Computer-aided diagnosis based on 3D deep convolutional neural network system using novel 3D magnetic resonance imaging sequences for high-grade prostate cancer

Abstract

Background: With the rising incidence of prostate cancer (PCa), there is a global demand for assistive tools that aid in the diagnosis of high-grade PCa. This study aimed to develop a diagnostic support system for high-grade PCa using innovative magnetic resonance imaging (MRI) sequences in conjunction with artificial intelligence (AI).

Materials and methods: We examined image sequences of 254 patients with PCa obtained from diffusion-weighted and T2-weighted imaging, using novel MRI sequences before prostatectomy, to elucidate the characteristics of the 3-dimensional (3D) image sequences. The presence of PCa was determined based on the final diagnosis derived from pathological results after prostatectomy. A 3D deep convolutional neural network (3DCNN) was used as the AI for image recognition. Data augmentation was conducted to enhance the image dataset. High-grade PCa was defined as Gleason grade group 4 or higher.

Results: We developed a learning system using a 3DCNN as a diagnostic support system for high-grade PCa. The sensitivity and area under the curve values were 85% and 0.82, respectively.

Conclusions: The 3DCNN-based AI diagnostic support system, developed in this study using innovative 3D multiparametric MRI sequences, has the potential to assist in identifying patients at a higher risk of pretreatment of high-grade PCa.

Keywords

Prostate cancer / Multiparametric magnetic resonance imaging / Convolutional neural networks / Artificial intelligence

Acknowledgments

None.

Statement of ethics

The use of prostate MRI data in this study received prior approval from the Ethics Committee at Toho University Sakura Medical Center (approval no. S16085). This study is conducted with the approval of the Ethics Committee of Toho University Medical Center Sakura Hospital. According to the local ethical approval, participants’ consent was not required for this study. The results obtained from this study had no personally identifiable information that could be leaked to outside parties. All procedures performed in this study involving human participants were in accordance with the ethical standards of the institutional and national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

Conflict of interest statement

HS reports research funding from Astellas, AstraZeneca, Bayer, Chugai, Eli-lily, Janssen, MSD, Nihon Kayaku, and Sanofi; advisory fees from Astra Zeneca, Bayer, Chuga-Roche, Eli Lilly, Ferring, Janssen, MSD, Novartis, Pfizer, and Sanofi; and lecture fees from Astellas, AstraZeneca, Bayer, Janssen, Novartis, Pfizer, and Sanofi.

Funding source

None.

Author contributions

RO: conceived the idea of the study, contributed to the interpretation of the results, and drafted the original manuscript;

BL: developed the statistical analysis plan and conducted statistical analyses;

SK: contributed to the interpretation of the results;

TU, TF, NK: contributed to the interpretation of the results;

TN, HS: supervised the conduct of this study.

All authors reviewed the manuscript draft, revised it critically on intellectual content, and approved the final version of the manuscript to be published.

Date availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

References

[1] Scott R, Misser SK, Cioni D, et al. PI-RADS v2.1:What has changed and how to report. SA J Radiol 2021; 25(1):2062.

[2] Loeb S, Bjurlin MA, Nicholson J, et al. Overdiagnosis and overtreatment of prostate cancer. Eur Urol 2014; 65(6):1046-1055.

[3] Ploussard G, Epstein JI, Montironi R, et al. The contemporary concept of significant versus insignificant prostate cancer. Eur Urol 2011; 60(2):291-303.

[4] Epstein JI, Egevad L, Amin MB, et al. The 2014 International Society of Urological Pathology (ISUP) consensus conference on Gleason grading of prostatic carcinoma: Definition of grading patterns and proposal for a new grading system. Am J Surg Pathol 2016; 40(2):244-252.

[5] Lemaître G, Martí R, Freixenet J, et al. Computer-aided detection and diagnosis for prostate cancer based on mono and multi-parametric MRI: A review. Comput Biol Med 2015;60:8-31.

[6] Giger ML, Chan H-P, Boone J. Anniversary paper: History and status of CAD and quantitative image analysis: The role of medical physics and AAPM. Med Phys 2008;35:5799-5820.

[7] Wang X, Yang W, Weinreb J, et al. Searching for prostate cancer by fully automated magnetic resonance imaging classification: Deep learning versus non-deep learning. Sci Rep 2017;7:15415.

[8] Tran Du, Bourdev L, Fergus R, et al. Learning spatiotemporal features with 3D convolutional networks. In: Proceedings of the IEEE International Conference on Computer Vision; December 7-13, 2015; Santiago, Chile.

[9] Kingma DP, Ba J. Adam: A method for stochastic optimization. Published as a conference paper at: 3rd International Conference on Learning Representations; May 7-9, 2015. San Diego, California.

[10] Tamada T, Sone T, Higashi H, et al. Prostate cancer detection in patients with total serum prostate-specific antigen levels of 4-10 ng/mL: Diagnostic efficacy of diffusion-weighted imaging, dynamic contrast-enhanced MRI, and T2-weighted imaging. AJR Am J Roentgenol 2011;197:664-670.

[11] Kitajima K, Kaji Y, Fukabori Y, et al. Prostate cancer detection with 3 T MRI: Comparison of diffusion-weighted imaging and dynamic contrast-enhanced MRI in combination with T2-weighted imaging. J Magn Reson Imaging 2010;31:625-631.

[12] Tanimoto A, Nakashima J, Kohno H, et al. Prostate cancer screening: The clinical value of diffusion-weighted imaging and dynamic MR imaging in combination with T2-weighted imaging. J Magn Reson Imaging 2007;25:146-152.

[13] Sato C, Naganawa S, Nakamura T, et al. Differentiation of noncancerous tissue and cancer lesions by apparent diffusion coefficient values in transition and peripheral zones of the prostate. J Magn Reson Imaging 2005;21:258-262.

[14] Langer DL, van der Kwast TH, Evans AJ, et al. Prostate cancer detection with multi-parametric MRI: Logistic regression analysis of quantitative T2, diffusion-weighted imaging, and dynamic contrast-enhanced MRI. J Magn Reson Imaging 2009;30:327-334.

[15] Lee DH, Koo KC, Lee SH, et al. Low-risk prostate cancer patients without visible tumor (T1c) on multiparametric MRI could qualify for active surveillance candidate even if they did not meet inclusion criteria of active surveillance protocol. Jpn J Clin Oncol 2013;43:553-558.

[16] Weinreb JC, Barentsz JO, Choyke PL, et al. PI-RADS prostate imaging—Reporting and data system: 2015, version 2. Eur Urol 2016;69:16-40.

[17] Le MH, Chen J, Wang L, et al. Automated diagnosis of prostate cancer in multi-parametric MRI based on multimodal convolutional neural networks. Phys Med Biol 2017;62:6497-6514.

[18] Du T, Lubomir B, Rob F, et al. Learning spatiotemporal features of DSA using 3D CNN and BiConvGRU for ischemic moyamoya disease detection. Int J Neurosci 2021;23:1-11.