Three-Dimensional Nonrigid MR-TRUS Registration Using Dual Optimization

Sun, Yue; Yuan, Jing; Qiu, Wu; Rajchl, Martin; Romagnoli, Cesare; Fenster, Aaron

doi:10.1109/tmi.2014.2375207

Cited by 33 publications

(15 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, the SM faces discontinuity in intensity problems along out of plane, since most medical imaging modalities keep spacing between the slices. Most studies on 3D-3D ISR have focused on optimization [13], [21], [29], local regularization [30], [31], multi resolution FFD [11], [12] and the application of the diffeomorphic log-demons algorithm [32], [33] to 3D-3D ISR. The most recently developed competitive registration algorithms are the multi resolution FFD and diffeomorphic log-demons approaches.…”

Section: Related Workmentioning

confidence: 99%

A Deformable 3D-3D Registration Framework Using Discrete Periodic Spline Wavelet and Edge Position Difference

et al. 2020

View full text Add to dashboard Cite

Neck pain is one of the most common symptoms of cervical spine disease, and segmenting neck muscles to create volumetric measurements may assist clinical diagnosis. While image registration is used to segment medical images, registration is highly challenging for neck muscles due to their tight proximity, shape and size variations among subjects, and similar appearance. These challenges cause conventional multi resolution-based registration methods to be trapped in local minima due to their low degree of freedom geometrical transforms. This paper presents a novel object-constrained hierarchical registration framework for aligning inter-subject neck muscles. First, to handle large scale local minima, the proposed framework uses a coarse registration technique, which optimizes the new edge position difference (EPD) similarity measure, to align large mismatches. Also, a new transformation based on the discrete periodic spline wavelet (DPSW), affine and free-form-deformation (FFD) transformations are exploited. Second, to avoid monotonous nature of using transformations in multiple stages, a fine registration technique is designed for aligning small mismatches. This technique uses a double-pushing system by changing edges in the EPD and switching transformation resolutions. The EPD helps in both coarse and fine techniques to implement object-constrained registration via controlling edges, which is not possible when using traditional similarity measures. Experiments are performed on clinical 3D magnetic resonance imaging (MRI) scans of the neck, with the results showing that the EPD is more effective than the mutual information (MI) and sum of squared difference (SSD) measure in terms of volumetric dice similarity coefficient (DSC). Additionally, the proposed method is compared with the diffeomorphic Demons and SyN state-of-the-art approaches with ablation studies in inter-subject deformable registration. The proposed method achieves better accuracy, robustness and consistency than the reference methods, with an average volumetric DSC of 0.7029 compared to 0.6654 and 0.6606 for the Demons and SyN methods, respectively. INDEX TERMS Neck muscles, edge position difference (EPD), discrete periodic spline wavelet (DPSW), magnetic resonance imaging (MRI), deformable registration.

show abstract

Section: Related Workmentioning

confidence: 99%

A Deformable 3D-3D Registration Framework Using Discrete Periodic Spline Wavelet and Edge Position Difference

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Landmarks were chosen as the maximum and minimum coordinates on the prostate surfaces in each of the imaging dimensions (x, y, and z), which corresponded approximately to the anatomical planes (sagittal, coronal, and axial). These landmarks are similar to the ones used by Sun et al 18 and described by Mahdavi et al 19 The landmarks L max pre and L min pre on the preoperative image were defined using the preoperative prostate surface P pre & R 3 as Medical Physics, 46 (2), February 2019…”

Section: B2 Linear Registrationmentioning

confidence: 99%

Evaluation of tumor coverage after MR‐guided prostate focal laser ablation therapy

et al. 2018

Self Cite

View full text Add to dashboard Cite

Purpose Prostate cancer is the most common noncutaneous cancer among men in the USA. Focal laser thermal ablation (FLA) has the potential to control small tumors while preserving urinary and erectile function by leaving the neurovascular bundles and urethral sphincters intact. Accurate needle guidance is critical to the success of FLA. Multiparametric magnetic resonance images (mpMRI) can be used to identify targets, guide needles, and assess treatment outcomes. In this study, we evaluated the location of ablation zones relative to targeted lesions in 23 patients who underwent FLA therapy in a phase II trial. The ablation zone margins and unablated tumor volume were measured to determine whether complete coverage of each tumor was achieved, which would be considered a clinically successful ablation. Methods Preoperative mpMRI was acquired for each patient 2–3 months preceding the procedure and the prostate and lesion(s) were manually contoured on 3 T T2‐weighted axial images. The prostate and ablation zone(s) were also manually contoured on postablation 1.5 T T1‐weighted contrast‐enhanced axial images acquired immediately after the procedure intraoperatively. The lesion surface was nonrigidly registered to the postablation image using an initial affine registration followed by nonrigid thin‐plate spline registration of the prostate surfaces. The margins between the registered lesion and ablation zone were calculated using a uniform spherical distribution of rays, and the volume of intersection was also calculated. Each prostate was contoured five times to determine the segmentation variability and its effect on intersection of the lesion and ablation zone. Results Our study showed that the boundaries of the segmented tumor and ablation zone were close. Of the 23 lesions that were analyzed, 11 were completely covered by the ablation zone and 12 were partially covered. A shift of 1.0, 2.0, and 2.6 mm would result in 19, 21, and all tumors completely covered by the ablation zone, respectively. The median unablated tumor volume across all tumors was 0.1 mm3 with an IQR of 3.7 mm3, which was 0.2% of the median tumor volume (46.5 mm3 with an IQR of 46.3 mm3). The median extension of the tumors beyond the ablation zone, in cases which were partially ablated, was 0.9 mm (IQR of 1.3 mm), with the furthest tumor extending 2.6 mm. Conclusion In all cases, the boundary of the tumor was close to the boundary of the ablation zone, and in some cases, the boundary of the ablation zone did not completely enclose the tumor. Our results suggest that some of the ablations were not clinically successful and that there is a need for more accurate needle tracking and guidance methods. Limitations of the study include errors in the registration and segmentation methods used as well as different voxel sizes and contrast between the registered T2 and T1 MRI sequences and asymmetric swelling of the prostate postprocedurally.

show abstract

“…To efficiently explore the solution space and also enhance the capture range, the proposed approach uses differential evolution, followed by the local second order algorithm -BFGS. The application of the developed approach shows promising performance and also advantages over the classical mutual information and MIND based approaches [23].…”

Section: Introductionmentioning

confidence: 98%

Learning deep similarity metric for 3D MR–TRUS image registration

et al. 2018

View full text Add to dashboard Cite

Purpose The fusion of transrectal ultrasound (TRUS) and magnetic resonance (MR) images for guiding targeted prostate biopsy has significantly improved the biopsy yield of aggressive cancers. A key component of MR-TRUS fusion is image registration. However, it is very challenging to obtain a robust automatic MR-TRUS registration due to the large appearance difference between the two imaging modalities. The work presented in this paper aims to tackle this problem by addressing two challenges: (i) the definition of a suitable similarity metric and (ii) the determination of a suitable optimization strategy.Methods This work proposes the use of a deep convolutional neural network to learn a similarity metric for MR-TRUS registration. We also use a composite optimization strategy that explores the solution space in order to search for a suitable initialization for the second-order optimization of the learned

show abstract

Three-Dimensional Nonrigid MR-TRUS Registration Using Dual Optimization

Cited by 33 publications

References 38 publications

A Deformable 3D-3D Registration Framework Using Discrete Periodic Spline Wavelet and Edge Position Difference

A Deformable 3D-3D Registration Framework Using Discrete Periodic Spline Wavelet and Edge Position Difference

Evaluation of tumor coverage after MR‐guided prostate focal laser ablation therapy

Learning deep similarity metric for 3D MR–TRUS image registration

Contact Info

Product

Resources

About