Abstract
Objectives
To build an artificial intelligence (AI) system to classify benign and malignant non-mass enhancement (NME) lesions using maximum intensity projection (MIP) of early post-contrast subtracted breast MR images.
Methods
This retrospective study collected 965 pure NME lesions (539 benign and 426 malignant) confirmed by histopathology or follow-up in 903 women. The 754 NME lesions acquired by one MR scanner were randomly split into the training set, validation set, and test set A (482/121/151 lesions). The 211 NME lesions acquired by another MR scanner were used as test set B. The AI system was developed using ResNet-50 with the axial and sagittal MIP images. One senior and one junior radiologist reviewed the MIP images of each case independently and rated its Breast Imaging Reporting and Data System category. The performance of the AI system and the radiologists was evaluated using the area under the receiver operating characteristic curve (AUC).
Results
The AI system yielded AUCs of 0.859 and 0.816 in the test sets A and B, respectively. The AI system achieved comparable performance as the senior radiologist (p = 0.558, p = 0.041) and outperformed the junior radiologist (p < 0.001, p = 0.009) in both test sets A and B. After AI assistance, the AUC of the junior radiologist increased from 0.740 to 0.862 in test set A (p < 0.001) and from 0.732 to 0.843 in test set B (p < 0.001).
Conclusion
Our MIP-based AI system yielded good applicability in classifying NME lesions in breast MRI and can assist the junior radiologist achieve better performance.
Key Points
• Our MIP-based AI system yielded good applicability in the dataset both from the same and a different MR scanner in predicting malignant NME lesions.
• The AI system achieved comparable diagnostic performance with the senior radiologist and outperformed the junior radiologist.
• This AI system can assist the junior radiologist achieve better performance in the classification of NME lesions in MRI.
Similar content being viewed by others
Abbreviations
- AI:
-
Artificial intelligence
- AUC:
-
Area under the receiver operating characteristic curve
- BI-RADS:
-
Breast Imaging Reporting and Data System
- BPE:
-
Background parenchymal enhancement
- CI:
-
Confidence interval
- CNB:
-
Core needle biopsy
- MIP:
-
Maximum intensity projection
- NME:
-
Non-mass enhancement
- NPV:
-
Negative predictive value
- PPV:
-
Positive predictive value
References
Morris EA, Comstock CE, Lee CH et al (2013) ACR BI-RADS® magnetic resonance imaging. In: ACR BI-RADS® atlas, Breast Imaging Reporting and Data System. American College of Radiology, Reston, VA
Thomassin-Naggara I, Trop I, Chopier J et al (2011) Nonmasslike enhancement at breast MR imaging: the added value of mammography and US for lesion categorization. Radiology 261:69–79
Chikarmane SA, Michaels AY, Giess CS (2017) Revisiting nonmass enhancement in breast MRI: analysis of outcomes and follow-up using the updated BI-RADS Atlas. AJR Am J Roentgenol 209:1178–1184
Uematsu T, Kasami M (2012) High-spatial-resolution 3-T breast MRI of nonmasslike enhancement lesions: an analysis of their features as significant predictors of malignancy. AJR Am J Roentgenol 198:1223–1230
Shimauchi A, Ota H, Machida Y et al (2016) Morphology evaluation of nonmass enhancement on breast MRI: effect of a three-step interpretation model for readers’ performances and biopsy recommendations. Eur J Radiol 85:480–488
Avendano D, Marino MA, Leithner D et al (2019) Limited role of DWI with apparent diffusion coefficient mapping in breast lesions presenting as non-mass enhancement on dynamic contrast-enhanced MRI. Breast Cancer Res 21:136
Kul S, Eyuboglu I, Cansu A, Alhan E (2014) Diagnostic efficacy of the diffusion weighted imaging in the characterization of different types of breast lesions. J Magn Reson Imaging 40:1158–1164
Lunkiewicz M, Forte S, Freiwald B, Singer G, Leo C, Kubik-Huch RA (2020) Interobserver variability and likelihood of malignancy for fifth edition BI-RADS MRI descriptors in non-mass breast lesions. Eur Radiol 30:77–86
Machida Y, Tozaki M, Shimauchi A, Yoshida T (2015) Two distinct types of linear distribution in nonmass enhancement at breast MR imaging: difference in positive predictive value between linear and branching patterns. Radiology 276:686–694
Ou WC, Polat D, Dogan BE (2021) Deep learning in breast radiology: current progress and future directions. Eur Radiol 31:4872–4885
Mendelson EB (2019) Artificial intelligence in breast imaging: potentials and limitations. AJR Am J Roentgenol 212:293–299
Meyer-Base A, Morra L, Tahmassebi A, Lobbes M, Meyer-Base U, Pinker K (2021) AI-enhanced diagnosis of challenging lesions in breast MRI: a methodology and application primer. J Magn Reson Imaging 54(3):686–702
Gallego-Ortiz C, Martel AL (2019) A graph-based lesion characterization and deep embedding approach for improved computer-aided diagnosis of nonmass breast MRI lesions. Med Image Anal 51:116–124
Youn I, Choi S, Choi YJ et al (2019) Contrast enhanced digital mammography versus magnetic resonance imaging for accurate measurement of the size of breast cancer. Br J Radiol 92:20180929
Kuhl CK, Schrading S, Strobel K, Schild HH, Hilgers RD, Bieling HB (2014) Abbreviated breast magnetic resonance imaging (MRI): first postcontrast subtracted images and maximum-intensity projection-a novel approach to breast cancer screening with MRI. J Clin Oncol 32:2304–2310
Choi WJ, Cha JH, Kim HH, Shin HJ, Chae EY (2016) The accuracy of breast MR imaging for measuring the size of a breast cancer: analysis of the histopathologic factors. Clin Breast Cancer 16:e145–e152
Liao GJ, Henze Bancroft LC, Strigel RM et al (2020) Background parenchymal enhancement on breast MRI: a comprehensive review. J Magn Reson Imaging 51:43–61
Han M, Kim TH, Kang DK, Kim KS, Yim H (2012) Prognostic role of MRI enhancement features in patients with breast cancer: value of adjacent vessel sign and increased ipsilateral whole-breast vascularity. AJR Am J Roentgenol 199:921–928
Kul S, Cansu A, Alhan E, Dinc H, Reis A, Çan G (2010) Contrast-enhanced MR angiography of the breast: evaluation of ipsilateral increased vascularity and adjacent vessel sign in the characterization of breast lesions. AJR Am J Roentgenol 195:1250–1254
Antropova N, Abe H, Giger ML (2018) Use of clinical MRI maximum intensity projections for improved breast lesion classification with deep convolutional neural networks. J Med Imaging (Bellingham) 5:014503
Hu Q, Whitney HM, Li H, Yu J, Liu P, Giger ML (2021) Improved classification of benign and malignant breast lesions using deep feature maximum intensity projection MRI in breast cancer diagnosis using dynamic contrast-enhanced MRI. Radiol Artif Intell 3(3):e200159
Adachi M, Fujioka T, Mori M et al (2020) Detection and diagnosis of breast cancer using artificial intelligence based assessment of maximum intensity projection dynamic contrast-enhanced magnetic resonance images. Diagnostics (Basel) 10:330
Fujioka T, Yashima Y, Oyama J et al (2021) Deep-learning approach with convolutional neural network for classification of maximum intensity projections of dynamic contrast-enhanced breast magnetic resonance imaging. Magn Reson Imaging 75:1–8
Russakovsky O, Deng J, Su H et al (2015) ImageNet large scale visual recognition challenge. Int J Comput Vis 115:211–252
Wang L, Harz M, Boehler T, Platel B, Homeyer A, Hahn HK (2014) A robust and extendable framework towards fully automated diagnosis of non-mass lesions in breast DCE-MRI. ISBI 1:129–132
Wang L, Wang D, Fei X et al (2014) A rim-enhanced mass with central cystic changes on MR imaging: how to distinguish breast cancer from inflammatory breast diseases? PLoS One 9:e90355
Tozaki M, Igarashi T, Fukuda K (2006) Breast MRI using the VIBE sequence: clustered ring enhancement in the differential diagnosis of lesions showing non-masslike enhancement. AJR Am J Roentgenol 187:313–321
Acknowledgements
The authors would like to acknowledge the contribution of Wu Xinyang and Zhou Shanshan for their valuable assistance and coordination with this project.
Funding
This study has received funding from the Special Research Program of the Shanghai Municipal Commission of Heath and Family Planning on medical intelligence (2018ZHYL0108), the Doctoral Innovation Fund of Shanghai Jiao Tong University School of Medicine (CBXJ201807), and the Program of Shanghai Science and Technology Committee (No. 21S31905000). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
This retrospective study was approved by the institutional review board of our hospital, and the need to obtain informed consent was waived (approval #, XHEC-D-2020-104). This study was in accordance with the Declaration of Helsinki.
Guarantor
The scientific guarantor of this publication is Dengbin Wang, MD, PhD, the chief of the Department of Radiology, Xinhua Hospital affiliated to Shanghai Jiao Tong University School of Medicine.
Conflict of interest
All the authors in this research declare no potential conflicts of interest on the work. Two authors of this research, Lufan Chang and Hao Liu, work for a medical company (Yizhun Medical AI Co. Ltd., Beijing, China). No disclosures of potential conflicts of Yizhun Medical AI Co. Ltd., Beijing, China, and no other potential conflict of interest relevant to this article are reported.
Statistics and biometry
No complex statistical methods were necessary for this paper.
Informed consent
Written informed consent was waived by the institutional review board.
Ethical approval
Institutional Review Board approval was obtained.
Study subjects or cohorts overlap
Eleven patients included in this study overlap with a prior study. The prior study focused on describing the MRI features of papillary breast lesions. This study focused on developing a deep learning model for the classification of NME lesions. We have uploaded the PDF of this study in the online submission system.
Methodology
• retrospective
• diagnostic or prognostic study
• performed at one institution
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
ESM 1
(DOCX 758 kb)
Rights and permissions
About this article
Cite this article
Wang, L., Chang, L., Luo, R. et al. An artificial intelligence system using maximum intensity projection MR images facilitates classification of non-mass enhancement breast lesions. Eur Radiol 32, 4857–4867 (2022). https://doi.org/10.1007/s00330-022-08553-5
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00330-022-08553-5