Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2022 Aug 11:13:884306.
doi: 10.3389/fendo.2022.884306. eCollection 2022.

Deep-learning image reconstruction for image quality evaluation and accurate bone mineral density measurement on quantitative CT: A phantom-patient study

Yali Li  1 , Yaojun Jiang  1 , Xi Yu  1 , Binbin Ren  1 , Chunyu Wang  1 , Sihui Chen  1 , Duoshan Ma  1 , Danyang Su  1 , Huilong Liu  1 , Xiangyang Ren  1 , Xiaopeng Yang  1 , Jianbo Gao  1 , Yan Wu  1
Affiliations

Deep-learning image reconstruction for image quality evaluation and accurate bone mineral density measurement on quantitative CT: A phantom-patient study

Yali Li et al. Front Endocrinol (Lausanne). .

Abstract

Background and purpose: To investigate the image quality and accurate bone mineral density (BMD) on quantitative CT (QCT) for osteoporosis screening by deep-learning image reconstruction (DLIR) based on a multi-phantom and patient study.

Materials and methods: High-contrast spatial resolution, low-contrast detectability, modulation function test (MTF), noise power spectrum (NPS), and image noise were evaluated for physical image quality on Caphan 500 phantom. Three calcium hydroxyapatite (HA) inserts were used for accurate BMD measurement on European Spine Phantom (ESP). CT images were reconstructed with filtered back projection (FBP), adaptive statistical iterative reconstruction-veo 50% (ASiR-V50%), and three levels of DLIR(L/M/H). Subjective evaluation of the image high-contrast spatial resolution and low-contrast detectability were compared visually by qualified radiologists, whilst the statistical difference in the objective evaluation of the image high-contrast spatial resolution and low-contrast detectability, image noise, and relative measurement error were compared using one-way analysis of variance (ANOVA). Cohen's kappa coefficient (k) was performed to determine the interobserver agreement in qualitative evaluation between two radiologists.

Results: Overall, for three levels of DLIR, 50% MTF was about 4.50 (lp/cm), better than FBP (4.12 lp/cm) and ASiR-V50% (4.00 lp/cm); the 2 mm low-contrast object was clearly resolved at a 0.5% contrast level, while 3mm at FBP and ASiR-V50%. As the strength level decreased and radiation dose increased, DLIR at three levels showed a higher NPS peak frequency and lower noise level, leading to leftward and rightward shifts, respectively. Measured L1, L2, and L3 were slightly lower than that of nominal HA inserts (44.8, 95.9, 194.9 versus 50.2, 100.6, 199.2mg/cm3) with a relative measurement error of 9.84%, 4.08%, and 2.60%. Coefficients of variance for the L1, L2, and L3 HA inserts were 1.51%, 1.41%, and 1.18%. DLIR-M and DLIR-H scored significantly better than ASiR-V50% in image noise (4.83 ± 0.34, 4.50 ± 0.50 versus 4.17 ± 0.37), image contrast (4.67 ± 0.73, 4.50 ± 0.70 versus 3.80 ± 0.99), small structure visibility (4.83 ± 0.70, 4.17 ± 0.73 versus 3.83 ± 1.05), image sharpness (3.83 ± 1.12, 3.53 ± 0.90 versus 3.27 ± 1.16), and artifacts (3.83 ± 0.90, 3.42 ± 0.37 versus 3.10 ± 0.83). The CT value, image noise, contrast noise ratio, and image artifacts in DLIR-M and DLIR-H outperformed ASiR-V50% and FBP (P<0.001), whilst it showed no statistically significant between DLIR-L and ASiR-V50% (P>0.05). The prevalence of osteoporosis was 74 (24.67%) in women and 49 (11.79%) in men, whilst the osteoporotic vertebral fracture rate was 26 (8.67%) in women and (5.29%) in men.

Conclusion: Image quality with DLIR was high-qualified without affecting the accuracy of BMD measurement. It has a potential clinical utility in osteoporosis screening.

Keywords: Catphan 500; European Spine Phantom; bone mineral density; deep learning iterative reconstruction; osteoporosis.

PubMed Disclaimer

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Figures

Figure 1
Figure 1
High-contrast images in helical mode reconstructed with FBP (a, f), ASiR-50% (b, g), and DLIR (L/M/H) (c, h; d, i; e, j) at 0.25mGy and 0.75mGy with a slice thickness of 1.25mm (A) and 5mm (B), respectively. CT, computed tomography; FBP, filtered back projection; ASiR-50%, adaptive statistical iterative reconstruction-veo 50%; DLIR(L/M/H), deep-learning image reconstruction, level low, medium, and high; mGy, milligray.
Figure 2
Figure 2
High-contrast images in axial mode reconstructed with FBP (a, f), ASiR-50% (b, g), and DLIR at three levels (L/M/H) (c, h; d, i; e, j) at 0.25mGy and 0.75mGy with a slice thickness of 1.25mm (A) and 5mm (B), respectively. CT, computed tomography; DLIR(L/M/H), deep-learning image reconstruction, level low, medium, and high; mGy, milligray; ASiR-50%, adaptive statistical iterative reconstruction-veo 50%.
Figure 3
Figure 3
MTF curves in helical mode reconstructed with FBP, ASiR-V50%, and DLIR (L/M/H) at 0.25mGy (A) and 0.75mGy (B). CT, computed tomography; FBP, filtered back projection; ASiR-50%, adaptive statistical iterative reconstruction-veo 50%; DLIR(L/M/H), deep-learning image reconstruction, level low, medium, and high; mGy, milligray.
Figure 4
Figure 4
MTF curves in axial mode reconstructed with FBP, ASiR-50%, and DLIR (L/M/H) at 0.25mGy (A) and 0.75mGy (B). CT, computed tomography; DLIR(L/M/H), deep-learning image reconstruction, level low, medium, and high; mGy, milligray; ASiR-50%, adaptive statistical iterative reconstruction-veo 50%.
Figure 5
Figure 5
Low-contrast detectability images in helical mode reconstructed with FBP (a, f), ASiR-50% (b, g), and DLIR(L/M/H) (c, h; d, i; e, j) at 0.25mGy and 0.75mGy with a slice thickness of 1.25mm (A) and 5mm (B), respectively. CT, computed tomography; FBP, filtered back projection; ASiR-50%, adaptive statistical iterative reconstruction-veo 50%; DLIR(L/M/H), deep-learning image reconstruction, level low, medium, and high; mGy, milligray.
Figure 6
Figure 6
Low-contrast detectability images in axial mode reconstructed with FBP (a, f), ASiR-50% (b, g), and DLIR(L/M/H) (c, h; d, i; e, j) at 0.25mGy and 0.75mGy with a slice thickness of 1.25mm (A) and 5mm (B), respectively. CT, computed tomography; DLIR(L/M/H), deep-learning image reconstruction, level low, medium, and high; mGy, milligray; ASiR-50%, adaptive statistical iterative reconstruction-veo 50%.
Figure 7
Figure 7
The curves of NPS and NNPS in helical mode reconstructed with FBP, ASiR-V50%, and DLIR (L/M/H) at 0.25mGy (A, B, E, F) and 0.75mGy (C, D, G, H) with a slice thickness of 1.25mm (A–D) and 5mm (E–H). NPS, noise power spectrum; NNPS, normalized noise power spectrum; HU, Hounsfield units; FBP, filtered back projection; ASiR-50%, adaptive statistical iterative reconstruction-veo 50%; DLIR(L/M/H), deep-learning image reconstruction, level low, medium, and high; mGy, milligray.
Figure 8
Figure 8
The curves of NPS and NNPS in axial mode reconstructed with FBP, ASiR-V50%, and DLIR (L/M/H) at 0.25mGy (A, B, E, F) and 0.75mGy (C, D, G, H) with a slice thickness of 1.25mm (A–D) and 5mm (E–H). NPS, noise power spectrum; NNPS, normalized noise power spectrum; HU, Hounsfield units; FBP, filtered back projection; ASiR-50%, adaptive statistical iterative reconstruction-veo 50%; DLIR(L/M/H), deep-learning image reconstruction, level low, medium, and high; mGy, milligray.
Figure 9
Figure 9
Accuracy deviation of bone mineral density in L1, L2, and L3 with ESP. Error bars standard deviation indicated the relative accuracy error (%) of 3 nominal HA concentrations (ESP, No.145; L1, 50.2; L2, 100.6; L3, 199.2 mg/cm3 HA) for helical (A, B) and axial (C, D) scan type. The relative measurement errors and coefficient of variation of L1, L2, and L3 were fell within the range of 4-15%, indicating no statistically significant differences among L1, L2, and L3 at different scan protocols (P>0.05). ESP, European Spine Phantom; HA, calcium hydroxyapatite; FBP, filtered back projection; ASiR-V50%, adaptive statistical iterative reconstruction-veo 50%; DLIR(L/M/H), deep-learning image reconstruction, level low, medium, and high; mGy, milligray.
Figure 10
Figure 10
Unenhanced CT images of a 67-year-old female for osteoporotic vertebral fracture in the L3 vertebrae. CT images were reconstructed with FBP (A, F), ASiR-V50% (B, G), DLIR-L (C, H), DLIR-M (D, I) and DLIR-H (E, J) with a slice thickness of 1.25mm at 0.75 mGy. The L3 vertebrae body was shown as a severe collapse in sagittal images (arrow), and the vertebral compression appearance was presented in axial images (arrow). The BMD values of FBP, ASiR-V50%, DLIR-L, DLIR-M and DLIR-H were 72.49, 72.74, 71.68, 70.11 and 69.24 mg/cm3 for L1 vertebrae, 67.33, 69.11, 70.25, 65.38, 68.49 mg/cm3 for L2 vertebrae, 62.08, 45.92, 49.57, 52.21, 50.93mg/cm3 for L3 vertebrae, respectively. CT, computed tomography; FBP, filtered back projection; ASiR-V50%, adaptive statistical iterative reconstruction-veo 50%; DLIR(L/M/H), deep-learning image reconstruction, level low, medium, and high; mGy, milligray; BMD, bone mineral density.
See this image and copyright information in PMC

References

    1. Liang C, Zhang G, Wang J, Tang X, Chen H, Yu R. An epidemiological investigation of patients with hip fracture over 50 years old in changning district. Chin J Bone Joint Injury (2013) 28:1122–4. doi: 10.7531/j.issn.1672-9935.2013.12.004 - DOI
    1. Nevitt MC, Ross PD, Palermo L, Musliner T, Genant HK, Thompson DE. Association of prevalent vertebral fractures, bone density, and alendronate treatment with incident vertebral fractures: Effect of number and spinal location of fractures. Fract Intervent Trial Res Group Bone (1999) 25(5):613–9. doi: 10.1016/s8756-3282(99)00202-1 - DOI - PubMed
    1. Patil S, Rawall S, Singh D, Mohan K, Nagad P, Shial B, et al. . Surgical patterns in osteoporotic vertebral compression fractures. Eur Spine J (2013) 22(4):883–91. doi: 10.1007/s00586-012-2508-4 - DOI - PMC - PubMed
    1. Wu Y, Guo Z, Fu X, Wu J, Gao J, Zeng Q, et al. . The study protocol for the China health big data (China biobank) project. Quant Imaging Med Surg (2019) 9(6):1095–102. doi: 10.21037/qims.2019.06.16 - DOI - PMC - PubMed
    1. Wu Y, Jiang Y, Han X, Wang M, Gao J. Application of low-tube current with iterative model reconstruction on philips brilliance iCT elite FHD in the accuracy of spinal QCT using a European spine phantom. Quant Imaging Med Surg (2018) 8(1):32–8. doi: 10.21037/qims.2018.02.03 - DOI - PMC - PubMed

Publication types