Table 1.
Experiments | Human counting | R-MOD counting | ||
---|---|---|---|---|
Duration | Count (mean ± s.d.) | Count (7/10/15-μm) Throughput | Number ratio (est. 100/34/10) | |
Exp-1 (6 μL/min) | 4 sec | 2,343 ± 6.3 | 2,346 (1,540/549/257) 587 cells/s | 100/35/10 |
Exp-2 (10 μL/min) | 4 sec | — | 3,103 (2,423/462/218) 776 cells/s | 100/19/9 |
Exp-3 (18 μL/min) | 4 sec | — | 5,362 (3,055/1,306/1001) 1,340 cells/s | 100/43/33 |
Exp-Large (18 μL/min) | 17 min | — | 635,844 (396,122/133,891/105,831) 616 cells/s | 100/34/27 |
From Exp-1 to Exp-3, 2,000 recorded images (during 4 sec) were processed by R-MOD. To quantitative evaluation of counting results, six human participants counted flowing beads by viewing the recorded image sequence of Exp-1. Error of counting result between human and R-MOD is 3. Since beads in the suspension are not uniformly distributed, every experiment shows different actual throughput (from 600–1,300 cells/s). Exp-Large analysed a large number of images (>510,000) during 17 minutes to show the high-throughput and real-time capability of the proposed system. Number ratio is calculated based on the number concentration of the mixed bead to roughly compare the actual detected counts.