SPATIAL FILTERING TECHNIQUE FOR MOTION CORRECTION OF CERVICAL-SPINE MAGNETIC RESONANCE IMAGES
A thesis submitted in partial fulfillment of the requirements for the degree of Master of Biotechnology (Biomedical Engineering)
Prepared by:
ERICA ROSELLA 09/291334/PMU/06091
DEPARTMENT OF NATIONAL EDUCATION GRADUATE SCHOOL OF UNIVERSITAS GADJAH MADA YOGYAKARTA 2012
Tesis
SPATIAL FILTERING TECHNIQUE FOR MOTION CORRECTION OF CERVICAL-SPINE MAGNETIC RESONANCE IMAGES
dipersiapkan dan disusun oleh: ERICA ROSELLA telah dipertahankan di depan dewan penguji pada tanggal : 2012
Susunan Dewan Penguji
Pembimbing Utama
Anggota Dewan Penguji Lain
Dr. Gede Bayu Suparta ......................................................... Pembimbing Pendamping I
Prof. Dr. Adhi Susanto, M.Sc. ......................................................... Penguji I
dr. Rahadian Magetsari, Sp. OT., Ph. D. ......................................................... Pembimbing Pendamping II
........................................................... Penguji II
Tesis ini telah diterima sebagai salah satu persyaratan untuk memperoleh gelar Master Tanggal : ...........................................
Prof. drg. Widowati Siswomiharjo, MS., Ph. D. Pengelola Program Studi: Rekayasa Biomedis
Statement of Originality
I hereby certify that the content of this Thesis is truly and genuinely my own work. It does not resemble or contain another work which ever proposed for obtaining a degree in any university, and as long as I observed, there is no ideas or thought in this content which has been written or published by other person, except those which is written and referred on this script and listed within the reference. In a case that my statement is proven false, I, herewith, will sincerely accept any academic consequences from the authority of the university.
Yogyakarta, …………………2012
Erica Rosella
I dedicate this Thesis for: My Father and Mother, Tjioe Sui Laij and Kwee Hiang Nio My kind Brother, Helmi Henara and Rendra Gry Hena My Guardian, Nixi Tirayoh and Arifa Khairiyanti
Thank you: For all the love, sacrifice, encouragement and best prayers
Acknowledgement
Praise be to The God to Whom we seek help and guidance, in which under His benevolence we exist and with His mercy this project and report are has been accomplished. This thesis, entitled “Spatial Filtering Technique for Motion Correction of Cervical-Spine Magnetic Resonance Images”, was proposed to fulfill one of the requirements for obtaining a master degree in Biomedical Engineering Program, Graduate School, Universitas Gadjah Mada. The work outlined in this thesis was carried out in Panti Rapih Hospital Yogyakarta, Physics of Imaging Laboratory-Departement of Physics Faculty of Mathematics and Natural Sciences, Universitas Gadjah Mada, over the period from March to July 2011. Many great people have been involved in the work leading to my master thesis. For their contributions to this study, I wish to express my appreciation. I would like to express my sincere and indebted gratitude to the Director of Post Graduate School of Gadah Mada University for the opportunity, facilities, encouragement and financial support during my study. I express my indebtedness and gratefulness to Prof. drg. Widowati Siswomihardjo, MS., as the chair of Biomedical Engineering Department for her continuous encouragement and guidance. I am obliged to her for her care, constructive critism, moral support and her valuable time. I would also send my sincere gratitude for financial support I have received through „beasiswa SPS‟ through all the stages during my Master study. I would like to express my gratitude and appreciation to my supervisors, Dr. Gede Bayu Suparta, as the first counselor, I greatly appreciate the inspiration, his guidance of my research and for introducing me to the world of Imaging, also for the constructive feedback and encouragement,
and dr. Rahardian Magetsari, Sp. OT., as the second counselor, I am thankful for inspiring supervision of my work, advice, time and efforts. I would also like to express my sincere gratitude to Prof. Adhi Susanto, from Department of Electronics Engineering for sharing his generous help, valuable time, suggestions and also his laboratory facilities, which greatly aided my work. I am also thankful to dr. Simeon Budi Mulyo, Sp. Rad. and Y. Purnomo Sigit , B. Sc., for their help, sharing and the insightful comments given throughout the time of completing the data collection in RS Panti Rapih Yogyakarta. I would like to thank all my colleagues and friends Suryono, M.Si, Sri Ratna Sulistiyanti, MT., Adrianus Inu Natalisanto, M.Si, Catur Edi Widodo, M.Si, I Wayan Sutrisna, ST, Krisnawati, S.Si, Agatha Mahardika, S.Si., Ayu Fitri Amalia, M. Sc., Sri Lestari, S.Si and I Gede Arya Wiguna, S. Si. for their company and cooperation and sharing during this period. I also owe my gratitude to Pak Soegi for his enthusiastic help in administrative matters. Not forgetting to all my fellow postgraduate students of Biomedical Engineering, class of '09 and friends for the lovely friendship and moral support that are given throughout the entire masters programme. Lastly, I am thankful to all those who have supported me directly or indirectly during the my master study and this research work. There were still many weaknesses in this report. Therefore, I will be grateful for any critics and constructive feedback to complete this thesis. Hopefully this simple report useful for extend of science and knowledge. Yogyakarta, February 2012 Author, Erica Rosella
INTISARI
TEKNIK PENAPISAN SPASIAL UNTUK KOREKSI GERAK CITRA SPINAL SERVIKS MAGNETIC RESONANCE IMAGING Oleh ERICA ROSELLA 09/291334/PMU/06091
Telah dilakukan penelitian metode penapisan spasial untuk memprediksi dan merestorasi citra MRI yang terdegradasi oleh pengaburan gerak. Beberapa kasus merujuk pada kegagalan pencitraan MRI selama proses pemindaian spinal serviks untuk pasien yang mengalami masalah struktur anatomis sendi. Pada setiap kasus yang memunculkan pergerakan, sehingga menghasilkan citra yang cacat, penyelesaiannya adalah dengan pengambilan ulang gambar yang lebih baik dan pengulangan prosedur. Proses restorasi citra bertujuan untuk menanggulangi pemindaian yang tak efisien. Pengaburan gerak pada citra MRI diamati berdasar pada koefisien Point Spread Function (PSF) yang dihasilkan oleh pergerakan objek selama pemindaian. Performa metode telah dievaluasi untuk ukuran blur, orientasi blur, dan ukuran jendela dari kernel. Pengujian dilakukan untuk dua kategori citra, yaitu terhadap sekumpulan data citra MRI asli dan terhadap citra sintetis yang dibuat menggunakan aplikasi perangkat lunak seperti GIMP dan MATLAB. Metode prediksi dan restorasi dilakukan dengan konvolusi toleran terhadap error pada PSF. Ketajaman citra dan detail tepi objek secara signifikan dapat terlihat secara visual. Hal demikian juga ditunjukkan oleh hasil analisa histogram dan profil garis. Namun, berdasarkan evaluasi kualitas menggunakan analisis koefisien korelasi ternormalisasi, metode yang diusulkan kurang menunjukkan hasil koreksi signifikan dan memuaskan.
Kata KuncI: citra MRI, restorasi citra, orientasi gerak, koreksi gerak, konvolusi, koefisien korelasi
ABSTRACT
SPATIAL FILTERING TECHNIQUE FOR MOTION CORRECTION OF CERVICAL-SPINE MAGNETIC RESONANCE IMAGES By ERICA ROSELLA 09/291334/PMU/06091 The research of spatial filtering method for predicting and restoring MRI images which was degraded by motion blur has been done. Some cases referred to the failure of MRI imaging during cervical-spine scanning process for patient with anatomical structure problem of joints. In each case which movement occured, and that resulting degraded image, the existing solution were retaking the better image and repeating procedure. Image restoration process intended to overcome that inefficient scanning. Motion blurring in MR-images were observed based on Point Spread Function (PSF) coefficients produced by object movement during scanning. Performance of the method has been evaluated regarding to the blur size, blur orientation and window size of the kernel. The experiment has been done for two categories of image, such as original data set from MRI and synthetic image generated with software application such as GIMP and MATLAB. Both prediction and restoration methods performed with convolution appear to be reasonable tolerant to the error in the PSF. The sharpness and detail of object boundary can be seen visually quite significant. Thus, results with line profile and histogram analysis are quite similiar. However, regarding to evaluation of the quality using normalized correlation coefficients, the proposed method provide less significant and unsatisfying correction.
Keywords: MR images, image restoration, orientation of motion, motion correction, convolution, and correlation coefficient
