Computer-based microscopy speeds automated cell morphology
CellaVision AB (Lund, Sweden; www.cellavision.com) is using off-the-shelf cameras, x-y positioning stages, and host PCs in the development of its DiffMaster system, a product that automates the time-consuming counting of white blood cells (leukocytes) and examining blood films and allows technicians to devote their attention to more-demanding tasks. To classify white blood cells, a technician places a slide under an automated microscope and starts the system. For each white blood cell the system finds, an image is taken for subsequent analysis and preclassification according to cell class. Up to eight 1 × 3-in. specimen slides can be analyzed in a batch. The results of the analysis are displayed on a computer screen, permitting the technician to study all of the cells from a given class in the same window.
Based on the company's Cytologica software, the DiffMaster system identifies and preclassifies white blood cells based on parameters that have been optimized on a large database of cells. This classification has been made by experts using an artificial neural-network engine. Once captured, cell images can be sent via e-mail to other physicians in-house or at other hospitals.
CellaVision is standardized on a BX50WI microscope from Olympus America (Melville, NY, USA; www.olympusamerica.com), a fixed-stage upright microscope that uses built-in Koehler illumination for transmitted light. The DiffMaster system uses a Model H138 x-y stage from Prior Scientific (Rockland, MA, USA; www.prior.com) that provides 9 × 3-in. travel, repeatability to ±1.0 µm, and step size from 0.1 to 5.0 µm. Under PC control, the x-y stage is driven by Prior's PSH29 x-y-z series of motor controllers using the PC RS-232 serial port.
Slides placed under the microscope stage are imaged by a DXC-9100P camera from Sony Electronics (Park Ridge, NJ, USA; www.sony.com/videocameras). This 1/2-in. 3-CCD color video camera uses progressive-scan technology and built-in frame memory that transfers images in a noninterlaced fashion to a PCI-based frame grabber. Designed to capture from standard or variable analog monochrome or component RGB frame scan sources, the Meteor-II Multichannel C/4 card from Matrox Imaging (Dorval, QC, Canada; www.matrox.com/imaging) supports acquisition from progressive-scan component RGB cameras. A proprietary autofocus algorithm digitizes the images. Then image data are transferred to the host computer where CelliVision's proprietary Cytologica image-acquisition and preclassification software identifies and preclassifies the cells.