Vertebra segmentation based on two-step refinement

Abstract

Knowledge of vertebra location, shape, and orientation is crucial in many medical applications such as orthopedics or interventional procedures. Computed tomography (CT) offers a high contrast between bone and soft tissues, but automatic vertebra segmentation remains difficult. Hence, the wide range of shapes, aging, and degenerative joint disease alterations as well as the variety of pathological cases encountered in an aging population make automatic segmentation sometimes challenging. Besides, daily practice implies a need for affordable computation time.This paper aims to present a new automated vertebra segmentation method (using a first bounding box for initialization) for CT 3D data which tackles these problems. This method is based on two consecutive steps. The first one is a new coarse-to-fine method efficiently reducing the data amount to obtain a coarse shape of the vertebra. The second step consists in a hidden Markov chain (HMC) segmentation using a specific volume transformation within a Bayesian framework. Our method does not introduce any prior on the expected shape of the vertebra within the bounding box and thus deals with the most frequent pathological cases encountered in daily practice.We experiment this method on a set of standard lumbar, thoracic, and cervical vertebrae and on a public dataset, on pathological cases, and in a simple integration example. Quantitative and qualitative results show that our method is robust to changes in shapes and luminance and provides correct segmentation with respect to pathological cases.

Unsupervised segmentation of stents corrupted by artifacts in medical X-ray images

“…This way, physical continuity in the path is simulated, gathering all the stent parts in a 1D sequence which allows efficient computations. Such a path is particularly suited to get a smoothed and regular surface for tubular structures [4]. Moreover, in the HMC model one has to learn the transition probabilities between the classes.…”

“…2) The HMC model: We follow the statistical approach developped in [4]. We refer the reader to the latter article for all the details about the statistical modeling.…”

Automatic Vertebral Body Segmentation using Semantic Segmentation

“…Even though CT images offer a high contrast between bone and soft tissues, automatic vertebra segmentation remains difficult. Factors like complex vertebral structure, images taken with restricted field of view causing lack of broader contextual Automatic Vertebral Body Segmentation using Semantic Segmentation information [1], unclear boundaries of the vertebrae, abnormal spine curves like scoliosis, adjacent bone regions, aging, degenerative joint disease alterations as well as the variety of pathological cases like osteoporosis encountered in an aging population, tumor, deformations, trauma, and fractures make automatic segmentation challenging [5]. In addition, the output of segmentation algorithm is affected due to intensity inhomogeneity and partial volume effect.…”

Visualization of stereoscopic anatomic models of the paranasal sinuses and cervical vertebrae from the surgical and procedural perspective

“…Computer models have gained attention as a means to delineate detailed clinical anatomy (Qayumi et al, ; Nicholson et al, ; Brown et al, ), but few medical schools use virtual models in their clinical anatomy course (Adams and Wilson, ; Yeung et al, ; Cui et al, ). The process of learning new surgical skills requires visual‐spatial ability and can be difficult for some learners; however, the impact of a 3D video learning module on the acquisition of new surgical skills has not been previously performed (Roach et al, ), although there have been reports demonstrating use of 3D reconstruction of cervical vertebrae for different purposes (Clogenson et al, ; Zheng et al, ; Courbot et al, ) as well as 3D models of paranasal sinuses in a virtual surgical environment (Parikh et al, ; Kapakin, ). Most of these reports use routine volume rendering rather than stereoscopic 3D presentation for simulation of these structures in a virtual reality environment for surgical and procedure training purposes.…”

“…Visualization of the regional anatomy of the vertebrae poses similar difficulties. Vertebra segmentation is a challenging task, and the pathologic alterations frequently encountered in real cases explain the difficulties of an automatic segmentation in daily practice patients (Courbot et al, ). To prevent complications of cervical epidural injections, it is also essential for the trainee to understand the spatial relationship between cervical vertebrae and the surrounding structures, such as spinal nerves and blood vessels (Botwin et al, ).…”