Detection of prostate cancer using temporal sequences of ultrasound data: a large clinical feasibility study
- 663 Downloads
This paper presents the results of a large study involving fusion prostate biopsies to demonstrate that temporal ultrasound can be used to accurately classify tissue labels identified in multi-parametric magnetic resonance imaging (mp-MRI) as suspicious for cancer.
We use deep learning to analyze temporal ultrasound data obtained from 255 cancer foci identified in mp-MRI. Each target is sampled in axial and sagittal planes. A deep belief network is trained to automatically learn the high-level latent features of temporal ultrasound data. A support vector machine classifier is then applied to differentiate cancerous versus benign tissue, verified by histopathology. Data from 32 targets are used for the training, while the remaining 223 targets are used for testing.
Our results indicate that the distance between the biopsy target and the prostate boundary, and the agreement between axial and sagittal histopathology of each target impact the classification accuracy. In 84 test cores that are 5 mm or farther to the prostate boundary, and have consistent pathology outcomes in axial and sagittal biopsy planes, we achieve an area under the curve of 0.80. In contrast, all of these targets were labeled as moderately suspicious in mp-MR.
Using temporal ultrasound data in a fusion prostate biopsy study, we achieved a high classification accuracy specifically for moderately scored mp-MRI targets. These targets are clinically common and contribute to the high false-positive rates associated with mp-MRI for prostate cancer detection. Temporal ultrasound data combined with mp-MRI have the potential to reduce the number of unnecessary biopsies in fusion biopsy settings.
KeywordsTemporal ultrasound data Deep learning Deep belief network Cancer diagnosis Prostate cancer
This work was supported in part by the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Canadian Institutes of Health Research (CIHR).
Compliance with ethical standards
Conflict of interest
The authors declare that they have no conflict of interest.
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.
- 1.Azizi S, Imani F, Zhuang B, Tahmasebi A, Kwak JT, Xu S, Uniyal N, Turkbey B, Choyke P, Pinto P, Wood B (2015) Ultrasound-based detection of prostate cancer using automatic feature selection with deep belief networks. Med Image Comput Comput-Assist Interv—MICCAI 2015, pp 70–77. Springer (2015)Google Scholar
- 2.Bengio Y, Lamblin P, Popovici D, Larochelle H (2007) Greedy layer-wise training of deep networks. Adv Neural Inf Process Syst 19:153Google Scholar
- 5.Epstein JI, Feng Z, Trock BJ, Pierorazio PM (2012) Upgrading and downgrading of prostate cancer from biopsy to radical prostatectomy: incidence and predictive factors using the modified Gleason grading system and factoring in tertiary grades. Eur Urol 61(5):1019–1024CrossRefPubMedPubMedCentralGoogle Scholar
- 6.Erhan D, Bengio Y, Courville A, Manzagol PA, Vincent P, Bengio S (2010) Why does unsupervised pre-training help deep learning? J Mach Learn Res 11:625–660Google Scholar
- 8.Feleppa, E., Porter, C., Ketterling, J., Dasgupta, S., Ramachandran, S., Sparks, D.: Recent advances in ultrasonic tissue-type imaging of the prostate. In: Acoustical imaging, pp 331–339. Springer (2007)Google Scholar
- 9.Goossen T, Wijkstra H (2003) Transrectal ultrasound imaging and prostate cancer. Archivio Italiano di Urologia Andrologia 75(1):68–74Google Scholar
- 10.Hinton G (2010) A practical guide to training restricted Boltzmann machines. Momentum 9(1):926Google Scholar
- 12.Imani F, Abolmaesumi P, Gibson E, Khojaste A, Gaed M, Moussa M, Gomez JA, Romagnoli C, Siemens DR, Leviridge M, Chang S, Fenster A, Ward AD, Mousavi P (2013) Ultrasound-based characterization of prostate cancer: an in vivo clinical feasibility study. Med Image Comput Comput-Assist Interv–MICCAI 2013, pp 279–286, SpringerGoogle Scholar
- 15.Imani, F., Zhuang, B., Tahmasebi, A., Kwak, J.T., Xu S, Agarwal H, Bharat S, Uniyal N, Turkbey I, Choyke P, Pinto P (2015) Augmenting MRI-transrectal ultrasound-guided prostate biopsy with temporal ultrasound data: a clinical feasibility study. Int J Comput Assist Radiol Surg, pp 1–9 (2015)Google Scholar
- 16.Kuru TH, Roethke MC, Seidenader J, Simpfendörfer T, Boxler S, Alammar K, Rieker P, Popeneciu V, Roth W, Pahernik S, Schlemmer H (2013) Critical evaluation of magnetic resonance imaging targeted, transrectal ultrasound guided transperineal fusion biopsy for detection of prostate cancer. J Urol 190(4):1380–1386CrossRefPubMedGoogle Scholar
- 20.Moradi, M., Mahdavi, S.S., Nir, G., Jones, E.C., Goldenberg, S.L., Salcudean, S.E.: Ultrasound RF time series for tissue typing: first in vivo clinical results. In: SPIE Medical Imaging, pp. 86,701I–86,701I. International Society for Optics and Photonics (2013)Google Scholar
- 21.Moradi M, Mahdavi SS, Nir G, Mohareri O, Koupparis A, Gagnon L, Fazli L, Casey R, Ischia J, Jones E, Goldenberg S (2014) Multiparametric 3D in vivo ultrasound vibroelastography imaging of prostate cancer: Preliminary results. Med Phys 41(7)Google Scholar
- 28.Tanaka M, Okutomi M (2014) A novel inference of a restricted Boltzmann machine. In: International conference on pattern recognition (ICPR), 2014 22nd, pp 1526–1531. IEEE (2014)Google Scholar
- 29.Turkbey B, Mani H, Aras O, Ho J, Hoang A, Rastinehad A, Agarwal H, Shah V, Bernardo M, Pang Y, Daar D (2013) Prostate cancer: Can mp-MR imaging help identify patients who are candidates for active surveillance? Radiology 268(1):144–152Google Scholar
- 32.Yosinski J, Lipson H (2012) Visually debugging restricted Boltzmann machine training with a 3D example. In: 29th international conference on machine learning, representation learning workshop 2012Google Scholar