Cellular Image Segmentation Using Watershed Algorithm Biology Essay

Published:

Segmentation is one of the most important preprocessing steps towards pattern recognition and image understanding. It is a significant step towards image compression and coding. The cells analysis is an important study in the medical industry. In order to assists the cells analysis in the medical industry, this project was carry out a research towards the watershed transform and apply it in the synthetic fluorescence cell image segmentation. On the other way, this project is also concern on building an interface for users to perform the cellular image segmentation. At the end of this project, the watershed algorithm was applied in the target application. The original input cells' images were segmented into appropriate size base on the input images. As a conclusion, the morphology watershed transform is being studied and implemented in the proposed system.

Image segmentation is a fundamental step in many areas of computer vision including stereo vision and object recognition. It is the process of dividing images into regions according to its characteristic e.g., colors and objects present in the images. These regions are sets of pixels and have some meaningful information about objects. The result of image segmentation is in the form of images that are more meaningful, easier to understand and easier to analyze. In order to locate objects and boundaries in images feature extraction of object shape, optical density, and texture, surface visualization, image registration and compression segmentation is used. Correct segmented results are very useful for the analysis, prediction and diagnoses.

Lady using a tablet
Lady using a tablet

Professional

Essay Writers

Lady Using Tablet

Get your grade
or your money back

using our Essay Writing Service!

Essay Writing Service

Nowadays, there are many studies that concern on the cellular image segmentation in the image processing. The cellular image is the image of cells where it is captured from the electronic microscope. In the cells analysis, the cells are observed under the video microscope. For further used in the remains analysis (i.e. cells recognition) the cells images is then captured out from the microscope. In order to approach the cells recognition in the cellular image, an algorithm for performing the segmentation is required and become the important key.

Hence, a suitable algorithm for the cellular image segmentation, Watershed Algorithm is implemented in this project. The watershed algorithm is the method for the image segmentation in the field of mathematical morphology. Because of the natural shapes and the characteristic of the cells, the use of watershed algorithm is more suitable than others algorithm in order to produce an accurate segmentation in the cellular image. Hence, in this proposed system the watershed algorithm is being implemented accurately in order to produce an accurate image segmentation system for the cellular image. By the end of this project, the target system can be assists in the cellular image analysis, i.e. the cell counting for the cell contained inside the cell image.

TERMINOLOGY

The Cell Image Analysis

Cell image analysis in microscopy is the core activity of cytology and cytopathology for assessing cell physiological (cellular structure and function) and pathological properties [1]. Biologists usually make evaluations by visually and quantitatively inspecting microscopic images: this way, they are particularly able to recognize deviations from normality. Nevertheless, automated, analysis is strongly preferable for obtaining objective, quantitative, detailed and reproducible measurements, i.e., features of cells. [1]

Automated analysis of medical cell images has been gaining more importance in pharmacology and toxicology practice [2]. Extraction of accurate quantitative data about the cell morphology is a critical task for biologists. An automated procedure for analysis of cell images is highly desirable since there may exist a hundreds of images for each patient. And the analysis by hand is very time-consuming and tedious. In such an automated analysis system the most critical step is the correct segmentation of cell bodies which are then used to obtain the quantitative data on each cell image. Hence, the cell image segmentation is an important base for retrieving the quantitative data in a huge quantity of cell image and it making the data retrieving on the cell image become more efficient and accurately.

B. Image Segmentation

Image segmentation is the pre-processing step in the image processing and it is consider as an important base before the image analysis. Generally, image segmentation is subdividing the image into constitute regions or objects. The level to which the subdivision is carried depends on the problem being solved. That is, segmentation should stop when the objects of interests in an application have been isolated. Furthermore, the result of image segmentation is a set of segments that collectively cover the entire image, or a set of contours extracted from the image. Each of the regions is similar with respect to some characteristic or computed property, such as color, intensity, or texture.

Lady using a tablet
Lady using a tablet

Comprehensive

Writing Services

Lady Using Tablet

Plagiarism-free
Always on Time

Marked to Standard

Order Now

The usage of image segmentation is widely use in the medical imaging. For instance locate tumors and other pathologies, measures tissue volumes, computer-guided surgery, diagnosis, treatment planning and the study of anatomical structure in the medical image. Beside of the medical imaging, the use of image segmentation is also cover the face recognition, fingerprint recognition, traffic control systems, brake light detection, machine vision, and also locate objects in satellite image.

In this project, the image segmentation is use to counting the cells contained inside the cell image. Besides of the cell counting, the cells contained inside the cell image are also being segmented into appropriate size.

C. The Watershed Transform

The watershed transform is the method of choice for image segmentation in the field of mathematical morphology. In grey scale mathematical morphology the watershed transform, originally proposed by Digabel and Lantuejoul and later improved by Beucher and Lantuejour in late of 70's as a tool for segmenting grayscale images [3].

The watershed transform can be classified as a region- based segmentation approach. The intuitive idea underlying this method comes from geography: it is that of landscape or topographic relief which is flood by water, watershed being the divide lines of the domains of attraction of rain failing over the region (as shown in Figure 1 below). An alternative approach is to imagine the landscape being immersed in a lake, with holes pierced in local minima. Basins, it also called 'catchment basins', till fill up with water starting at these local minima, and, at point where water coming from different basins would meet, dams are built. When the water level has reached the highest peak in the landscape, the process is stopped. As a result, the landscape is partitioned into regions or basins separated by dams, called watershed lines or simply watersheds.

Figure 1, the diagram of watershed transform.

SYSTEM DESIGN

The work flow of entire system

The watershed transform is being used in the target research and application for the cell image segmentation. Figure 2 shows that the workflow of the entire system.

Figure 2, the work flow of the entire system.

Once the cell image is loaded into the system, the first processing step is the noise filtering towards the original image. Then, the cell image is transformed into grey scale image. Because the watershed algorithm can only perform on the grey scale image so we need to transform the image into grey scale image before we perform the watershed segmentation.

In order to enhance the edge of each cells contained inside the cell image, the homogeneity edge detection is applied in the processed image. Lastly, the edge detected image is then passes to perform the watershed segmentation. In the end of this segmentation process, the segmented image and the total amount of cell contained inside the cell image is produced. The results were discussed in the Section IV.

EXPERIMENT RESULT

Tested Image Sets

The accuracy of the system is tested by using the benchmark image from the Broad Bioimage Benchmark Collection by Broad Institute [4]-[5]. The Broad Bioimage Benchmark Collection (BBBC) is a collection of freely downloadable microscopy image sets. Besides, the BBBC is organized by the Broad Institute's Imaging Platform.

There are two sets of synthetic cell image being tested in this project. Each of the tested set contains 20 images. Both of the tested set is the synthetic images generated with the SIMCEP simulating platform for fluorescent cell population images. The first set is the synthetic fluorescent cell population images with probability clustering of 0.0. The second set is the synthetic fluorescent cell population images with probability clustering value of 0.15.

Result

Figure 3(a): Original image.

Figure 3(b): After apply noise filtering.

Figure 3(c): Segmented image with the result of 303 cells counted.

The accuracy on the cell counting is calculated based on the benchmark image. The color of these original images is in the grey scale. Hence, in the flow of the system, the process for transform the image into grey scale is skipped for these tested images.

Each of the tested benchmark image consists of 300 objects, and there are 20 images for this synthetic fluorescent cell population images with probability value 0.0. One of the result get based on the segmentation on the tested images is shown in Figure 3(a)-(c). The system was over segmented the cell and with the output value of 303 cells.

Lady using a tablet
Lady using a tablet

This Essay is

a Student's Work

Lady Using Tablet

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Examples of our work

Figure 3(d) shows the 20 set of synthetic fluorescence cell population images and the cell counted result generated in this system. The figure 3(e) represented the mean and standard deviation of the result. The mean and the standard deviation of the result are calculated by using the SSPS Data Editor. The mean showing that the average cell counted is around 299 cells. Which means the accuracy is 95%-99% towards this synthetic fluorescence cell population images with clustering probability value of 0.0.

Figure 3(d), cell counted result of 20 set of synthetic fluorescence cell population images.

Figure 3(e), the mean and standard deviation of the result.

On the others way, the accuracy towards the synthetic fluorescent cell population images with probability clustering value of 0.15 is also tested in this system. The Figure 4(a)-(c) below shows the segmented results of the tested image. The overall of the segmentation on the synthetic fluorescent cell population images with probability of clustering 0.15 is less accuracy than the synthetic fluorescence cell population images with clustering probability value of 0.0 (Figure 3(a)-(c)). The Figure 4(d) shows the cell counted result of the 20 images for this synthetic fluorescent cell population images with probability clustering value of 0.15. Whereby, the Figure 4(e) shows the mean and standard deviation of the 20 set of generated data. The mean and standard deviation for these 20 images is 242 and 7.118 respectively and which are calculated by using the SPSS Data Editor. The accuracy for this segmentation towards the synthetic fluorescent cell population images with probability clustering value of 0.15 is around 70%-80%.

Figure 4(a): Original image.

Figure 4(b): After apply noise filtering.

Figure 4(c): Segmented image with the result of 236 cells counted.

Figure 4(d), cell counted result of 20 set of synthetic fluorescence cell population images with probability of clustering value 0.15.

Figure 4(e), the mean and standard deviation of the result.

CONCLUSION

As a conclusion, the cell image segmentation has been successfully achieved by using the watershed algorithm. The watershed algorithm applied with the assists of some preprocessing steps. Without using the preprocessing steps, the watershed algorithm will not get the fully accurate result in the image segmentation. The watershed algorithm only can apply on a grey scale image but not RGB image. Rather than this, the pixel format of a pictures act as a very important key in the image segmentation. As in the image segmentation, each of the pixels contained in an image is taking considered and the segmentation on the image is based on the pixels or region scanned inside an image. The classification of pixels format is the core procedure to be done in the image segmentation. Without a specific algorithm, the image segmentation cannot be easily approach.

On the others way, this system might be improved in the future for further use in the cell image analysis. As a result of this project, the cell image segmentation has been successfully achieved. For the future work, this system might be improved in term of the accuracy towards the cell segmentation on the clustering nuclei cell images. Besides, the system can be extended to become a system to be used in the cell recognition. Since, the cell image segmentation is the preprocessing steps towards the pattern recognition in cell image analysis so this project has been achieved the based for the pattern recognition on the cell image analysis. Last but not least, the watershed transform has been studied and applied successfully in this project.