AN EFFICIENT TRAPEZOIDAL COMPRESSION ALGORITHM USING WAVELET TRANSFORMATION FOR MEDICAL IMAGE

Image compression processes moderate the number of bits essential to signify an image, which can improve the performance of systems during storage and transmission without compromising image quality. They are classified into lossy compression and lossless compression. In this work, a new algorithm called trapezoidal algorithm has been proposed for image compression. The encoding part of proposed algorithm is like trapezoid shape so it is named as trapezoidal algorithm. Two transformation techniques DWT, IWT with three wavelets such as Haar, Sym4 and Coif1 have been combined for image compression to confer good characteristics of these methods. In this each pixel coordinates are encoded using the logic from SPECK algorithm. In SPECK, the set S is encoded when the pixel level is reached whereas in proposed algorithm (trapezoidal) the set S is encoded when the size of set S is reached 4x4. The approximation of image is named as set S. set S can be formed into three subset termed as s1, s2, s3. S is grouped into many subsets each set can be defined based on a pattern of proposed algorithm. Magnetic Resonance Imaging (MRI) of brain and Computer Tomography (CT) of lung images are used for analyzing compression. The proposed algorithm gives high PSNR compared to existing algorithms EZW, SPIHT and SPECK. The performance metrics such as Peak signal-to-noise ratio (PSNR), Structural Similarity Index (SSIM), Mean square error (MSE), Bits Per Pixel (BPP), Compression Ratio (CR) and Compression Time (CT) are measured for lung and brain images. The dataset has been collected from various scan centers.


INTRODUCTION
Compression is the process of coding that will effectively reduce the total number of bits needed to represent certain information. The image compression techniques are classified into lossy compression and lossless compression. In lossy compression techniques reconstructed image contains loss of information. In lossless compression techniques reconstructed image contains no loss of information as exact reconstruction of the original image is possible. In this work, both lossless and lossy compression has been examined with medical image data such as lungs and brain image.
In this work, the compressions of two medical image modalities such as CT and MRI are analyzed. Proposed algorithm Trapezoidal is a wavelet based image compression technique. In this work three wavelets such as Haar, Sym4 and Coif1 are applied in DWT-Trapezoidal and IWT-Trapezoidal for compression of images. DWT is a mathematical tool for decomposing an image.
It does not change the information content present in the signal. IWT is used for lossless compression. Coefficients of this transform are represented by finite precision numbers and this allows for lossless coding.  Preeti V. Joshi and C.D.Rawat proposed a region based hybrid compression for medical images. In this brain image the ROI part is compressed by using arithmetic coding and NROI part is compressed by using SPIHT. The hybrid method is evaluated by Peak Signal to Noise Ratio (PSNR), Structural Similarity Index (SSIM) and Virtual Information Fidelity (VIF) for MRI brain images. PSNR for SPIHT is 23.90 and SSIM for SPIHT is 0.8398 VIF for SPIHT is 0.3455. [9] Surabhi N and Sreeleja N Unnithan proposed a review of various image compression algorithm EZW, DWT, SPIHT and also the performance metric (PSNR, MSE, CR) for these algorithm. The main objective of image compression is to decrease the redundancy of the image thereby increasing the capacity of storage and efficient transmission. [10] Chih-chien Chao and Robert M. Gray proposed Image Compression With A Vector Speck Algorithm. Vector SPECK shows a performance improvement over JPEG 2000 at the cost of added complexity. In this work, variation on SPECK using vector quantization to code the significant coefficients. Different VQ techniques including TSVQ and ECVQ are also proposed. An image block is divided in subblocks of equal size. This algorithm has very fast encoding and decoding which makes it very efficient in multimedia communication. [12] Gloria Menegaz proposed Trends in Medical Image Compression. This paper presents an overview on the state-of-the-art in the field of medical image coding. A versatile model-based coding scheme for three dimensional medical data is introduced. The potential of the proposed 5569 AN EFFICIENT TRAPEZOIDAL COMPRESSION ALGORITHM system is in the fact that it copes with many of the requirements characteristic of the medical imaging field without sacrificing compression efficiency. [13] Sudhakar Radhakrishnan and Jayaraman Subramaniam proposed Novel Image Compression Using Multiwavelets with SPECK Algorithm. This paper is to develop an efficient compression scheme and to obtain better quality and higher compression ratio using Multi-wavelet transform with Set Partitioned Embedded Block Coder algorithm (SPECK). [14] Shalini Prasad, PrashantAnkur Jain and Satendra Singh, "Lossless medical image the set S is encoded when the size of set S is reached 4x4. The approximation of image is named as set S. set S can be formed into three subset termed as s1, s2, s3. S is grouped into many subsets each set can be defined based on a pattern of proposed algorithm. Each pixel coordinates is encoded using the logic from SPECK algorithm. SPECK algorithm -If the size of the Set S is greater than 4x4 then it is partitioned into 4 sub sets by quad-tree decomposition (Figure 3). This . This partition is applied in proposed algorithm with three S sets recursively until the size of the sub set is equal to 4x4 and set is empty. (Figure 4). The performance of trapezoidal coder is compared with SPIHT, SPECK and EZW coder.  Next, if the set I is significant against the same threshold n, then the same partitioning is applied to set I recursively until the size of the sub set is equal to 4x4 and set is empty. This partition generates three S sets and one reduced I set ( Figure 6). This procedure is repeated until the set I is empty.

Results and Discussion
In this section, the results of DWT-Trapezoidal and IWT-Trapezoidal for medical image compression are discussed. Table 1 to Table4 shows the lossy performance of IWT compared with the DWT for a set of lung and brain images. Table 1 shows the PSNR value of DWT and IWT for CT lung images and MRI brain images using three wavelets such as Haar, Sym4 and Coif1. Table   2 shows the MSE value of DWT and IWT for CT lung images and MRI brain images using three wavelets such as Haar, Sym4 and Coif1. Table 3 shows the SSIM value of DWT and IWT for CT lung images and MRI brain images using three wavelets such as Haar, Sym4 and Coif1. Table 4 shows the CT (Compression Time) value of DWT and IWT for CT lung images and MRI brain images using three wavelets such as Haar, Sym4 and Coif1. The lossless compression results of IWT are obtained in terms of achieved bit rate. Table 5 shows the lossless compression PSNR value of IWT for lung and brain images using Haar, sym4 and COIF1wavelets.

COMPARATIVE ANALYSIS
In this section, comparison of existing algorithms such as EZW, SPIHT and SPECK are compared with proposed algorithm called Trapezoidal is discussed. Table 6 shows the performance metrics (PSNR, MSE, SSIM, CT) of Lung Image1 with BPP=2 and CR=4 for Haar wavelet. Table   7 shows the performance metrics (PSNR, MSE, SSIM, CT) of Lung Image1 with BPP=2 and CR=4 for Sym4 wavelet. Table 6 shows the performance metrics (PSNR, MSE, SSIM, CT) of Lung Image1 with BPP=2 and CR=4 for Coif1 wavelet. From this analysis, the proposed algorithm Trapezoidal is given high PSNR value using Haar Wavelet and low MSE value using Sym4 wavelet and achieve good SSIM and CT value of an image. From this discussion, the proposed algorithm is better than the existing algorithm.