DIMSCAN Description

DIMSCAN is a semi-automatic digital image microscopy system for quantitating relative cell numbers in tissue culture plates. Cytotoxicity assays measured by DIMSCAN using fluorescein diacetate (FDA), a dye accumulating selectively in viable cells, can achieve a 4 log dynamic range at 4 to 7 days and correlate with colony forming assays (U.S. Patent 6,459,805, Oct 1, 2002).

The system consists of an inverted fluorescence microscope, stepper motor scanning stage, the stage controller, CCD camera and a microcomputer running the main application, which controls stage movement and processes CCD camera images (see system diagram).

Relative numbers of viable cells are determined by evaluating FDA fluorescence intensity, with background fluorescence eliminated by digital thresholding and a quenching of florescence in non-viable cells with the vital stain eosin Y. The current system uses an Olympus IX50 inverted florescence microscope, a Prior motorized stage, a Qimaging Microimager II camera, and a Pentium III computer running Windows 2000 (see system figure).

Rectangular plates containing 6 to 384 wells per plate can be employed. Average scan time for a 96 well plate is 6 minutes. To maximize the automation of the system, an autofocus feature based on autocorrelation function was implemented. The program stores reconstructed images of scanned wells, which are displayed to the user when scanning is completed, allowing rescanning of selected areas. The picture of the main program window shows the thumbnail images for the 60 wells analyzed by the instrument. One of the 60 wells has been expanded in a magnified view. The plate contains FDA-stained cells (the CHLA-233 retinoblastoma cell line).

Sensitivity and linearity of the system was validated with FDA stained CEM leukemia cells deposited using a fluorescence activated cell sorter such that there was 1, 10, 20, 50, 100, 200, 500, 1000, 2000, 5000 and 10000 cells per well, 8 replicates per plate. A linear relation (linear regression of the first order) between number of viable cells deposited vs. DIMSCAN quantitation was observed: r² = 0.99967 ± 0.0003, documenting a dynamic range over 4 logs (see the graph). The linearity and dynamic range were preserved during repeated scanning of the same plate up to five times with the correlation coefficient r² > 0.99967. Detection of single cells by DIMSCAN was achieved in 93 % of wells seeded, while no false-positive detection has been observed.

Linearity test was also performed for 384 well plates. Similar to 96 well plates the system was validated with FDA-stained CB-33 cells (a B-lymphoblastoid cell line) deposited using Precision 2000 robot so that there was 1, 10, 50, 100, 300, 500, 700, 1000, 2000, 3000 and 5000 cells per well. A linear relation between number of viable cells deposited vs. DIMSCAN quantitation was observed again: r² = 0.99617, documenting a dynamic range over 3 logs (see the graph).