Application of deep learning reconstruction techniques in optimizing breast MRI image quality

Share this content in WeChat

• Special Focus •

FAN Wenwen FENG Qianqian LI Erni LIU Kan QUAN Guangnan WANG Peng LU Tongsuo HU Sijie LANG Yu ZHANG Hongmei

Cite this article as: FAN W W, FENG Q Q, LI E N, et al. Application of deep learning reconstruction techniques in optimizing breast MRI image quality[J]. Chin J Magn Reson Imaging, 2024, 15(10): 43-49. DOI:10.12015/issn.1674-8034.2024.10.008.

Abstract

[Abstract] Objective To investigate the impact of deep learning reconstruction (DLR) technology on the quality of breast MRI images and scan time.Materials and Methods A total of 60 patients with a pathological diagnosis of breast cancer at first diagnosis were prospectively enrolled in this study. Conventional fast recovery fast spin echo T2-Weighted imaging, DLR fast fast recovery fast spin echo (FRFSE)-T2WI and conventional short tau inversion recovery-diffusion weighted imaging (STIR-DWI), DLR fast STIR-DWI scanning were performed, respectively. The overall image quality score and artifacts score of two T2WI and DWI (conventional FRFSE-T2WI, DLR fast FRSE-T2WI, and STIR-DWI, DLR fast STIR-DWI) were evaluated subjectively (5-point scale) by two radiologists. One senior radiologist measured the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR). Shapiro-Wilk test was used to evaluate the normal distribution of quantitative values and subjective scores. Wilcoxon signed rank test was used to evaluate the statistical difference of data that did not conform to the normal distribution. The study compared the differences in subjective scores and objective metrics between conventional and DLR-accelerated FRFSE-T2WI scans, as well as conventional STIR-DWI and DLR-accelerated STIR-DWI images. The consistency of researchers' ratings of breast lesion images was quantified using Weighted-Kappa to ensure the reliability of the evaluations.Results A total of 60 patients [25-68 (49.8±8.2) years old] with breast tumors were enrolled in this study. The FRFSE-T2WI scan time was reduced by 47.8% compared to conventional FRFSE-T2WI, and the STIR-DWI scan time was reduced by 47.6% compared to conventional STIR-DWI. The subjective evaluations by two senior physicians reveal that both FRFSE-T2WI and DLR-accelerated FRFSE-T2WI, as well as standard STIR-DWI and DLR-accelerated STIR-DWI, demonstrate significantly superior overall image quality, reduced artifact levels, and enhanced clarity in breast lesion visualization compared to conventional FRFSE-T2-weighted imaging and STIR-DWI. These differences are statistically significant (P<0.05). The SNR for conventional FRFSE-T2WI and DLR-accelerated FRFSE-T2WI were 102.37 (63.24, 141.85) and 132.37 (77.25, 218.62), respectively, with a statistically significant difference (P<0.001). The CNR for lesions were 2.87 (6.35, 57.01) and 3.10 (8.94, 22.34), also showing a statistically significant difference (P<0.001). For conventional STIR-DWI with a b-value of 1000 s/mm² and DLR-accelerated STIR-DWI, the SNRs were 197.34 (157.01, 202.52) and 387.32 (265.06, 464.30), with a statistically significant difference (P<0.001). The CNRs were 1.86 (0.96, 3.23) and 2.22 (1.46, 5.89), also demonstrating a statistically significant difference (P<0.001).Conclusions DL Recon can significantly improve the image quality of rapidly acquired breast MRI sequences while shortening scan time across different modalities. This advancement is beneficial for promoting the clinical application of fast breast MRI sequences.

[Keywords] breast cancer；deep learning；magnetic resonance imaging；signal-to-noise ratio；contrast-to-noise ratio

Contributor Information

FAN Wenwen¹ FENG Qianqian¹ LI Erni¹ LIU Kan¹ QUAN Guangnan² WANG Peng¹ LU Tongsuo¹ HU Sijie¹ LANG Yu¹ ZHANG Hongmei^1*

¹ Department of Radiology, National Cancer Center, National Clinical Research Center for Cancer, Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing 100021, China

² GE Healthcare, Beijing 100176, China

Corresponding author: ZHANG H M, E-mail: 13581968865@163.com

Conflicts of interest None.

Publication Information

Received 2024-06-11

Accepted 2024-10-08

DOI: 10.12015/issn.1674-8034.2024.10.008

Text

References

[1]

XIA C F, DONG X S, LI H, et al. Cancer statistics in China and United States, 2022: profiles, trends, and determinants[J]. Chin Med J, 2022, 135(5): 584-590. DOI: 10.1097/CM9.0000000000002108.

[2]

ZHENG R S, CHEN R, HAN B F, et al. Cancer incidence and mortality in China, 2022[J]. Chin J Oncol, 2024, 46(3): 221-231. DOI: 10.3760/cma.j.cn112152-20240119-00035.

[3]

LEI S Y, ZHENG R S, ZHANG S W, et al. Global patterns of breast cancer incidence and mortality: a population-based cancer registry data analysis from 2000 to 2020[J]. Cancer Commun, 2021, 41(11): 1183-1194. DOI: 10.1002/cac2.12207.

[4]

LI J, CHEN C, NIE J J, et al. Changes in the disease burden of breast cancer along with attributable risk factors in China from 1990 to 2019 and its projections: an analysis of the global burden of disease study 2019[J]. Cancer Med, 2023, 12(2): 1888-1902. DOI: 10.1002/cam4.5006.

[5]

NOLAN E, LINDEMAN G J, VISVADER J E. Deciphering breast cancer: from biology to the clinic[J]. Cell, 2023, 186(8): 1708-1728. DOI: 10.1016/j.cell.2023.01.040.

[6]

HO P J, LIM E H, HARTMAN M, et al. Breast cancer risk stratification using genetic and non-genetic risk assessment tools for 246, 142 women in the UK Biobank[J/OL]. Genet Med, 2023, 25(10): 100917 [2024-06-10]. https://pubmed.ncbi.nlm.nih.gov/37334786/. DOI: 10.1016/j.gim.2023.100917.

[7]

GULLO R L, PARTRIDGE S C, SHIN H J, et al. Update on DWI for breast cancer diagnosis and treatment monitoring[J/OL]. AJR Am J Roentgenol, 2024, 222(1): e2329933 [2024-06-10]. https://pubmed.ncbi.nlm.nih.gov/37850579/. DOI: 10.2214/AJR.23.29933.

[8]

LOTHER D, ROBERT M, ELWOOD E, et al. Imaging in metastatic breast cancer, CT, PET/CT, MRI, WB-DWI, CCA: review and new perspectives[J/OL]. Cancer Imaging, 2023, 23(1): 53 [2024-06-10]. https://pubmed.ncbi.nlm.nih.gov/37254225/. DOI: 10.1186/s40644-023-00557-8.

[9]

YU J P, DU S Y, HAN R, et al. Application value of IDEAL-IQ sequence in differential diagnosis of benign and malignant breast masses[J]. Chin J Magn Reson Imag, 2024, 15(1): 14-20, 42. DOI: 10.12015/issn.1674-8034.2024.01.003.

[10]

BERG W A. Breast MRI for "the masses"[J]. Eur Radiol, 2022, 32(6): 4034-4035. DOI: 10.1007/s00330-022-08782-8.

[11]

LIU L S, LIU P F, MA W J, et al. Correlation analysis of MRI features of breast and feasibility of breast conserving surgery[J]. Chin J Magn Reson Imag, 2024, 15(1): 43-47, 60. DOI: 10.12015/issn.1674-8034.2024.01.007.

[12]

LEI K, SYED A B, ZHU X C, et al. Artifact- and content-specific quality assessment for MRI with image rulers[J/OL]. Med Image Anal, 2022, 77: 102344 [2024-06-10]. https://pubmed.ncbi.nlm.nih.gov/35091278/. DOI: 10.1016/j.media.2021.102344.

[13]

HASKELL M W, NIELSEN J F, NOLL D C. Off-resonance artifact correction for MRI: a review[J/OL]. NMR Biomed, 2023, 36(5): e4867 [2024-06-10]. https://pubmed.ncbi.nlm.nih.gov/36326709/. DOI: 10.1002/nbm.4867.

[14]

WU H F, CHEN X Z, ZHANG M Y, et al. Application of head enhanced T1WI sequences based on deep learning reconstruction technology in the transformation of pituitary neuroendocrine neoplasms[J]. Chin J Magn Reson Imag, 2024, 15(4): 133-138. DOI: 10.12015/issn.1674-8034.2024.04.021.

[15]

JOHNSON P M, LIN D J, ZBONTAR J, et al. Deep learning reconstruction enables prospectively accelerated clinical knee MRI[J/OL]. Radiology, 2023, 307(2): e220425 [2024-06-10]. https://pubmed.ncbi.nlm.nih.gov/36648347/. DOI: 10.1148/radiol.220425.

[16]

OSCANOA J A, MIDDIONE M J, ALKAN C, et al. Deep learning-based reconstruction for cardiac MRI: a review[J/OL]. Bioengineering, 2023, 10(3): 334 [2024-06-10]. https://pubmed.ncbi.nlm.nih.gov/36978725/. DOI: 10.3390/bioengineering10030334.

[17]

LIU H Y, LIU J Q, LI J B, et al. DL-MRI: a unified framework of deep learning-based MRI super resolution[J/OL]. J Healthc Eng, 2021, 2021: 5594649 [2024-06-10]. https://pubmed.ncbi.nlm.nih.gov/33897991/. DOI: 10.1155/2021/5594649.

[18]

NJEH I, MZOUGHI H, SLIMA M BEN, et al. Deep Convolutional Encoder-Decoder algorithm for MRI brain reconstruction[J]. Med Biol Eng Comput, 2021, 59(1): 85-106. DOI: 10.1007/s11517-020-02285-8.

[19]

EDUPUGANTI V, MARDANI M, VASANAWALA S, et al. Uncertainty quantification in deep MRI reconstruction[J]. IEEE Trans Med Imaging, 2021, 40(1): 239-250. DOI: 10.1109/TMI.2020.3025065.

[20]

ZHANG X X, WANG Y C, WANG S C, et al. Feasibility study of deep learning reconstruction in the clinical application of MRI in bladder cancer[J]. Chin J Magn Reson Imag, 2023, 14(5): 36-40. DOI: 10.12015/issn.1674-8034.2023.05.008.

[21]

JANNUSCH K, LINDEMANN M E, BRUCKMANN N M, et al. Towards a fast PET/MRI protocol for breast cancer imaging: maintaining diagnostic confidence while reducing PET and MRI acquisition times[J]. Eur Radiol, 2023, 33(9): 6179-6188. DOI: 10.1007/s00330-023-09580-6.

[22]

LIU Z J, WEN B H, WANG Z Y, et al. Deep learning-based reconstruction enhances image quality and improves diagnosis in magnetic resonance imaging of the shoulder joint[J]. Quant Imaging Med Surg, 2024, 14(4): 2840-2856. DOI: 10.21037/qims-23-1412.

[23]

ALTMANN S, GRAUHAN N F, BROCKSTEDT L, et al. Ultrafast brain MRI with deep learning reconstruction for suspected acute ischemic stroke[J/OL]. Radiology, 2024, 310(2): e231938 [2024-06-10]. https://pubmed.ncbi.nlm.nih.gov/38376403/. DOI: 10.1148/radiol.231938.

[24]

SINGH D, MONGA A, DE MOURA H L, et al. Emerging trends in fast MRI using deep-learning reconstruction on undersampled k-space data: a systematic review[J/OL]. Bioengineering, 2023, 10(9): 1012 [2024-06-10]. https://pubmed.ncbi.nlm.nih.gov/37760114/. DOI: 10.3390/bioengineering10091012.

[25]

WANG K, TAMIR J I, DE GOYENECHE A, et al. High fidelity deep learning-based MRI reconstruction with instance-wise discriminative feature matching loss[J]. Magn Reson Med, 2022, 88(1): 476-491. DOI: 10.1002/mrm.29227.

[26]

DEMIREL O B, ZHANG C, YAMAN B, et al. High-fidelity database-free deep learning reconstruction for real-time cine cardiac MRI[J/OL]. Annu Int Conf IEEE Eng Med Biol Soc, 2023, 2023: 1-4 [2024-06-10]. https://pubmed.ncbi.nlm.nih.gov/38083374/. DOI: 10.1109/EMBC40787.2023.10340709.

[27]

RIEDERER S J, BORISCH E A, FROEMMING A T, et al. Comparison of model-based versus deep learning-based image reconstruction for thin-slice T2-weighted spin-echo prostate MRI[J]. Abdom Radiol (NY), 2024, 49(8): 2921-2931. DOI: 10.1007/s00261-024-04256-1.

[28]

LI K J, ZHANG J T, REN K X, et al. The value of machine learning model for predicting prostate cancer bone metastases based on MRI radiomics[J]. Chin J Magn Reson Imag, 2023, 14(1): 100-104, 115. DOI: 10.12015/issn.1674-8034.2023.01.018.

[29]

MANGANARO L, CIULLA S, CELLI V, et al. Impact of DWI and ADC values in Ovarian-Adnexal Reporting and Data System (O-RADS) MRI score[J]. Radiol Med, 2023, 128(5): 565-577. DOI: 10.1007/s11547-023-01628-3.

[30]

VAN LOHUIZEN Q, ROEST C, SIMONIS F F J, et al. Assessing deep learning reconstruction for faster prostate MRI: visual vs. diagnostic performance metrics[J/OL]. Eur Radiol, 2024 [2024-06-10]. https://pubmed.ncbi.nlm.nih.gov/38724765/. DOI: 10.1007/s00330-024-10771-y.

[31]

SHANBHOGUE K, TONG A, SMEREKA P, et al. Accelerated single-shot T2-weighted fat-suppressed (FS) MRI of the liver with deep learning-based image reconstruction: qualitative and quantitative comparison of image quality with conventional T2-weighted FS sequence[J]. Eur Radiol, 2021, 31(11): 8447-8457. DOI: 10.1007/s00330-021-08008-3.

[32]

WILPERT C, NEUBAUER C, RAU A, et al. Accelerated diffusion-weighted imaging in 3 T breast MRI using a deep learning reconstruction algorithm with superresolution processing: a prospective comparative study[J]. Invest Radiol, 2023, 58(12): 842-852. DOI: 10.1097/RLI.0000000000000997.

[33]

ESTLER A, HAUSER T K, BRUNNÉE M, et al. Deep learning-accelerated image reconstruction in back pain-MRI imaging: reduction of acquisition time and improvement of image quality[J]. Radiol Med, 2024, 129(3): 478-487. DOI: 10.1007/s11547-024-01787-x.

[34]

YANG S, BIE Y F, PANG G D, et al. Impact of novel deep learning image reconstruction algorithm on diagnosis of contrast-enhanced liver computed tomography imaging: comparing to adaptive statistical iterative reconstruction algorithm[J]. J Xray Sci Technol, 2021, 29(6): 1009-1018. DOI: 10.3233/XST-210953.

[35]

KIM J H, YOON J H, KIM S W, et al. Application of a deep learning algorithm for three-dimensional T1-weighted gradient-echo imaging of gadoxetic acid-enhanced MRI in patients at a high risk of hepatocellular carcinoma[J]. Abdom Radiol, 2024, 49(3): 738-747. DOI: 10.1007/s00261-023-04124-4.

[36]

SAUER S T, CHRISTNER S A, LOIS A M, et al. Deep learning k-space-to-image reconstruction facilitates high spatial resolution and scan time reduction in diffusion-weighted imaging breast MRI[J]. J Magn Reson Imaging, 2024, 60(3): 1190-1200. DOI: 10.1002/jmri.29139.

[37]

BAE S H, HWANG J, HONG S S, et al. Clinical feasibility of accelerated diffusion weighted imaging of the abdomen with deep learning reconstruction: comparison with conventional diffusion weighted imaging[J/OL]. Eur J Radiol, 2022, 154: 110428 [2024-06-10]. https://pubmed.ncbi.nlm.nih.gov/35797791/. DOI: 10.1016/j.ejrad.2022.110428.

[38]

ZERUNIAN M, PUCCIARELLI F, CARUSO D, et al. Artificial intelligence based image quality enhancement in liver MRI: a quantitative and qualitative evaluation[J]. Radiol Med, 2022, 127(10): 1098-1105. DOI: 10.1007/s11547-022-01539-9.

[39]

KE Z, LI L, SONG X Y, et al. Research on Improving Prostate T2WI Image Quality Based on Deep Learning Reconstruction Technology[J]. Chin J Magn Reson Imag, 2023, 14(5): 41-47. DOI: 10.12015/issn.1674-8034.2023.05.009.

[40]

ENSLE F, KANIEWSKA M, TIESSEN A, et al. Diagnostic performance of deep learning-based reconstruction algorithm in 3D MR neurography[J]. Skeletal Radiol, 2023, 52(12): 2409-2418. DOI: 10.1007/s00256-023-04362-z.

[41]

VAN DER VELDE N, HASSING H C, BAKKER B J, et al. Improvement of late gadolinium enhancement image quality using a deep learning-based reconstruction algorithm and its influence on myocardial scar quantification[J]. Eur Radiol, 2021, 31(6): 3846-3855. DOI: 10.1007/s00330-020-07461-w.

PREV Deep learning-based reconstruction of diffusion-weighted imaging images to assess the activity of thyroid-associated ophthalmopathy

NEXT Application value of high-resolution single-shot fast spin-echo ovarian MRI based on deep learning reconstruction in follicle counting

LINK

Tel & Fax: +8610-67113815 E-mail: editor@cjmri.cn