Chapter 1 Introduction
Pathogen microorganism detection is a worldwide issue, and scientists already invented several methods for pathogen microorganism diagnosis, but there still have some disadvantage for current biometric pathogen diagnostic methods, such as complicated process involved, time-consuming, large quantity sample needed, non-in situ detection...etc. Took some conventional biometric detection methods for example. The culture and colony counting method such as the standard procedures for the detection of Escherichia coli(E.
coli) in water[1] is a conventional bacteria detection method, which already being invented
for over decades, and still is a popular method for the diagnosis of bacteria. For executing the culture and colony counting method, solution sample need to be diluted to different concentrations and plated on petri dishes. The petri dishe with proper concentration of sample, bacteria are uniformly spread on petri dishes in single bacterium level, can be used for further diagnosis. After bacteria culturing process, each bacterium will divide and breeding into a single colony. By analyzing those colonies, scientist can find out what kinds of bacteria are in the sample solution, and the concentration of each kind of bacteria during the beginning of diagnosis. The concept of this method is simple and it can be used for obtaining reliable results, but it is a time-consuming and procedure complicated method.Immunology-based methods[2] rely on the ability of an antibody to capture a specific antigen, and producing measurable signal in response to the antibody-antigen coupling process. Most of the Immunology-based methods are involved with some sort of labeling process (ex. fluorescent label). Immunology-based method is a much faster detection method, but still is a matter of hours. The detection accuracy is enormously relied on the quality of the antibody and this detection method can only be used for the detection of well-known target antigen.
In recent days, several interdisciplinary types of research show the potential of combining techniques from different field, which might be able to highly enhance the performance of traditional methods. Electrical measurement-based biosensors [3] is a good example of an interdisciplinary technique. In proper conditions, measurement of electric signal can be very fast and sensitive. Electrical data is digital data, which can be directly analyzed by computer. Most of the electrical measurement-based biosensors can be fabricate as portable device, making in situ detection possible. Impedance measurement method is one popular kind of electrical measurement-based detection method, and generally it can be
‧
method[3]. Bacteria metabolism can affect the impedance of sample solution. By measuring the impedance changes of sample solution, this method can be used for finding out the concentration of targeting bacteria. It is a fast detection method, but still is a matter of hours, and can only obtain ensemble results. The second type is bioreceptors-based impedance measurement methods. Similar as immunology-based methods, this method rely on the ability of a bioreceptor to capture a specific target. Before the measurement, the corresponding bioreceptor of the target were immobilized on the electrode, that impedance value between electrode will change when the targeting particle captured by the bioreceptor.
This method is a very fast detection method, can obtain result within a hour, but it can only be used for the detection of well-known target.
In 2014, Leonardo Lesser-Rojas and co-workers fabricated a device with nano-electrodes and a nano-channel lies between the gap of nano-electrodes and perpendicular to them, which is called as the tandem array nano-gap biosensor[4]. The nano-channel is used for confining the quantity of targeting particles from macro-scale level to single particle level, and the size of electrode pairs are similar with targeting sample, that this tandem array nano-gap biosensor can be used for rapid (real-time) electrical properties measurement of nanoparticles in single particle level. If the difference of electrical properties between different kinds of particle is large enough for distinguishment, this device might be able to be used for identifying nanoparticles in label-free. Inspired by their works, the main goal of this thesis is developing a device which can directly detect larger electrical properties of bioparticles in single particle level. This device has potential to achieve small quantity, single particle, label-free, in situ detection and identification of pathogen microorganism.
Since electrical properties for lots of bioparticle in single particle level still remain unknown, some other conventional data is needed as control group, and image data, a common type of data already used for the studies of microorganism in past few decades, has been chosen as our target. After we obtain image data, it can be compared with electrical data to see if there is any correlation between them. Abstracting image data for thousands of microorganism one by one manually will consume too much time, and the result might not be objective enough, that the secondary goal in this thesis is building up an image processing system for cell identification as a supporting tool, which can be used for obtaining image data of microorganism in single particle level faster and with more subjective standard.
The critical step for bacteria identification is the cell segmentation process. Most cell segmentation methods involve in multiple segmentation process, making whole process very complicated. MicrobeTracker [5] use Otsu’s thresholding method for obtaining initial
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
segmentation image, then applying edge detection and watershed segmentation. In the last step, using active contours model for obtaining edge of bacteria. MAMLE [6] use multiple threshold for learning shape feature of bacteria, then segment cells and update the shape of cells by maximizing likelihood parameter. These kind of cell segmentation program are complicated, and is very consumption of computational resource. Other than that, most of them cannot work well for low resolution and crowded images.
In 2016, Sajith Kecheril Sadanandan and co-workers found out that principal curvatures[7] for phase image of bacteria can be used for enhancing the edge of bacteria, which is a relatively easy method for cell segmentation process. Other than that, it also works in low resolution and crowded images. Combine this technique with other several common image processing techniques, image data of single particle cells or bacterium can be obtained in more efficiently way.