5055
AIR recon deep learning on oZTEo technique in cortical bone MRI: an ex vivo ankle joint of adolescent pig and a knee joint of adolescent child1West China Second University Hospital, Sichuan University, Chengdu, China, 2GE HealthCare MR Research, Beijing, China
Synopsis
Keywords: MSK, MSK, oZTEo, DL
Motivation: Exploring oZTEo sequence with NEX=2 after AIR recon deep learning (DL) alternatives to 3D-ZTE for diagnosing bone diseases in children is clinically desirable.
Goal(s): To compare the image quality among different kinds of oZTEo sequences.
Approach: A specimen and a child were subjected to different oZTEo MRI with and witout DL. The specimen was subjected to CT examination. The thicknesses of bone cortex and SNR of neighboring muscles were measured.
Results: The image quality of oZTEo CT-like with NEX=2 after AIR recon DL was superior than that with NEX=6 before AIR recon DL.
Impact: The oZTEo technique with NEX=2 after AIR recon DL, with shorter scanning time and witout ionizing radiation, is a safer choice for diagnosing children bone diseases.
Inroduction
Compared to routine MRI sequences, zero echo time (3D-ZTE) MR sequence has been proved to be capable of capturing weak bone signals and show good agreement with CT images [1-7]. However, the scanning time is 3min37s and a little long for children. Another similar sequence oZTEo can decrease the time by decreasing NEX and AIR recon deep learning (DL) can make sure the image quality to meet the diagnosis [2]. Therefore, the objective of this study was to compare the image quality among different NEXs in oZTEo sequence before and after AIR recon DL on both adolescent pig and child. For the ex vivo pig, CT examination was performed to compare with MRI.
Methods
One specimen of ankle joint of adolescent pig was subjected to both oZTEo MRI and CT examination. oZTEo acquisition was performed on a 3.0T MR scanner (SIGNATM Premier, GE Healthcare Systems, Milwaukee, WI, USA) equipped with a 16-channel flex coil. The main parameters as follows: FOV=20mm2, voxel size=0.8x0.8x0.8mm3, TR=385.5ms, TE=0, FA=1, bandwidth=83.33, total scan time=3min33s while NEX=6 and that was 1min13s while NEX=2. AIR recon DL were performed on the two kinds of images. Black and white inversion post-processing was conducted in AW4.7 workstation to obtain CT-like images. Then the thickness of bone cortex were measured in inner sides of tibia and from the coronal plane (figure 2). The region of interest (ROI) was set to neighboring muscles measure and calculate signal noise raito (SNR) (figure 2). CT acquisition was performed on a 256-row CT scanner (Revolution CT, GE Healthcare, Waukesha, WI, USA) with main parameters including: tube voltage using the fast switching mode between 80 kVp and 140kVp, with tube current 275mA to obtain a reset noise index (NI) of 10 and a pitch of 0.992:1, with scanning thickness 5mm and reconstructed to 0.8mm, total scan time=5.8s. Coronal plane images of bone window were obtained also in AW4.7 workstation to measure the thicknesses of bone cortices and SNR on neighboring muscles in aforementioned locations (figure 3). With institutional review board approval and informed consent, one knee joint of adolescent child was subjected to oZTEo MRI examination. The methods of scanning and post-processing were the same as the specimen. The thickness of bone cortex were measured in lower femur and SNR on neighboring muscles was also measured and calculate with setting ROI (figure 4).Results
(1) For the specimen, the thickness of bone cortex in inner side of tibia were 3.1mm and 3.3mm in oZTEo CT-like images with NEX=2 and 6, those were 3.6mm and 3.6mm after AIR recon DL respectively, that was 3.2mm in CT image. The SNR on neighboring muscles were 62.24, 109.64, 122.97, 158.09 and 14.90 respectively (figure 5). (2) For the child, the thickness of bone cortex in lower femur were 1.6mm and 1.6mm in oZTEo CT-like images with NEX=2 and 6, those were 1.7mm and 1.7mm after AIR recon DL respectively. The SNR on neighboring muscles were 6.94, 9.47, 20.78 and 38.27 respectively.Discussion
For the specimen, the thicknesses of bone cortex was found to be consistent among the four kinds of oZTEo CT-like and CT images. The SNR on neighboring muscles in oZTEo CT-like images with NEX=2 after AIR recon DL was superior to that with NEX=6 before AIR recon DL. For the adolescent child, similar results were obtained. Therefore, oZTEo CT-like images with NEX=2 after AIR recon DL could apply to clinical practice. The scanning time was only 1min13s and suitable for adolescent children.Conclusions
The CT-like images obtained from oZTEo technique with NEX=2 after AIR recon DL can be used to adolescent children for diagnosing bone diseases, without too long scanning time and without ionizing radiation.Acknowledgements
No acknowledgement found.References
1. Fujisaki A, Tsukamoto J, Narimatsu H, et al. Zero echo time magnetic resonance imaging: techniques and clinical utility in musculoskeletal system. JMRI, 2023, online.
2. Sneag DB, Albe F, Potter HG, et al. MRI advancements in musculoskeletal clinical and research practice. Radiology, 2023, 308(2): e230531.
3. Sandberg JK, Young VA, Yuan JM, et al. Zero echo time pediatric musculoskeletal magnetic resonance imaging: initial experience. Pediatric radiology, 2021, 51(13): 2549-2560.
4. Bharadwaj UU, Coy A, Motamedi D, et al. CT‑like MRI: a qualitative assessment of ZTE sequences for knee osseous abnormalities. Skeletal radiology, 2022, 51(8): 1585-1594.
5. de Mello RAF, Ma YJ, Ashir A, et al. 3D Zero Echo Time (ZTE) MRI versus 3D CT for Glenoid Bone Assessment. Arthroscopy, 2020, 36(9): 2391–2400.
6. Aydıngöz Ü, Yildiz AE, Ergen FB. Zero echo time musculoskeletal MRI: technique, optimization, applications, and pitfalls. Radiographics, 2022, 42(5): 1398-1414.
7. Sandberg JK, Young VA, Yuan JM, et al. Zero echo time pediatric musculoskeletal magnetic resonance imaging: initial experience. Pediatr Radiol, 2021, 51(3): 2549-2560.
Figures
