To Align Multimodal Lumbar Spine Images via Bending Energy Constrained Normalized Mutual Information

August 1, 2020

. 2020 Jul 10;2020:5615371.

doi: 10.1155/2020/5615371.

eCollection 2020.

Affiliations

¹ Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, China.
² Department of Radiology, Shenzhen Second People’s Hospital, The First Affiliated Hospital of Shenzhen University, Shenzhen 518035, China.
³ National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital & Shenzhen Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Shenzhen 518116, China.
⁴ Department of Radiation Oncology, University of Texas, Southwestern Medical Center, Dallas, TX 75390, USA.

Item in Clipboard

Shibin Wu et al.

Biomed Res Int.

2020.

. 2020 Jul 10;2020:5615371.

doi: 10.1155/2020/5615371.

eCollection 2020.

Affiliations

¹ Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, China.
² Department of Radiology, Shenzhen Second People’s Hospital, The First Affiliated Hospital of Shenzhen University, Shenzhen 518035, China.
³ National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital & Shenzhen Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Shenzhen 518116, China.
⁴ Department of Radiation Oncology, University of Texas, Southwestern Medical Center, Dallas, TX 75390, USA.

Item in Clipboard

Abstract

To align multimodal images is important for information fusion, clinical diagnosis, treatment planning, and delivery, while few methods have been dedicated to matching computerized tomography (CT) and magnetic resonance (MR) images of lumbar spine. This study proposes a coarse-to-fine registration framework to address this issue. Firstly, a pair of CT-MR images are rigidly aligned for global positioning. Then, a bending energy term is penalized into the normalized mutual information for the local deformation of soft tissues. In the end, the framework is validated on 40 pairs of CT-MR images from our in-house collection and 15 image pairs from the SpineWeb database. Experimental results show high overlapping ratio (in-house collection, vertebrae 0.97 ± 0.02, blood vessel 0.88 ± 0.07; SpineWeb, vertebrae 0.95 ± 0.03, blood vessel 0.93 ± 0.10) and low target registration error (in-house collection, ≤2.00 ± 0.62 mm; SpineWeb, ≤2.37 ± 0.76 mm) are achieved. The proposed framework concerns both the incompressibility of bone structures and the nonrigid deformation of soft tissues. It enables accurate CT-MR registration of lumbar spine images and facilitates image fusion, spine disease diagnosis, and interventional treatment delivery.

Conflict of interest statement

The authors declare there is no conflict of interest. The founding sponsors had no role in the design of this study, in the collection, analysis or interpretation of data, in the writing of this manuscript, nor in the decision to publish the experimental results.

Figures

**Figure 1**

Perceived visual difference between CT and MR images of lumbar spine from three perspective views. The difference of imaging characteristics, fields of view, and unavoidable motion make the registration challenging. Red arrows show different imaging contrast, green arrows direct to the undesirable artifact of bias field in MR images, and blue arrows indicate different field of views. Note that images are cropped and scaled for display purpose.

**Figure 2**

The proposed coarse-to-fine framework for aligning CT-MR lumbar spine images. It consists of two stages. The first stage is for global positioning via NMI based rigid registration (highlighted in red), and the second stage is for the local deformation of soft tissues via the bending energy penalized NMI (highlighted in yellow). Both stages utilize the same workflow for iterative optimization.

**Figure 3**

Jaccard index of the vertebrae and blood vessel overlapping on the in-house dataset (a) and the online dataset (b). Box-and-whisker plots represent the median Jaccard index (horizontal line) and total range (whiskers). The red ⁺ indicates an outlier that causes failure in image registration.

**Figure 4**

Tissue overlapping metric of Dice coefficient of the vertebrae and blood vessel on the in-house dataset (a) and the online dataset (b). Box-and-whisker plots show the median coefficient (horizontal line) and total range (whiskers). The red ⁺ indicates a failure case.

**Figure 5**

TRE values of anatomical landmarks on the in-house collection.

**Figure 6**

TRE values of anatomical landmarks on the SpineWeb dataset.

**Figure 7**

Perceived visual difference of CT-MR images before and after image registration. The regions directed by the arrows are for comparison before and after registration. In addition, images are cropped for display purpose.

Similar articles

Volume-preserving nonrigid registration of MR breast images using free-form deformation with an incompressibility constraint.

Rohlfing T, Maurer CR Jr, Bluemke DA, Jacobs MA.
Rohlfing T, et al.
IEEE Trans Med Imaging. 2003 Jun;22(6):730-41. doi: 10.1109/TMI.2003.814791.
IEEE Trans Med Imaging. 2003.

PMID: 12872948

Clinical Trial.
Accurate CT∕MR vessel-guided nonrigid registration of largely deformed livers.

Vasquez Osorio EM, Hoogeman MS, Mendez Romero A, Wielopolski P, Zolnay A, Heijmen BJ.
Vasquez Osorio EM, et al.
Med Phys. 2012 May;39(5):2463-77. doi: 10.1118/1.3701779.
Med Phys. 2012.

PMID: 22559617
Multimodal image registration with joint structure tensor and local entropy.

Zhang J, Wang J, Wang X, Feng D.
Zhang J, et al.
Int J Comput Assist Radiol Surg. 2015 Nov;10(11):1765-75. doi: 10.1007/s11548-015-1219-9. Epub 2015 May 28.
Int J Comput Assist Radiol Surg. 2015.

PMID: 26018848
Multi-segmental spine image registration supporting image-guided interventions of spinal metastases.

Hille G, Saalfeld S, Serowy S, Tönnies K.
Hille G, et al.
Comput Biol Med. 2018 Nov 1;102:16-20. doi: 10.1016/j.compbiomed.2018.09.003. Epub 2018 Sep 13.
Comput Biol Med. 2018.

PMID: 30236968
Performance evaluation of MIND demons deformable registration of MR and CT images in spinal interventions.

Reaungamornrat S, De Silva T, Uneri A, Goerres J, Jacobson M, Ketcha M, Vogt S, Kleinszig G, Khanna AJ, Wolinsky JP, Prince JL, Siewerdsen JH.
Reaungamornrat S, et al.
Phys Med Biol. 2016 Dec 7;61(23):8276-8297. doi: 10.1088/0031-9155/61/23/8276. Epub 2016 Nov 3.
Phys Med Biol. 2016.

PMID: 27811396
Free PMC article.

References

1. Zheng G., Li S. Medical image computing in diagnosis and intervention of spinal diseases. Computerized Medical Imaging and Graphics. 2015;45:99–101. doi: 10.1016/j.compmedimag.2015.08.006.
  
  –
  
  DOI
  
  –
  
  PubMed
1. Raciborski F., Gasik R., Klak A. Disorders of the spine. A major health and social problem. Annals of Physics. 2016;4(4):196–200. doi: 10.5114/reum.2016.62474.
  
  –
  
  DOI
  
  –
  
  PMC
  
  –
  
  PubMed
1. Zhang G., Yang Y., Hai Y., Li J., Xie X., Feng S. Analysis of Lumbar Sagittal Curvature in Spinal Decompression and Fusion for Lumbar Spinal Stenosis Patients under Roussouly Classification. BioMed Research International. 2020;2020:8. doi: 10.1155/2020/8078641.
  
  –
  
  DOI
  
  –
  
  PMC
  
  –
  
  PubMed
1. Toussaint A., Richter A., Mantel F., et al. Variability in spine radiosurgery treatment planning – results of an international multi-institutional study. Radiation Oncology. 2016;11(1):p. 57. doi: 10.1186/s13014-016-0631-9.
  
  –
  
  DOI
  
  –
  
  PMC
  
  –
  
  PubMed
1. Helm P. A., Teichman R., Hartmann S. L., Simon D. Spinal navigation and imaging: history, trends, and future. IEEE Transactions on Medical Imaging. 2015;34(8):1738–1746. doi: 10.1109/TMI.2015.2391200.
  
  –
  
  DOI
  
  –
  
  PubMed