Lung Deformation Estimation with Non-rigid Registration for Radiotherapy Treatment

IEEE Trans. Med. Imaging

Desbat

2009

Abstract-Respiratory motion is a major concern in cone-beam (CB) computed tomography (CT) of the thorax. It causes artifacts such as blur, streaks, and bands, in particular when using slow-rotating scanners mounted on the gantry of linear accelerators. In this paper, we compare two approaches for motion-compensated CBCT reconstruction of the thorax. The first one is analytic; it is heuristically adapted from the method of Feldkamp, Davis, and Kress (FDK). The second one is algebraic: the system of linear equations is generated using a new algorithm for the projection of deformable volumes and solved using the Simultaneous Algebraic Reconstruction Technique (SART). For both methods, we propose to estimate the motion on patient data using a previously acquired 4-D CT image. The methods were tested on two digital and one mechanical motion-controlled phantoms and on a patient dataset. Our results indicate that the two methods correct most motion artifacts. However, the analytic method does not fully correct streaks and bands even if the motion is perfectly estimated due to the underlying approximation. In contrast, the algebraic method allows us full correction of respiratory-induced artifacts.

Section: Patient Motion Estimationmentioning

confidence: 99%

Comparison of Analytic and Algebraic Methods for Motion-Compensated Cone-Beam CT Reconstruction of the Thorax

Rit

IEEE Trans. Med. Imaging

Desbat

2009

“…It is based on the physical equations of the deforming material, assuming that the relationship between strain and stress is linear. Other regularizing energies are the membrane or Laplacian model (which can be considered as a simplification of the linear elastic model [46]), bi-harmonic [38] (TPS correspond to an exact solution to this energy minimization), viscous fluid [47,48,49] (same equations as for the elastic model but applied to the velocity field instead of the displacement field), Jacobian-based [50,51], Gaussian [52,48] (which can be related to elastic regularization under some assumptions), etc. Interested readers should refer to Cachier et al [53] who report that almost all regularization energies are based on the same small set of differential quadratic forms.…”

Section: Transformation Modelsmentioning

confidence: 99%

Deformable registration for image-guided radiation therapy

Zeitschrift für Medizinische Physik

2006

Self Cite

Purpose. In this paper, we review several applications of deformable registration algorithms in the field of image-guided radiotherapy.Materials and methods. We first describe the input and output of deformable registration algorithms. Section 3 is dedicated to a brief review of common methods. Several examples of the use of deformable registration are reviewed in section 4(some equations are given in appendix A). The first four sets of examples deal with intrapatient registration: two sections are dedicated to inter-fraction registration (time interval between images in the order of several days), and the other two sections with intra-fraction registration (displacement during image acquisition or patient treatment, mainly due to respiratory movement; time interval in the order of several tenths of seconds). The last examples (section 4.5) focus on inter-patient registration. Conclusion. Deformable registration has increasing applications in radiotherapy.Extensive validation of the numerous existing methods is required before extending its clinical use. Nevertheless, deformable registration is already a fundamental image analysis tool for radiotherapy, and will probably be included into all treatment planning systems in the near future.

“…the corresponding 3D image. Our team had produced such a 4D model not elaborated from a complete 4D acquisition but from two 3D breathhold acquisitions, one at the end of normal expiration (I1) and one at the end of normal inspiration (I2), acquired using spiral CT imaging and the Active Breathing Coordinator (ABC, Elekta Oncology Systems) [15]. The non-rigid registration of I2 on I1 produced a dense vector-field representing the displacement of each point of I2 toward I1.…”

Section: D Modelmentioning

confidence: 99%

Respiratory Signal Extraction for 4D CT Imaging of the Thorax from Cone-Beam CT Projections

Rit

Lecture Notes in Computer Science

Ginestet

2005

Self Cite

Abstract. Current methods of four-dimensional (4D) CT imaging of the thorax synchronise the acquisition with a respiratory signal to restrospectively sort acquired data. The quality of the 4D images relies on an accurate description of the position of the thorax in the respiratory cycle by the respiratory signal. Most of the methods used an external device for acquiring the respiratory signal. We propose to extract it directly from thorax cone-beam (CB) CT projections. This study implied two main steps: the simulation of a set of CBCT projections, and the extraction, selection and integration of motion information from the simulation output to obtain the respiratory signal. A real respiratory signal was used for simulating the CB acquisition of a breathing patient. We extracted from CB images a respiratory signal with 96.4% linear correlation with the reference signal, but we showed that other measures of the quality of the extracted respiratory signal were required.