A 4D biomechanical lung phantom for joint segmentation/registration evaluation

Abstract

At present, there exists few openly available methods for evaluation of simultaneous segmentation and registration algorithms. These methods allow for a combination of both techniques to track the tumor in complex settings such as adaptive radiotherapy. We have produced a quality assurance platform for evaluating this specific subset of algorithms using a preserved porcine lung in such that it is multi-modality compatible: positron emission tomography (PET), computer tomography (CT) and magnetic resonance imaging (MRI). A computer controlled respirator was constructed to pneumatically manipulate the lungs in order to replicate human breathing traces. A registration ground truth was provided using an in-house bifurcation tracking pipeline. Segmentation ground truth was provided by synthetic multi-compartment lesions to simulate biologically active tumor, background tissue and a necrotic core. The bifurcation tracking pipeline results were compared to digital deformations and used to evaluate three registration algorithms, Diffeomorphic demons, fast-symmetric forces demons and MiMVista's deformable registration tool. Three segmentation algorithms the Chan Vese level sets method, a Hybrid technique and the multi-valued level sets algorithm. The respirator was able to replicate three seperate breathing traces with a mean accuracy of 2-2.2%. Bifurcation tracking error was found to be sub-voxel when using human CT data for displacements up to 6.5 cm and approximately 1.5 voxel widths for displacements up to 3.5 cm for the porcine lungs. For the fast-symmetric, diffeomorphic and MiMvista registration algorithms, mean geometric errors were found to be [Formula: see text], [Formula: see text] and [Formula: see text] voxels widths respectively using the vector field differences and [Formula: see text], [Formula: see text] and [Formula: see text] voxel widths using the bifurcation tracking pipeline. The proposed phantom was found sufficient for accurate evaluation of registration and segmentation algorithms. The use of automatically generated anatomical landmarks proposed can eliminate the time and potential innacuracy of manual landmark selection using expert observers.

Clinical implementations of 4D pencil beam scanned particle therapy: Report on the 4D treatment planning workshop 2016 and 2017

“…The future of the 4D phantoms is in new material compositions, including approaches to build phantom components from 3D printing materials on the basis of real lung patient CT scans [70]. Also creating phantoms which partly consist of (preserved) real tissue, e.g porchine lungs which can be externally inflated [69] are more and more popular. This is especially used for 4D imaging, segmentation and registration validation but have the disadvantage of having a short lifetime.…”

Advances in anthropomorphic thorax phantoms for radiotherapy: a review

“…Biederer et al have also constructed a dynamic lung phantom based on porcine lungs [7]. In a study, Markel et al have designed a phantom using actual tissues for testing both segmentation and registration [73]. They utilized porcine lungs to simulate human lungs owing to similarities in anatomy and capacity.…”

“…They utilized porcine lungs to simulate human lungs owing to similarities in anatomy and capacity. Then, those lungs connected to an air flow system to provide inflation and deflation of the lung [73]. The proposed phantom was compatible with CT, MRI, and PET.…”

“…To this end, novel deformable lung phantom providing the advantage of emulating features of the human lungs has been designed [78]. Although the phantom consists of a driving motor, a fluid pump, and a lung part, similar to the typical phantom designs, the key feature of this phantom is the discretization of the lung part by small bricks, which offer two advantages over previously developed lung phantoms in the literature [26,44,45,59,73]. First, a fairly large deformation of the lung part that is essential for radiotherapy treatment planning.…”

Complete region of interest reconstruction by fusing multiview deformable three‐dimensional transesophageal echocardiography images

“…17,18 Following Thirion's Demons method, a range of variants were proposed to improve convergence speed and precision. 18 Among these, the diffeomorphic Demons algorithm 19 which can preserve the topology of the objects achieves remarkable success in various computational anatomy studies, such as breast computed tomography (CT), 20 brain magnetic resonance (MR), 21 and lung CT. 22 In recent years, learning-based registration methods have attracted a lot of attention because of their high efficiency after training. 23,24 The state-of -the-art learning-based method VoxelMorph has shown comparable accuracy to the nonlearning-based methods.…”