Abstract
An important factor for the development of spinal degeneration, pain and the outcome of spinal surgery is known to be the balance of the spine. It must be analyzed in an upright, standing position to ensure physiological loading conditions and visualize load-dependent deformations. Despite the complex 3D shape of the spine, this analysis is currently performed using 2D radiographs, as all frequently used 3D imaging techniques require the patient to be scanned in a prone position. To overcome this limitation, we propose a deep neural network to reconstruct the 3D spinal pose in an upright standing position, loaded naturally. Specifically, we propose a novel neural network architecture, which takes orthogonal 2D radiographs and infers the spine’s 3D posture using vertebral shape priors. In this work, we define vertebral shape priors using an atlas and a spine shape prior, incorporating both into our proposed network architecture. We validate our architecture on digitally reconstructed radiographs, achieving a 3D reconstruction Dice of , indicating an almost perfect 2D-to-3D domain translation. Validating the reconstruction accuracy of a 3D standing spine on real data is infeasible due to the lack of a valid ground truth. Hence, we design a novel experiment for this purpose, using an orientation invariant distance metric, to evaluate our model’s ability to synthesize full-3D, upright, and patient-specific spine models. We compare the synthesized spine shapes from clinical upright standing radiographs to the same patient’s 3D spinal posture in the prone position from CT.
Highlights
A biomechanical load analysis of the spine in an upright standing position is highly warranted in various spine disorders to understand their cause and guide therapy [1]
In order to analyze the contribution of various architectural components of the TransVert+ and to validate its performance on clinical radiographs, we propose three sets of experiments
We introduced TransVert+, a neural network architecture to reconstruct a full 3D spinal model from 2D orthogonal radiographs by deforming vertebral shapes
Summary
A biomechanical load analysis of the spine in an upright standing position is highly warranted in various spine disorders to understand their cause and guide therapy [1]. Typical approaches for load estimation either use a computational shape model of the spine for all patients or obtain a subject-specific spine model from a 3D imaging modality such as magnetic resonance imaging (MRI) or computed tomography (CT) [2]. Even though MRI and CT images can capture 3D anatomical information, they need the patient to be in a prone or supine position (lying flat on a table) during imaging. To analyze the spinal alignment in a physiologically upright standing position under weight bearing, orthogonal 2D plain radiographs (as depicted in Figure 1) are the de facto choice. A combination of both these worlds is of clinical interest to fully assess the true biomechanical situation, that is, to capture the patient-specific complex pathological spinal arrangement in a standing position with full 3D information [2,3,4]. Typical projections—that is, lateral and AP radiographs— cover most of these variations, both on a local (per vertebra) and global (overall spinal alignment) level
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have