Medical image compression using IIR

wavelet filter banks

Cite as: AIP Conference Proceedings 2591, 030040 (2023); https://doi.org/10.1063/5.0119538
Published Online: 29 March 2023

Rasha Waleed

AIP Conference Proceedings 2591, 030040 (2023); https://doi.org/10.1063/5.0119538 2591, 030040

© 2023 Author(s).
Medical Image Compression Using IIR Wavelet Filter Banks
Rasha Waleed
Technical Computer Engineering Depatment, Al-Hadba University College, Mosul, Iraq.

rashawaleedhamad@gmail.com

Abstract. The proper utilization of digital medical imaging technology and applications requires image compression. The
proposal in this paper is based on the IIR wavelet filter bank, and it suggests an image compression algorithm. The
performance of the scheme has been measured using prescribed performance measures. The results demonstrate that the
proposed method performs preferable compression than other compression methods. The results obtained are compared
with wavelet-based filters bank in order to improve the compression using our approach. When compared to standard
approaches, the results are quite excellent in terms of compression ratio and compressed image quality.

INTRODUCTION
Digital medical imaging is an important and well-established part of the full chain of patient dealing in all modern
hospitals. Medical imaging styles include: magnetic resonance imaging, computerized tomography, X radiographs ,
and ultrasonography, etc. These styles allow you to view anatomical cross sections and physiological states in a variety
of ways. The medicinal images have large storage requirement. Due to network bandwidth and storage space
constraints, the images must be compressed before being sent and saved. As a result, image compression techniques
are required to save storage and transmission space and time [1][2]. Lossless compression and lossy compression are
the two types of medical image compression techniques [3].

LITERATURE REVIEW

In 2012 M. M. H. Chowdhury and A. Khatun proposed in [4] an image compression technique based on DWT.
This technique achieves high compression ratios while avoiding image quality degradation. Using a set of genuine
images, the effectiveness and resilience of this approach has been demonstrated. The suggested compression technique
was compared to other commonly used compression strategies. When compared to alternative compression strategies,
the proposed methodology performs better.
In 2016 R. Sharma,et al. Discussed in [5] a lossless compression model with a high compression rate. The medical
photos are accurately coded thanks to the hybrid technicality.
In 2017 F. Liu,et al. The current state of image compression standards in medical imaging applications, as well as
the legal and regulatory issues surrounding compression in medical settings, were discussed in [6].
Based on IIR wavelet filter banks, this paper proposed a new algorithm for medical image compression that could
provide the best compression performance.

IIR WAVELET FILTER BANKS BASED COMPRESSION

DWT is an efficient tool for image processing applications. DWT that implemented efficiently with a filter banks,
which can be accomplished using IIR or FIR filters [7]. The wave digital filter is one of the IIR digital filters' structures.
[8]. A Bireciprocal Lattice Wave digital filters (BLWDFs) are special form of lattice wave digital filters(LWDFs) [9].
Wavelet transform implementations and wavelet bases are obtained from orthonormal perfect reconstruction
quadrature mirror filter bank. Figure 1 shows the structure of a 2-channel QMF bank. [10].

Al-Kadhum 2nd International Conference on Modern Applications of Information and Communication Technology
AIP Conf. Proc. 2591, 030040-1–030040-7; https://doi.org/10.1063/5.0119538
Published by AIP Publishing. 978-0-7354-4324-2/$30.00

030040-1
 FIGURE 1. Structure of a 2-channel Quadrature Mirror Filter bank [10].

An efficient Quadrature Mirror Filter bank structure is achieved by an all-pass based polyphase network
implementation according to Fig. 2.

FIGURE 2. Efficient structure of the wavelet filter bank with IIR filters [9].

As shown in Fig. 2, the transfer function of a low-pass BLWDF is:

1
𝐻𝑜 (𝑧) = 2 [𝐴𝑜 (𝑧) + 𝑧 −1 𝐴1 (𝑧)] (1)

where 𝐴𝑜 (𝑧) and 𝐴1 (𝑧) are all-pass sections and can be expressed, as: [9]

(𝑁+1)/2 𝛼𝑖 +𝑧 −1
𝐴0 (𝑧) = ∏𝑖=2,4,.. 1+𝛼𝑖 𝑧 −1
(2)

(𝑁+1)/2 𝛼𝑖 +𝑧 −1
𝐴1 (𝑧) = ∏𝑖=3,5,.. 1+𝛼𝑖 𝑧 −1
(3)

Where αi is the multiplier coefficient's value in the ith all-pass function and N is the 𝐻𝑜 (𝑧) filter order, which are
selected as in refrences [11] [9] [12] for getting (7th ,9th ,11th ) IIR Wavelet Filter Banks.

IMAGE QUALITY MEASUREMENT

Throughout this paper numbers are given for four measures of compression performance. These measures are:

030040-2
 Compression Ratio (CR)
The compression ratio (CR) is the proportion of the original image size to the compressed image size [13].

𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙 𝑖𝑚𝑎𝑔𝑒 𝑠𝑖𝑧𝑒

𝐶𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑜𝑛 𝑟𝑎𝑡𝑖𝑜 = (4)
𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑒𝑑 𝑖𝑚𝑎𝑔𝑒 𝑠𝑖𝑧𝑒

SNR and PSNR

The basic methods for quantitatively comparing a reconstructed image with the original are SNR and PSNR.
SNR is measured in decibels (dBs).

1 𝑅
√ ∑ ∑𝐶 [𝑦(𝑟,𝑐)]2
𝑅𝐶 𝑟=1 𝑐=1
𝑆𝑁𝑅 = 20 𝑙𝑜𝑔10 (5)
√𝑀𝑆𝐸

MSE: mean square error is computed as:

1
𝑀𝑆𝐸 = 𝑅𝐶 ∑𝑅𝑟=1 ∑𝐶𝑐=1|𝑦(𝑟, 𝑐) − 𝑦̃(𝑟, 𝑐)|2 (6)

 𝑦(𝑟, 𝑐) is the original image.

 𝑦̃(𝑟, 𝑐) is the reconstructed image.

and PSNR is calculated in dBs by the following equation:

[𝑆−1]2
𝑃𝑆𝑁𝑅 = 10 𝑙𝑜𝑔10 ( ) (7)
𝑀𝑆𝐸

Where S =2υ, υ is number of bits/pixel (S=256 for images with resolution of 8 bits/pixel) [14].

Correlation Factor (Cor)

Correlation factor (Cor) is a metric that determines how similar two images as [15]:

∑𝑅 𝐶 ̅) (𝑦(𝑟,𝑐)− 𝑦̅̃ )
𝑟=1 ∑𝑐=1(𝑦(𝑟,𝑐)−𝑦
𝐶𝑜𝑟 = (8)
√[∑𝑅 𝐶 ̅) 2 ][∑𝑅
𝑟=1 ∑𝑐=1(𝑦(𝑟,𝑐)−𝑦
𝐶 ̅ 2
𝑟=1 ∑𝑐=1(𝑦̃(𝑟,𝑐)− 𝑦̃ ) ]

Where:

𝑦̅ is the mean of 𝑦(𝑟, 𝑐), that is,

1
𝑦̅ = 𝑅×𝐶 ∑𝑅𝑟=1 ∑𝐶𝑐=1 𝑦(𝑟, 𝑐) (9)

𝑦̅̃ is the mean of 𝑦̃(𝑟, 𝑐) , that is,

𝑅 𝐶
1
̅𝑦 =
̃ ∑ ∑ 𝑦̃(𝑟, 𝑐) (10)
𝑅×𝐶
𝑟=1 𝑐=1

SIMULATION RESULTS
The presented approaches are simulated in Matlab. A set of MRI images are selected as test data. All are gray level
images shown in figure 3.The Decomposition of MRI images using three IIR wavelet filter bank structures are shown

030040-3
in figure 4. Table 1 gives the performance of designed IIR wavelet filter bank on the input images. The performance
of these wavelet filter banks was analyzed and plotted in figure 5.

FIGURE 3. Input MRI images.

(a)

030040-4
 (b)

(c)

FIGURE 4. Decomposition of MRI images using (a) 11th order Wavelet Filter Bank
(b) 9th order Wavelet Filter Bank (c) 7th order Wavelet Filter Bank.

030040-5
 TABLE 1. Performance of designed IIR wavelet filter banks on medical images.

Image Parameter 7th IIR Filter Banks 9th IIR Filter Banks 11th IIR Filter Banks
CR 17.8029 18.045 22.679
MRI1 PSNR(db) 62.1761 49.457 45.8379
SNR(db) 54.3650 41.6458 38.0258
Corr. 1 0.9999 0.9999
MRI2 CR 39.1376 39.1376 39.1376
PSNR(db) 74.9069 62.1476 58.5278
SNR(db) 66.6928 53.9335 50.3137
Corr. 1 1 1
MRI3 CR 48.7029 46.4891 56.3727
PSNR(db) 65.2457 63.8763 47.8546
SNR(db) 58.5235 57.1541 41.1323
Corr. 1 1 0.9998
MRI4 CR 29.5721 22.247 33.3936
PSNR(db) 65.3794 54.7146 51.14
SNR(db) 56.1996 45.5377 41.961
Corr. 1 1 0.9999
MRI5 CR 28.25 28.25 28.25
PSNR(db) 56.9628 44.8623 39.8046
SNR(db) 48.7074 36.6069 31.5492
Corr. 0.9999 0.9992 0.9973
MRI6 CR 58.0918 53.0197 69.006
PSNR(db) 63.3656 50.6466 47.0266
SNR(db) 53.8963 41.1773 37.5573
Corr. 1 0.9999 0.9998

80

70

60

50

40 7th IIR
Filter Banks
30
9th IIR
20
Filter Banks
10

0
PSNR

PSNR

PSNR

PSNR

PSNR

PSNR
SNR
SNR

SNR

SNR

SNR

SNR
CR
CR

CR

CR

CR

CR
Corr.
Corr.

Corr.

Corr.

Corr.

Corr.

FIGURE 5. Performance of designed IIR wavelet filter banks on medical images.

030040-6
 CONCLUSIONS
A medical image compression method is proposed that preserves the original image's quality. It is based on IIR
wavelet filter bank. Many image quality measurements, such as the SNR and PSNR, have been used to show the
efficiency and performance of the proposed design. The design's simplicity makes real-time implementation easy.

REFERENCES

1. T. Bruylants, A. Munteanu and P. Schelkens," Wavelet based volumetric medical image compression", Signal
Processing: Image Communication Vol. 31,pp. 112–133,2015.
2. V. K. Sudha and R. Sudhakar, "Two Dimensional Medical Image Compression Techniques-A Survey", ICGST-
GVIP Journal, Vol.11, Issue 1, March 2011.
3. H. Jiang, Z. Ma, Yang Hu, B. Yang, and L. Zhang, " Medical Image Compression Based on Vector Quantization
with Variable Block Sizes in Wavelet Domain", Computational Intelligence and Neuroscience, Article ID
541890, 8 pages,2012, On http://dx.doi.org/10.1155/2012/541890
4. M. M. H. Chowdhury and A. Khatun , "Image Compression Using Discrete Wavelet Transform", IJCSI
International Journal of Computer Science Issues, Vol. 9, Issue 4, No 1, July 2012.
5. R. Sharma, C Kamargaonkar and M. Sharma, " hybrid Medical Image Compression: Survey", International
Journal of Advanced Research in Computer Engineering & Technology (IJARCET) Vol. 5, Issue 4, April 2016.
6. F. Liu , M. H. Cabronero, V. Sanchez, M. W. Marcellin and Ali Bilgin, "The Current Role of Image Compression
Standards in Medical Imaging", www.mdpi.com/journal/information/ Vol.8, No.131,2017.
7. M. Estes, "The Discrete Wavelet Transform". ECE 402–Digital Signal Processing, Spring 2001.
8. W. Sootkaneung, “The Design of Bit-Serial Lattice Wave Digital Filter Using FPGA”, IEEE Trans. On
communication and signal processing, No.5, pp. 559 - 563, Bnakok, 2005.
9. J. M. Abdul-Jabbar and R. W. Hamad, “Design and Multiplierless Implementations of 9th order linear-Phase
Bireciprocal Lattice Wave Digital Wavelet Filter Banks ”, Al-Rafidain Engineering Journal, Vol.21, No. 4, pp.
68 – 86, August 2013.
10. H. W. L ̈llmann and P. Vary, "Design of IIR QMF Banks with Near-Perfect Reconstruction And Low
Complexity", Proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing
(ICASSP), (Las Vegas, NV, USA), pp. 3521–3524, ISBN 978-1-42441-483-3, March 2008.
11. J. M. Abdul-Jabbar, and R. W. Hamad, “Efficient IIR Wavelet Filter Banks with Approximate Linear-phase in
Pass-band”, 2011 IEEE Jordan Conference on Applied Electrical Engineering and Computing Technologies
(AEECT 2011), Dec 6-8, 2011,Amman, pp. 227 – 232.
12. R. W. Hamad, "Design and FPGA implementation of 11thorder Efficient IIR Wavelet Filter Banks with
Approximate Linear-phase". Academic Journal of Nawroz University (AJNU) Volume 7, No 4, pp.207-212,
2018.
13. M. Dridi, B. Bouallegue, M. A. Hajjaji and A. Mtibaa,,, An Enhencment Medical Image Compression Algorithm
Based on Neural Network’’, (IJACSA) International Journal of Advanced Computer Science and
Applications,Vol. 7, No. 5, pp.484-489, 2016.
14. R. Fakeh and A. A. Abd Ghani , “Empirical Evaluation of Decomposition Strategy for Wavelet Video
Compression”, International Journal of Image Processing, Vol. 3, Issue 1, pp. 31-54, 2009.
15. H. H. Abdul zahrah, ‘‘Encryption Using Wavelet coded Image Data’’, M.Sc. Thesis in Computer Engineering,
Collage of Engineering, University of Basrah, Basrah, Iraq, 2004.

030040-7