3-D Ultrasound Diagnosis And Visualization Of Scoliosis

Scoliosis is a common 3-D deformity of the spine,including sequence abnormalities in the coronal,sagittal,and traverse plane.Adolescent Idiopathic Scoliosis(AIS)is the most common type of scoliosis.It is a common condition that seriously affects the physical and mental health of adolescents.The golden standard for clinical diagnosis of AIS is Cobb’s angle detection based on X-ray imaging.However,the long-term use of this method to follow the spine curve progress produces ionizing radiation that will affect the growth of adolescents and increase the risk of cancer.Therefore,the freehand 3-D ultrasound imaging technology,which has the advantages of real-time,low cost,and free-radiation,has been actively used to assess spine anatomy to reduce the harm of conventional X-ray examination.However,the existing methods have not fully utilized the volume data to reveal the 3-D distortion of the spine truly.It will ultimately affect the diagnosis of scoliosis by clinicians.Therefore,volume data visualization and the key algorithms of image processing are studied in this thesis,and the imaging and measurement methods are proposed for scoliosis.Also,a freehand 3-D ultrasound system is developed and verified for scoliosis assessment.The main work of this thesis is combining classic image processing algorithms and visualization algorithms to design a 3-D diagnostic system for the assessment of scoliosis.It includes the following points:(i)Based on the coronal volume projection image(VPI)of the 3-D ultrasound spinal data,an adaptive method combined with oriented phase congruency is proposed for spinous column profile detection by further using the spatial distribution of spinous processes and other anatomical structures.It excavates the prior knowledge of vertebral anatomy better,realizes the spinous column profile identification and the spine curvature measurement automatically in the coronal VPI image.The experiments demonstrate that the proposed method has a more excellent performance in terms of accuracy and speed and can extract the spine curve automatically from the coronal VPI image acquired at various depths.(ii)A sagittal volume projection imaging method based on spinal anatomy is proposed in this thesis.The 3-D ultrasound diagnosis methods do not fully use the volume data at the present phase and only detect the spine curvatures on a single visual image.Due to the particularity of the spine anatomy,the conventional sagittal slice method can only present part of vertebrae in an image.Thus,the spine curvature of the sagittal can not be derived from these conventional projection images.Therefore,we extract the spine curves from VPI images at various depths of one subject using the mentioned automatic extraction algorithm.The vertebrae’s spatial position with the corresponding spinous process is obtained.It motivates us to generate a sagittal referred profile based on the spinal anatomy.Based on this sagittal reference profile,the sagittal volume projection image,including all vertebrae,can generate,and the measurement of the sagittal spine curvature is therefore achieved.The results show that the spine curvature measured in the sagittal projection image has a good correlation and consistency with the measurement results of the X-ray sagittal plane(lateral)Cobb’s method.(iii)Based on the above research and existing freehand 3-D ultrasound system,a freehand 3-D ultrasound system for scoliosis assessment is developed in this thesis combined with mentioned algorithms.The system realizes the functions of acquisition,reconstruction,visualization,and interactive observation often found in the previous freehand 3-D ultrasound system and integrates the function of spine curvature detection with multi-visualization images.Furthermore,based on the manually selected position of the key anatomical features in the coronal and sagittal images,the system also loads the pre-designed standard 3-D vertebral models to perform a simulated 3-D spine model by matching the corresponding position in ultrasound data with rigid transformation.Thus,it proposes a new solution for the 3-D evaluation and analysis of spine deformity.In general,compared with previous methods,this system can generate coronal,sagittal,and transverse sections with a single scan at the same time,and the whole spine anatomy is visualized by selecting some specific feature points manually.Moreover,the system is nonradiative,flexible,and inexpensive,conducive to deployment in hospitals,and has huge development and application potential in the future.