Transfer Learning for Breast Cancer Malignancy Classification based on Dynamic Contrast-Enhanced MR Images
In clinical contexts with very limited annotated data, such as breast cancer diagnosis, training state-of-the art deep neural networks is not feasible. As a solution, we transfer parameters of networks pretrained on natural RGB images to malignancy classification of breast lesions in dynamic contrast-enhanced MR images. Since DCE-MR images comprise several contrasts and timepoints, a direct finetuning of pretrained networks expecting three input channels is not possible. Based on the hypothesis that a subset of the acquired image data is sufficient for a computer-aided diagnosis, we provide an experimental comparison of all possible subsets of MR image contrasts and determine the best combination for malignancy classification. A subset of images acquired at three timepoints of dynamic T1-weighted images which closely corresponds to human interpretation performs best with an AUC of 0.839.
Unable to display preview. Download preview PDF.
- 1.Siegel R, Ma J, Zou Z, et al. Cancer statistics, 2014. CA Cancer J Clin. 2014;64(1):9–29.Google Scholar
- 2.Kuhl CK. The current status of breast MR imaging: part I: choice of technique, image interpretation, diagnostic accuracy, and transfer to clinical practice. Radiology. 2007;244(2):356–378.Google Scholar
- 3.Kuhl CK. Current status of breast MR imaging: part 2: clinical applications. Radiology. 2007;244(3):672–691.Google Scholar
- 4.Esteva A, Kuprel B, Novoa RA, et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature. 2017 Jan;542(7639):115–118.Google Scholar
- 5.Revealing hidden potentials of the q-space signal in breast cancer. Proc MICCAI. 2017; p. 664–671.Google Scholar
- 6.Tajbakhsh N, Shin JY, Gurudu SR, et al. Convolutional neural networks for medical image analysis: full training or fine tuning? IEEE Trans Med Imaging. 2016 May;35(5):1299–1312.Google Scholar
- 7.Hadad O, Bakalo R, Ben-Ari R, et al. Classification of breast lesions using crossmodal deep learning. Proc ISBI. 2017; p. 109–112.Google Scholar
- 8.Marrone S, Piantadosi G, Fusco R, et al. An investigation of deep learning for lesions malignancy classification in breast DCE-MRI. Proc ICIAP. 2017; p. 479–489.Google Scholar
- 9.Antropova N, Huynh B, Giger M. Performance comparison of deep learning and segmentation-based radiomic methods in the task of distinguishing benign and malignant breast lesions on DCE-MRI. Proc SPIE. 2017;(10134).Google Scholar
- 10.Tustison NJ, Avants BB, Cook PA, et al. N4ITK: improved N3 bias correction. IEEE Trans Med Imaging. 2010 June;29(6):1310–1320.Google Scholar
- 11.He K, Zhang X, Ren S, et al. Deep residual learning for image recognition. Proc CVPR. 2016 June; p. 770–778.Google Scholar
- 12.Russakovsky O, Deng J, Su H, et al. ImageNet large scale visual recognition challenge. Int J Comput Vis;(3):211–252.Google Scholar