3D SPECT/CT fusion using image data projection of bone SPECT onto 3D volume-rendered CT images: feasibility and clinical impact in the diagnosis of bone metastasis

Yuji Ogata, Tadaki Nakahara, Kenichi Ode, Yoji Matsusaka, Mari Katagiri, Yu Iwabuchi, Kazunari Itoh, Akira Ichimura, Masahiro Jinzaki

Research output: Contribution to journalArticle

7 Citations (Scopus)

Abstract

Purpose: We developed a method of image data projection of bone SPECT into 3D volume-rendered CT images for 3D SPECT/CT fusion. The aims of our study were to evaluate its feasibility and clinical usefulness. Methods: Whole-body bone scintigraphy (WB) and SPECT/CT scans were performed in 318 cancer patients using a dedicated SPECT/CT systems. Volume data of bone SPECT and CT were fused to obtain 2D SPECT/CT images. To generate our 3D SPECT/CT images, colored voxel data of bone SPECT were projected onto the corresponding location of the volume-rendered CT data after a semi-automatic bone extraction. Then, the resultant 3D images were blended with conventional volume-rendered CT images, allowing to grasp the three-dimensional relationship between bone metabolism and anatomy. WB and SPECT (WB + SPECT), 2D SPECT/CT fusion, and 3D SPECT/CT fusion were evaluated by two independent reviewers in the diagnosis of bone metastasis. The inter-observer variability and diagnostic accuracy in these three image sets were investigated using a four-point diagnostic scale. Results: Increased bone metabolism was found in 744 metastatic sites and 1002 benign changes. On a per-lesion basis, inter-observer agreements in the diagnosis of bone metastasis were 0.72 for WB + SPECT, 0.90 for 2D SPECT/CT, and 0.89 for 3D SPECT/CT. Receiver operating characteristic analyses for the diagnostic accuracy of bone metastasis showed that WB + SPECT, 2D SPECT/CT, and 3D SPECT/CT had an area under the curve of 0.800, 0.983, and 0.983 for reader 1, 0.865, 0.992, and 0.993 for reader 2, respectively (WB + SPECT vs. 2D or 3D SPECT/CT, p < 0.001; 2D vs. 3D SPECT/CT, n.s.). The durations of interpretation of WB + SPECT, 2D SPECT/CT, and 3D SPECT/CT images were 241 ± 75, 225 ± 73, and 182 ± 71 s for reader 1 and 207 ± 72, 190 ± 73, and 179 ± 73 s for reader 2, respectively. As a result, it took shorter time to read 3D SPECT/CT images than 2D SPECT/CT (p < 0.0001) or WB + SPECT images (p < 0.0001). Conclusions: 3D SPECT/CT fusion offers comparable diagnostic accuracy to 2D SPECT/CT fusion. The visual effect of 3D SPECT/CT fusion facilitates reduction of reading time compared to 2D SPECT/CT fusion.

Original languageEnglish
Pages (from-to)1-11
Number of pages11
JournalAnnals of Nuclear Medicine
DOIs
Publication statusAccepted/In press - 2017 Feb 27

Fingerprint

Cone-Beam Computed Tomography
Single-Photon Emission-Computed Tomography
Neoplasm Metastasis
Bone and Bones
Radionuclide Imaging
Single Photon Emission Computed Tomography Computed Tomography

Keywords

  • Bone metastasis
  • Image fusion
  • SPECT/CT
  • Three-dimensional
  • Volume rendering

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging

Cite this

@article{82ea9832d2144b2caddc4f6bc9dd3638,
title = "3D SPECT/CT fusion using image data projection of bone SPECT onto 3D volume-rendered CT images: feasibility and clinical impact in the diagnosis of bone metastasis",
abstract = "Purpose: We developed a method of image data projection of bone SPECT into 3D volume-rendered CT images for 3D SPECT/CT fusion. The aims of our study were to evaluate its feasibility and clinical usefulness. Methods: Whole-body bone scintigraphy (WB) and SPECT/CT scans were performed in 318 cancer patients using a dedicated SPECT/CT systems. Volume data of bone SPECT and CT were fused to obtain 2D SPECT/CT images. To generate our 3D SPECT/CT images, colored voxel data of bone SPECT were projected onto the corresponding location of the volume-rendered CT data after a semi-automatic bone extraction. Then, the resultant 3D images were blended with conventional volume-rendered CT images, allowing to grasp the three-dimensional relationship between bone metabolism and anatomy. WB and SPECT (WB + SPECT), 2D SPECT/CT fusion, and 3D SPECT/CT fusion were evaluated by two independent reviewers in the diagnosis of bone metastasis. The inter-observer variability and diagnostic accuracy in these three image sets were investigated using a four-point diagnostic scale. Results: Increased bone metabolism was found in 744 metastatic sites and 1002 benign changes. On a per-lesion basis, inter-observer agreements in the diagnosis of bone metastasis were 0.72 for WB + SPECT, 0.90 for 2D SPECT/CT, and 0.89 for 3D SPECT/CT. Receiver operating characteristic analyses for the diagnostic accuracy of bone metastasis showed that WB + SPECT, 2D SPECT/CT, and 3D SPECT/CT had an area under the curve of 0.800, 0.983, and 0.983 for reader 1, 0.865, 0.992, and 0.993 for reader 2, respectively (WB + SPECT vs. 2D or 3D SPECT/CT, p < 0.001; 2D vs. 3D SPECT/CT, n.s.). The durations of interpretation of WB + SPECT, 2D SPECT/CT, and 3D SPECT/CT images were 241 ± 75, 225 ± 73, and 182 ± 71 s for reader 1 and 207 ± 72, 190 ± 73, and 179 ± 73 s for reader 2, respectively. As a result, it took shorter time to read 3D SPECT/CT images than 2D SPECT/CT (p < 0.0001) or WB + SPECT images (p < 0.0001). Conclusions: 3D SPECT/CT fusion offers comparable diagnostic accuracy to 2D SPECT/CT fusion. The visual effect of 3D SPECT/CT fusion facilitates reduction of reading time compared to 2D SPECT/CT fusion.",
keywords = "Bone metastasis, Image fusion, SPECT/CT, Three-dimensional, Volume rendering",
author = "Yuji Ogata and Tadaki Nakahara and Kenichi Ode and Yoji Matsusaka and Mari Katagiri and Yu Iwabuchi and Kazunari Itoh and Akira Ichimura and Masahiro Jinzaki",
year = "2017",
month = "2",
day = "27",
doi = "10.1007/s12149-017-1158-3",
language = "English",
pages = "1--11",
journal = "Annals of Nuclear Medicine",
issn = "0914-7187",
publisher = "Springer Japan",

}

TY - JOUR

T1 - 3D SPECT/CT fusion using image data projection of bone SPECT onto 3D volume-rendered CT images

T2 - feasibility and clinical impact in the diagnosis of bone metastasis

AU - Ogata, Yuji

AU - Nakahara, Tadaki

AU - Ode, Kenichi

AU - Matsusaka, Yoji

AU - Katagiri, Mari

AU - Iwabuchi, Yu

AU - Itoh, Kazunari

AU - Ichimura, Akira

AU - Jinzaki, Masahiro

PY - 2017/2/27

Y1 - 2017/2/27

N2 - Purpose: We developed a method of image data projection of bone SPECT into 3D volume-rendered CT images for 3D SPECT/CT fusion. The aims of our study were to evaluate its feasibility and clinical usefulness. Methods: Whole-body bone scintigraphy (WB) and SPECT/CT scans were performed in 318 cancer patients using a dedicated SPECT/CT systems. Volume data of bone SPECT and CT were fused to obtain 2D SPECT/CT images. To generate our 3D SPECT/CT images, colored voxel data of bone SPECT were projected onto the corresponding location of the volume-rendered CT data after a semi-automatic bone extraction. Then, the resultant 3D images were blended with conventional volume-rendered CT images, allowing to grasp the three-dimensional relationship between bone metabolism and anatomy. WB and SPECT (WB + SPECT), 2D SPECT/CT fusion, and 3D SPECT/CT fusion were evaluated by two independent reviewers in the diagnosis of bone metastasis. The inter-observer variability and diagnostic accuracy in these three image sets were investigated using a four-point diagnostic scale. Results: Increased bone metabolism was found in 744 metastatic sites and 1002 benign changes. On a per-lesion basis, inter-observer agreements in the diagnosis of bone metastasis were 0.72 for WB + SPECT, 0.90 for 2D SPECT/CT, and 0.89 for 3D SPECT/CT. Receiver operating characteristic analyses for the diagnostic accuracy of bone metastasis showed that WB + SPECT, 2D SPECT/CT, and 3D SPECT/CT had an area under the curve of 0.800, 0.983, and 0.983 for reader 1, 0.865, 0.992, and 0.993 for reader 2, respectively (WB + SPECT vs. 2D or 3D SPECT/CT, p < 0.001; 2D vs. 3D SPECT/CT, n.s.). The durations of interpretation of WB + SPECT, 2D SPECT/CT, and 3D SPECT/CT images were 241 ± 75, 225 ± 73, and 182 ± 71 s for reader 1 and 207 ± 72, 190 ± 73, and 179 ± 73 s for reader 2, respectively. As a result, it took shorter time to read 3D SPECT/CT images than 2D SPECT/CT (p < 0.0001) or WB + SPECT images (p < 0.0001). Conclusions: 3D SPECT/CT fusion offers comparable diagnostic accuracy to 2D SPECT/CT fusion. The visual effect of 3D SPECT/CT fusion facilitates reduction of reading time compared to 2D SPECT/CT fusion.

AB - Purpose: We developed a method of image data projection of bone SPECT into 3D volume-rendered CT images for 3D SPECT/CT fusion. The aims of our study were to evaluate its feasibility and clinical usefulness. Methods: Whole-body bone scintigraphy (WB) and SPECT/CT scans were performed in 318 cancer patients using a dedicated SPECT/CT systems. Volume data of bone SPECT and CT were fused to obtain 2D SPECT/CT images. To generate our 3D SPECT/CT images, colored voxel data of bone SPECT were projected onto the corresponding location of the volume-rendered CT data after a semi-automatic bone extraction. Then, the resultant 3D images were blended with conventional volume-rendered CT images, allowing to grasp the three-dimensional relationship between bone metabolism and anatomy. WB and SPECT (WB + SPECT), 2D SPECT/CT fusion, and 3D SPECT/CT fusion were evaluated by two independent reviewers in the diagnosis of bone metastasis. The inter-observer variability and diagnostic accuracy in these three image sets were investigated using a four-point diagnostic scale. Results: Increased bone metabolism was found in 744 metastatic sites and 1002 benign changes. On a per-lesion basis, inter-observer agreements in the diagnosis of bone metastasis were 0.72 for WB + SPECT, 0.90 for 2D SPECT/CT, and 0.89 for 3D SPECT/CT. Receiver operating characteristic analyses for the diagnostic accuracy of bone metastasis showed that WB + SPECT, 2D SPECT/CT, and 3D SPECT/CT had an area under the curve of 0.800, 0.983, and 0.983 for reader 1, 0.865, 0.992, and 0.993 for reader 2, respectively (WB + SPECT vs. 2D or 3D SPECT/CT, p < 0.001; 2D vs. 3D SPECT/CT, n.s.). The durations of interpretation of WB + SPECT, 2D SPECT/CT, and 3D SPECT/CT images were 241 ± 75, 225 ± 73, and 182 ± 71 s for reader 1 and 207 ± 72, 190 ± 73, and 179 ± 73 s for reader 2, respectively. As a result, it took shorter time to read 3D SPECT/CT images than 2D SPECT/CT (p < 0.0001) or WB + SPECT images (p < 0.0001). Conclusions: 3D SPECT/CT fusion offers comparable diagnostic accuracy to 2D SPECT/CT fusion. The visual effect of 3D SPECT/CT fusion facilitates reduction of reading time compared to 2D SPECT/CT fusion.

KW - Bone metastasis

KW - Image fusion

KW - SPECT/CT

KW - Three-dimensional

KW - Volume rendering

UR - http://www.scopus.com/inward/record.url?scp=85013998579&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85013998579&partnerID=8YFLogxK

U2 - 10.1007/s12149-017-1158-3

DO - 10.1007/s12149-017-1158-3

M3 - Article

C2 - 28243844

AN - SCOPUS:85013998579

SP - 1

EP - 11

JO - Annals of Nuclear Medicine

JF - Annals of Nuclear Medicine

SN - 0914-7187

ER -