Autofocus And Positioning Research Of DNA Chip Electrophoresis Detection

Microchip-based electrophoresis moves the conventional capillary electrophoresis to a chip with size of square centimeters by using microfluidic technology. It is a kind of fast, efficient, low cost micro-technology. It integrates sample injection, reaction, separation, detection and other processes together. Autofocus (AF) and positioning of microchip mainly studied in this thesis plays an important role in detection system. Software implementation of autofocus based on microscope includes focus evaluation function, focus search algorithm and the control algorithm, which bring the detection window entered and in focus.In this thesis, we design the overall framework of the software, implement and optimize AF system based on microscope for finding channel’s focus position in microchip.Microchip-based electrophoresis is detected using laser-induced fluorescence detection and this method has high sensitivity. To ensure the intensity and SNR (signal noise ratio) of fluorescence signal detected by the mentioned method, the channel in microchip should lie in the focus plane of the microscope. Two pivotal issues of AF technology based on digital image technology are focus evaluation functions and search algorithms.The object to be focused is a three-dimensional channel in microchip. Since the size of the channel in microfluidic chip is in range of microns, if the channel is not at the focus plane, the signal intensity detected will be reduced or lost. As a result of AF, focus plane will lie on the focus position of the upper surface of the channel, the lower surface of the channel or other position between them. Then the detection plane will be adjusted to the center of channel according to the depth of channel by using focus plane as reference plane, finally the strongest signal will be detected.In this study, Tenengrad function and variance function are chosen as focus measures. However, Tenengrad curve has two peaks. The first peak corresponds to the focus position of upper edge of the channel and the second one corresponds to the focus position of lower edge. The variance curve has only a relatively flat peak and the peak corresponds to the space from the beginning of the first peak of Tenengrad curve to the end of the second peak.There are some limitations for traditional search methods due to focus measures and three optimized search schemes are proposed. The first method is improved hill-climbing search method. It combines traversal search method based on variance function and hill-climbing search method based on Tenengrad function. This method can get clear image even if focus evaluation curves have two obvious peaks. The second one is fast search method which utilizes Fibonacci search method based on variance function to approximate focus position and engages hill-climbing search method based on Tenengrad function to find the exact focus position. The last one is valley search method. It use focus evaluation curves with obvious valley to narrow search range to half every time repeatedly until search precision meets the system requirement. Valley search method is the most time consuming searching method with good accuracy. Improved hill-climbing search method costs relatively long time but with better robustness. It can be used in various situations and has accurate focus results. Fast search method can not only reduce searching time but also guarantee auto-focusing accuracy. However, it requires unimodal evaluation curves.To find the best conditions for AF, we do comparative experiments with different media in channel like air, ultra-pure water and gel. With different media we get different channel images, and the same evaluation function will show different performance. We choose six focus strategies when medium is air and four for ultra-pure water medium, three for gel medium. It turns out that all six focus strategies used for air medium give right results but with different efficiencies. However, focus strategies used for ultra-pure water medium and gel medium might have error for AF results. When medium in channel has larger refractive index, more light will go through the channel and we will get brighter images. Under this circumstance, it will affect the performance of auto-focus. When the channel has no medium but air in it, we will get the best focus results.A lot of researches about detection of microchip-based electrophoresis have been done. But there are no details about autofocus in detection.In this thesis two problems are solved. A microscope-based AF system suitable for microchip-based electrophoresis is designed. After experimental verification, we get accurate focus results. And we further study the optimum focus condition, when medium is air the focus accuracy is the highest by comparing experiments of air, ultra-pure water or gel medium in channel.