with the real-time identification of unknown living organisms in the context of security and disease control. Conventional inspection methods are usually laboratory-based and involve biochemical processing, which requires trained personnel and takes several hours if not days to complete. Moreover, the target organisms are not rigid objects and vary in size and in shape. There can also be many morphological variations among the same species.1,2
Optical information systems have proven to be very useful for 2D pattern recognition systems.3 However, research has recently been focused on 3D optical information systems because of their vast potential for applications such as object recognition, image encryption and 3D display. Among imaging techniques, digital holography has become very attractive for the acquisition and visualization of 3D information.4,5 In digital microscopic holography, 3D information about micro-organisms can be recorded and numerically reconstructed to provide various slices of their complex (magnitude and phase) images. Our research effort is based on the idea that such 3D information, reconstructed at arbitrary depths and perspectives, might be able to provide more discriminating features for micro-organism recognition.
In our work, we use a type of holography called single-exposure on-line (SEOL) digital holography to acquire 3D information.6Figure 1 shows a schematic of a SEOL microscopic digital holography system. The complex wave transmitted from micro-organisms by Fresnel diffraction is recorded as a digital hologram by means of a Mach-Zehnder interferometer. A complex amplitude image is then reconstructed from the recorded Fresnel diffraction field on a computer. SEOL digital holography outperforms competitors, such as off-axis and phase-shifting on-axis digital holography.7 Since only a single-exposure is required to record the hologram, SEOL holography is more robust to sensor noise and environmental fluctuations and can be used to monitor real-time dynamic events such as micro-organism movement. Another benefit is that 3D scenes can be focused numerically rather than mechanically, as required in conventional microscopy.
Figure 1. A single-exposure on-line microscopic digital holography system shown with its various components.
Figure 2 depicts the flowchart of a 3D imaging and recognition system. During the first stage of the process, a SEOL digital holographic system performs 3D imaging of the unknown micro-organism sample. An inverse Fresnel transformation is then used to numerically reconstruct complex waveforms at different depths and perspectives. The reconstructed objects are segmented and features are extracted at the next stage. Gabor-based wavelets8 extract geometrical shape features by decomposing the complex micro-organism information into the spatial-frequency domain. The next stage matches features, using a technique called rigid graph matching (RGM).9 During RGM, a reference database is searched for shapes with morphological features similar to those of the unknown micro-organism by measuring similarities and difference functions. Feature vectors are defined at the nodes of two identical graphs on the reference database images and the reconstructed micro-organism images, respectively. The reference shape represents the unique morphological features of a given micro-organism that can potentially match unknown micro-organism samples.
Figure 2. A flowchart depicting a 3D imaging and recognition system.
Experimental results obtained for the visualization and recognition of two filamentous micro-organisms, sphacelaria and tribonema aequale alga, are shown in Fifores 3 and 4. Figure 3(a) shows a 2D image of sphacelaria alga. Figures 3(b) and 3(c) are reconstructed images (magnitude only) of sphacelaria and tribonema aequale alga samples, respectively.
Figure 3. A 2D image of sphacelaria is shown in (a). Also shown are the reconstructed images of sphacelaria (b) and tribonema aequale (c) using single-exposure on-line digital holography.
Our recognition process used feature extraction and graph matching to localize the predefined shapes of sphacelaria alga. Figure 4(a) shows its reference graph that consists of 25 ×4vc 3 nodes. Figure 4(b) shows another sphacelaria alga sample used as the input image with the graph matching results. In this case, the reference shapes were detected 65 times along the filamentous structure.
Figure 4. (a) A reference image of sphacelaria with a reference graph, and (b) an input image of sphacelaria and input graphs after the matching process.
We have presented preliminary results on the real-time automatic recognition of micro-organisms based on examining their simple morphological traits with SEOL microscopic digital holography. In the future, we plan to use the features of the reference shapes to design further training procedures. Also, we plan to investigate combining several morphological traits to improve unknown micro-organism recognition.
This work is supported by a grant from Defense Advanced Research Projects Agency.
Electrical and computer engineering, University of Connecticut
Dr. Bahram Javidi is Board Of Trustees Distinguished Professor at the University of Connecticut. He is a fellow of several professional societies including SPIE. He was awarded the Dennis Gabor Award in Diffractive Wave Technologies by SPIE in 2005.
Image Recognition and Classification: Algorithms, Systems, and Applications,
Marcel Dekker, New York, 2002.
Handbook of Holographic Interferometry,
Wiley, VCH, 2005.
5. J. W. Goodman,
Introduction to Fourier Optics 2nd,
McGraw Hill, Boston, 1996.
9. M. Lades, J. C. Vorbruggen, J. Buhmann, J. Lange, C. v. d. Malsburg, R. P. Wurtz, W. Konen, Distortion invariant object recognition in the dynamic link architecture,
IEEE Trans. on Computers,
Vol: 42, no. 3, pp. 300-311, 1993.