Reliability of Three-dimensional Ultrasound Parameters and Their Correlation with the Progression of Adolescent Idiopathic Scoliosis by Quang N. Vo A thesis submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Biomedical Engineering Department of Biomedical Engineering University of Alberta © Quang N. Vo, 2016 Abstract Adolescent idiopathic scoliosis (AIS) is a three-dimensional (3D) spinal deformity with unknown causes and with prevalence of 1. If AIS is left untreated, it may progress, leading to back pain, cardiopulmonary problems, and psychosocial concerns, and eventually resulting in surgical intervention.
Four types of scoliosis treatment exist and the selection of management depends on the severity and the risk of progression. Currently, the Cobb angle is the gold standard to measure the severity of the spinal curvature on a two-dimensional (2D) postero-anterior (PA) radiograph. However, this 2D measurement may underestimate the true severity of scoliosis, which affects treatment decisions. To report the actual severity, the Cobb angle on the plane of maximum curvature (PMC) must be measured, requiring a 3D spinal image.
Although X-ray based imaging modalities such as computed tomography (CT) and multi-planar radiography provide good 3D images of the spine, the cumulative amount of ionizing radiation increases the risk of cancer. Therefore, 3D ultrasound was proposed in this PhD research as an alternative imaging method to measure spinal severity. In addition to the Cobb angle, the axial vertebral rotation (AVR) and the lateral deviation were also measured from the ultrasound images. To reconstruct 3D spinal images from 2D B-scans (B-mode images), a software was developed using the voxel-based reconstruction method with bi-linear interpolation.
This software could also measure the AVR, the Cobb angle and the lateral deviation on the PA plane and the PMC. In order to obtain an optimal reconstructed image, in-vitro and in-vivo experiments were performed to investigate the optimal ultrasound configurations that consisted of the ii ultrasound frequency, the minimum spacing between two adjacent B-scans, and the reconstruction resolution. From both in-vitro and in-vivo studies, it was recommended that the frequency of 2.5 MHz, the spacing of 0.2 mm, and the reconstruction resolution of 0.6 mm constituted the best results. To measure the AVR, the Cobb angle and the lateral deviation on the PA plane and the PMC, the centre-of-lamina method was used.
In-vitro and in-vivo studies were performed and the results demonstrated that the intra- and inter-rater reliabilities were high for all five parameters (ICC > 0. In addition, the Cobb angle measurements from the PA ultrasound images agreed well with the Cobb angle measurements from scoliosis clinics with a small variation (MAD < 3) and high correlation (ICC > 0. The measurements of the lateral deviations also showed high reliabilities (ICC > 0.90 and MAD < 7 mm). Furthermore, the average difference between the PMC Cobb angle and the PA Cobb angle was 1.0 within the range of 0 and 7.
This result agreed with reports from literature. In addition, the AVR from the in-vitro study showed a strong correlation and high agreement between the ultrasound and CT images (ICC > 0. Unfortunately, the in-vivo intervertebral rotations reported from the EOS system did not match to the ultrasound measurements. Further studies will be required to understand the reasons for the discrepancies.
Since the 3D ultrasound was able to provide true spinal deformity information, a study to investigate which demographic and 3D ultrasound parameters correlated with the progression of AIS was conducted. A preliminary predictive model was developed using multi-linear regression and 23 retrospective subjects’ data. The results demonstrated the PMC Cobb angle and the number of vertebrae within the largest curve were the most reliable predictors. A preliminary validation using 6 subjects was performed.
The variation iii between the measured and the predicted Cobb angles was 2. The adjusted r2 was 0.87, indicating a good fit of data to the model. In conclusion, this PhD thesis demonstrated that the 3D freehand ultrasound method could be used to reconstruct 3D images of the scoliotic spine. The AVR, the PMC Cobb angle and lateral deviation could be measure reliably to assess the true severity of AIS.
The PMC Cobb angle and the number of vertebrae within the largest curve were the potential parameters that could be used to predict the progression of AIS. iv Preface This thesis is an original intellectual product of the author, Quang N. The research described in this thesis received ethics approval from the Health Research Ethics Board of the University of Alberta with the project name: “Using ultrasound to assess spinal deformity for AIS”, reference number: Pro00005707, starting January 22, 2010. Portions of the material in this thesis have been published in the following papers: VO, Q.
Investigation of the optimal freehand three- dimensional ultrasound configuration to image scoliosis: An in-vitro study. The fifth international conference on the development of biomedical engineering in Vietnam, Ho Chi Minh City, Vietnam. The materials described in this conference paper are reported in Chapter 4. I perceived, executed, and composed the manuscript of the work described in the paper.
Lou endorsed technical advice and direction, and assisted with manuscript preparation. Le edited the manuscript. Measurement of axial vertebral rotation using three-dimensional ultrasound images. Scoliosis and Spinal Disorders, 10, 1-4.
The materials from this journal article are also reported in Chapter 4. I was responsible for conducting the experiment, acquiring data, developing the software, performing measurements, and analyzing the results. I also composed the manuscript with assistance from Dr. Lou and Dr.
Reconstruction of a scoliotic spine using a three-dimensional medical ultrasound system. The 10th meeting of the International Research Society of Spinal Deformities, Sapporo, Japan, June 29 – July 2 2014. Part of the materials from this conference abstract is included in Chapter 4 of this thesis. I conceived of and executed the work from designing and conducting the experiment, collecting data, developing the software, performing measurements, and analyzing the results.
I composed the abstract with assistance from Dr. Lou and Dr. 3D ultrasound imaging method to assess the true spinal deformity. The 37th annual international conference of the IEEE Engineering in Medicine and Biology Society, Milan, Italy, August 25-29 2015, 1540-3.
The materials from this conference paper are reported in Chapter 4. I contributed to design the experiment, collecting data, developing the software, performing measurements, and analyzing the results. I composed the manuscript with assistance from all co-authors. Prediction of scoliosis progression using three- dimensional ultrasound images: A pilot study.
The first combined meeting of the International Research Society of Spinal Deformities and the Society on Scoliosis Orthopaedic and Rehabilitation Treatment, Banff, Canada, May 25-28 2016, 22. Part of the materials from this conference abstract is included in Chapter 7 of this thesis. I conceived of and executed the work from collecting data, performing measurements, analyzing the results, and developing the predictive model. I composed the abstract with assistance from Dr.
Lou and Dr. Submission of additional papers is planned related to Chapters 4, 5, 6, and 7. vi Acknowledgements This thesis is made as a completion of the PhD education in Biomedical Engineering. Several persons and organizations have contributed academically, financially, and practically to the completion of this PhD thesis.
I would firstly like to acknowledge my supervisor Dr. Edmond Lou and co-supervisor Dr. Le for their time, guidance, valuable input and support throughout my entire PhD period. My gratitude also goes to Doug Hill, Jim Raso, Dr.
Hong Zhao, and Dr. Rui Zheng, to name a few, for their encouragement, advice or support. I am grateful to Dr. Douglas Hedden, Dr.
Marc Moreau, and Dr. James Mahood for supporting the recruitment of patients during their scoliosis clinics. I would also like to appreciate the financial supports from the Vietnam International Education Development (VIED), the Natural Sciences and Engineering Research Council of Canada (NSERC), the Scoliosis Research Society (SRS), the Woman and Children’s Health Research Institute (WCHRI), and the University of Alberta Graduate Students’ Association (GSA). I would have not been able to pursue this PhD work and spread it out internationally without these supports.
Finally, I would like to thank my family, especially my wife Tran, for their love, patience, and support during my time studying at the University of Alberta. vii Table of Contents Abstract. vii List of Figures. xiii List of Abbreviations .1 Anatomy of the spine .1 Anatomical planes of the human body .2 Structures of the human spine .3 Structures of the human vertebrae.
Literature review on imaging modalities and the prediction of AIS progression .1 Imaging modalities to assess the severity of scoliosis .3 Magnetic resonance imaging (MRI) .4 Multi-planar radiography (MPR) .2 Prediction of progression in patients who have AIS. Development of 3D freehand ultrasound reconstruction and determination of the optimal configuration .1 The ultrasound system .3 Position and orientation tracking system .2 Computer hardware and software .3 Three-dimensional ultrasound reconstruction method .1 Step a: Three-dimensional freehand ultrasound data acquisition .2 Step b: Image processing .3 Step c: Formation of a regular volume .4 Step d: Volume visualization.4 Determination of the optimal ultrasound configuration to image a cadaveric vertebra: An in-vitro study.1 Experimental setup and scanning procedures .5 Determination of the optimal ultrasound configuration to image scoliotic spines: An in- vivo study. Subject recruitment and scanning procedure. In-vitro validation of the optimal configuration in imaging AIS .2 Accuracy of the 3D reconstruction of individual cadaveric vertebrae .3 Measurement of the AVR of cadaveric vertebrae .4 Accuracy and reliability of the measurement of the AVR, the tilt angle, and the Cobb angle on the spine phantoms.1 The experimental setup and scanning procedures.
In-vivo repeatability of the ultrasound method, reliability of the ultrasound measurements, and validity of the PA Cobb angle and AVR measurements .1 Repeatability of the 3D ultrasound method in imaging scoliosis .1 Subject recruitment and scanning procedure .2 Reliability of the measurement of the AVR, the Cobb angles, and the lateral deviations 128 6.3 Validity of the measurements of the PA Cobb angle .4 Validity of the measurement of the AVR .1 EOS imaging system. The correlation of 3D ultrasound parameters with the progression of AIS: A pilot study .1 Recruitment of patient data .3 Selection of predictors (independent variables) .4 Development of the PMC Cobb angle predictive model. Conclusions, limitations, and recommendations .1 Summary of the work .3 Limitations and recommendations for future work. 177 xii List of Figures Figure 2.1 Planes of human anatomy (Modified from (Interactive Biology, 2016)).2 The anatomy of the spine (Martini et al.3 Vertebral anatomy: (a) The lateral and inferior view of a vertebra; (b) The inferior view of a vertebra; (c) The posterior view of three vertebrae stacked together; (d) The lateral view of three vertebrae stacked together (.4 The illustration of (a) The AVR; (b) The Cobb angle.5 The illustration of the PMC in scoliosis (b) via the projection of a vector in physics (a).1 An example of a spinal radiograph showing the shadows of the vertebral body, the spinous process and the pedicle.2 The Cobb method of determining the AVR: (a) Normal vertebra without rotation; (b) A vertebra with rotation.
If the spinous process is at a, b, c, d, and beyond d, the vertebral rotation is neutral, +, ++, +++, and ++++, respectively ( .3 The pedicle method of determining vertebral rotation (Nash and Moe, 1969).4 The Aaro-Dahlborn’s methods to measure the AVR on the CT image.5 The Ho’s methods to measure the AVR on the CT image.6 The Krismer’s methods to measure the AVR on the CT image.7 The illustration of SCPs and NSCPs of a 6-vertex object. Vertices 1 and 3 are SCPs while vertices 2, 4, 5, and 6 are NSCPs.8 The Direct linear transformation with stereo corresponding points.9 The pulse-echo technique.10 3D ultrasound scanning configurations (a) Dedicated 3D tilt scanning; (b) Dedicated 3D linear scanning; (c) Dedicated 3D rotational scanning; (d) Freehand 3D scanning.11 The Voxel Nearest Neighbor (VNN) method.