| 研究実績の概要 |
In the first year, we have constructed a high-frame-rate cell recognition system by designing multi-object feature extraction circuits with FPGA and a PC-based, multi-object tracking algorithm.
(A) Prepare and confirm the real-time frame-straddling function of high-speed vision. We have improved our IDP Express system and verified that the frame-straddling function is working correctly with resolution 512×256 at 4000 fps and that the frame-straddling time can be adjusted from 0 to 0.25 ms in 9.9-ns steps. Furthermore, perform required calibrations for the two camera inputs (get affine transformation parameters). We also confirmed that the captured images contain no motion blur when we set the flow speed to 2 m/s by setting set the exposure time of the two camera heads to the minimum possible exposure time for the IDP Express, 6.25 us.
(B) Design and verify the multi-object tracking and recognition algorithm using offline videos. Firstly, we captured offline, high-frame-rate videos for sea urchin egg cells flowing rapidly in straight type microchannels (wide: 200 us, deep: 100 us). And then, we created a multi-object tracking algorithm on the basis of the offline videos and the multi-object features for two frame-straddling camera heads extracted by using software version of cell-based labeling algorithm.
(C) We designed cell-based labeling circuits for rapidly extracting multi-object features. In this hardware design, the cell size is 4×4, which can provide a more accurate segmentation result, compared to my previous study (cell size: 8×8).
|
| 今後の研究の推進方策 |
In the second year, we plan to analysis different type of cells fast-flowing in different type microchannels, to verify the performance of our system.
1. Analysis shape and motion of cells fast-flowing in microchannels. We plan to perform several experiments using sea urchin egg cells fast-flowing in a straight microchannel. To compare the difference between normal and Bouin's fixed sea urchin egg cells, all experiments are performed using those two kinds of sea urchin egg cells. In the first experiment, sea urchin egg cells flowing at different speeds are observed to quantify how their shapes are deformed in fast microchannel flows (flow speed from 125 ul/min to 2000 ul/min). In the second experiment, sea urchin egg cells are observed to quantify their aged deterioration after spawning by inspecting their deformed shapes in fast microchannel flow (500 ul/min).
2. Perform fast recognition of fertilized cells fast-flowing in microchannels.
The shape and inner tissue of fertilized sea urchin egg cells will changed rapidly in two or three days after fertilization. We plan to get technical support from Sakamoto-sensei at Hiroshima University, who can provide us adequate fertilized sea urchin eggs. In this experiment, we perform fast cell recognition of fertilized sea urchin egg cells fast-flowing in straight microchannels hour by hour using hardware extracted bounding box of cells. This experiment shows our system can be used for high-frame-rate image-based (in ROI region of cells) shape recognition and analysis.
|
