Image processing and soft-computing processes
Presented here is a novel methodology to automatically measure a number of brain cancer cells using optimized image processing and soft-computing for classification. The former approach is used to prepare the cell image from the medical laboratory, such as background removal, image adjustment, and cell detection including noise reduction. Then, Gabor filter is applied to retrieve the key features before feeding into different soft-computing techniques to identify the actual cells. The results show that the performance of Fuzzy C-Mean with image processing optimization is outstanding against the others (neural network, genetic algorithm, and support vector machine), i.e., 95% vs. less than 90% in precision in addition to the superior computational time, around 2 seconds.
Image processing and soft-computing processes
The process starts with a original images applying image process and soft-computing based classification into cell detection. Image processing is one of the computational processes to automatically recognize or interpret the image given its characteristic, such as size, shape, and direction, one of these is to apply into the area of cell recognition as one of the pattern recognition techniques. In general, as discussed in the research by Al-tarawneh, M. S. [1], there are four main methods for image processing .Soft computing techniques are generally used to resolve the problem which involves with an uncertainty and non-linear solver [2]. There are several classes of soft computing, e.g., NN, Fuzzy Logic (FL), SVM, and Evolutionary Computation (EC) such as GA [3], all of which are used as our comparative study then the output is result of counting a number of brain cancer cell in image.
The brain cancer cell images were provided from the department of Physiology, Faculty of Medicine, Khon Kaen University.
Click here to view/download the brain cancer cell images data.
Please send the email to chakso@kku. ac. th for downloading!
This paper investigates the possibility to enhance the recognition rate of brain tumor cell images acquired from the medical laboratory. A simplified image processing is first applied to the tumor cell as pre-processing, such as Otsu and unsharp masking methods; then, Histogram Orientation Gradient is our selection of feature extraction based on tumor shape characteristics which is then integrated into Extreme Learning Machine (ELM) as cell classification, called H-ELM. Its precision performance is confirmed against Support Vector Machine and a traditional ELM, i.e., 90% against 64% and 70%, respectively. To further improve H-ELM in aspects of computational complexity with high dimension and large image datasets, the feasibility to utilize the parallelism is investigated and implemented using Compute Unified Device Architecture in Graphics Processing Unit resulting into 3 and 7 times speedup over its non-parallel scheme (CPU) and the traditional ELM, called Parallel H-ELM.
Enhanced image processing and classification processes over GPU
This research investigates an alternate approach for brain tumor cell recognition using a simplified image processing as pre-processing. Then, Histogram Orientation Gradients (HOG) [1] is applied to extract key features for further classification steps using Extreme Learning Machine (ELM) [2] to achieve a high recognition rate, called H-ELM. To enhance H-ELM in a dimension of computational time, this research discusses the possibility to utilize the parallelism implemented on Graphics Processing Unit resulting (GPU) [3] using Compute Unified Device Architecture (CUDA) frameworks, and all these are called Parallel H-ELM (PH-ELM).
The brain cancer cell images were provided from the department of Physiology, Faculty of Medicine, Khon Kaen University.
Click here to view/download the brain cancer cell images data.
Please send the email to chakso@kku. ac. th for downloading!