A unified framework for image compression and segmentation by using an incremental neural network
Introduction
Medical images like magnetic resonance (MR) and computer tomography (CT) acquired from different modalities comprise huge amounts of data, rendering them impracticable for storage and transmission. Archiving this large amount of image data in the computer memory is very difficult without any compression. An important issue in lossy compression of medical images is the risk of destroying diagnostically relevant information. Current lossy compression standards, such as JPEG (Wallace, 1991) and MPEG, are designed for conventional still-image and video display.
Transform-based techniques have been proposed for the efficient reduction of the high redundancy usually encountered in real life images (Strintzis, 1998). Unsupervised neural networks can perform nonlinear principal component analysis as a transform-based method in image compression (Tzovaras & Strintzis, 1998). They outperform linear principal component analysis, and are relatively easy to implement.
Another common method to compress images is to code them through vector quantization (VQ) techniques. Self-organizing Kohonen maps have been used to achieve the VQ process of image compression (Amerijckx, Verleysen, Thissen, & Legat, 1998). They represent an efficient compression scheme based on the fact that consecutive blocks in an image are often similar, and thus coded by similar codewords with a VQ algorithm. Early studies of lossy compressed medical images performed compression using variations on the standard discrete cosine transform coding algorithm combined with scalar quantization and lossless coding. More recent studies of efficient lossy image compression combined with scalar/vector quantization (Tsekouras, 2005, Pal et al., 2000), or neural networks (Jiang, 1999).
The constitution of the right feature space plays a critical role in decision making. In order to construct realistic classifiers, the features that are sufficiently representative of the physical process must be searched. In the literature, it is observed that different transforms are used to extract desired information from medical images. Image intensities at one or two neighborhood of the pixel (Dokur et al., 2000) and image intensities in multiple images (T1, T2 and proton density) (Qian, Cheng, & Liang, 1989) are utilized to represent the tissues in magnetic resonance and computer tomography images. Wavelet transform (Qian, Li, & Clarke, 1999), co-occurrence matrix (Arrowsmith et al., 1999, Haering and Lobo, 1999), Fourier transform (Feleppa et al., 1996) and spatial gray-level dependence matrices (Wu, Chen, & Hsieh, 1992) are used to extract textures in ultrasound images. Zhang, Yoshida, Nishikawa, and Doi (1998) used wavelet transform for the detection of the microcalcifications in digital mammograms.
Recent developments in spatial-scale (or frequency) analysis such as Gabor transform, Wigner distribution (Reed & Wechsler, 1990), and wavelet transform (Zhang et al., 1998) have provided a new set of multi-resolution analytical tools. The scales that represent the textures in different image types must be determined before the classification process. Different scales are used to extract information from ultrasound images and digital mammograms. The determination of the scales and also the mother wavelet is time consuming.
The main difficulty of the above methods is due to the lack of an adequate tool that characterizes different scales of textures effectively. In this study, the elements of the feature vectors are formed by the coefficients calculated by applying 2D-discrete cosine transform (2D-DCT) (Ahmed, Natarajan, & Rao, 1974) to image blocks of 8 × 8 pixels. Only two parameters are adjusted to represent the textures in the image: The amount of ignored coefficients, and the size of the image blocks. The dimension of the feature vectors can be decreased by discarding some coefficients which represent high frequencies. This leads to a decrease in the computational times of learning and classification processes.
In this paper, a unified framework for image compression and segmentation schemes based on discrete cosine transform and vector quantization by artificial neural networks is presented.
Section snippets
Method
In this study, 2D-DCT is preferred due to the following reasons: (i) only two parameters are adjusted to represent the textures in the image: the amount of ignored coefficients, and the size of the image blocks. (ii) The dimension of the feature vectors can be decreased by discarding some coefficients which represent high frequencies. This leads to a decrease in the computational times of processes. (iii) 2D-DCT coefficients which represent low frequencies give a coarse representation of image
Artificial neural networks
There are four reasons to use an artificial neural network (ANN): (i) Weights representing the solution are found by iteratively training, (ii) ANN has a simple structure for physical implementation, (iii) ANN can easily map complex distributions, and (iv) generalization property of the ANN produces appropriate results for the input vectors that are not present in the training set.
In this study, two neural networks are comparatively examined for the compression and segmentation of medical
Computer simulations
In this study, Kohonen map, ISOM, and JPEG standard are comparatively examined for medical image compression. MR, CT and ultrasound images are compressed by these three methods. The original medical images are shown in Fig. 4a, Fig. 4b, Fig. 4c. The size of the MR and CT head images are 256 × 256, and the size of the ultrasound image of kidney cyst is 437 × 420. In the study, CT, MR and ultrasound images are segmented into four, five and five categories respectively. All simulations are performed
Conclusions
In this study, two neural networks with unsupervised learning; the ISOM and the Kohonen network are compared for compression and segmentation of tissues in medical images. Kohonen network is a non-incremental unsupervised neural network. The topology of the Kohonen network has to be predefined. In the study, ten different trials for the same training set were done to choose the appropriate network structure and to define various parameters. ISOM is an incremental unsupervised neural network.
References (18)
- et al.
Features and classification methods to locate deciduous trees in images
Computer Vision and Image Understanding
(1999) Image compression with neural networks – a survey
Signal Processing: Image Communication
(1999)An introduction to neural computing
Neural Networks
(1988)A fuzzy vector quantization approach to image compression
Applied Mathematics and Computation
(2005)- et al.
Discrete cosine transform
IEEE Transactions on Computers
(1974) - et al.
Image compression by self-organized Kohonen map
IEEE Transactions on Neural Networks
(1998) - Arrowsmith, M. J., Varley, M. R., Picton, P. D., & Heys, J. D. (1999). Hybrid neural network system for texture...
- Dokur, Z., & Ölmez, T. (2000). Classification of magnetic resonance images using a novel neural network. In Proceedings...
- et al.
Typing of prostate tissue by ultrasonic spectrum analysis
IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control
(1996)
Cited by (35)
Improved watershed transform for tumor segmentation: Application to mammogram image compression
2012, Expert Systems with ApplicationsCitation Excerpt :General segmentation is the process of partitioning the image into disjointed regions so as to the characteristics of each region are homogeneous (Harlick & Shapiro, 1985). A large variety of image segmentation methods have been presented (Dokur, 2008; Dokur & Ölmez, 2008; Hsu, 2010; Hsu, Poon, & Sun, 2008; Lai & Chang, 2009; Lin, Tsai, Hung, & Shih, 2006; Pal & Pal, 1993; Sahba, Tizhoosh, & Salama, 2008). Among these methods, the watershed (Beucher & Meyer, 1993) that is a region-based approach is a traditional but popular method from mathematical morphology.
Medical image analysis with artificial neural networks
2010, Computerized Medical Imaging and GraphicsCitation Excerpt :In the algorithm, a linear convolution neural network was employed to seek a wavelet that minimises errors and maximises compression efficiency for an image or a defined image pattern such as microcalcifications in mammograms and bone in computed tomography (CT) head images. In Dokur [75], ANN was applied to medical images like magnetic resonance (MR), computer tomography (CT) head images and ultrasound imaging for compression and decision-making, where Kohonen map and incremental self-organising map (ISOM) were employed. In the proposed method, the image was first decomposed into blocks of 8 × 8 pixels, from which 2D discrete cosine transform (DCT) coefficients were computed.
A Moment Based Fractional Fourier Transform Scheme for MR Image Classification
2021, Automatic Control and Computer SciencesA comprehensive review on brain tumor segmentation and classification of MRI images
2021, Multimedia Tools and ApplicationsDNN-Based Brain MRI Classification Using Fuzzy Clustering and Autoencoder Features
2021, International Journal of Fuzzy Logic and Intelligent Systems