Abstract
We address the problem of learning Deep Learning Radiomics (DLR) that are not redundant with Hand-Crafted Radiomics (HCR). To do so, we extract DLR features using a VAE while enforcing their independence with HCR features by minimizing their mutual information. The resulting DLR features can be combined with hand-crafted ones and leveraged by a classifier to predict early markers of cancer. We illustrate our method on four early markers of pancreatic cancer and validate it on a large independent test set. Our results highlight the value of combining non-redundant DLR and HCR features, as evidenced by an improvement in the Area Under the Curve compared to baseline methods that do not address redundancy or solely rely on HCR features.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Aerts, H.J., et al.: Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach. Nat. Commun. 5(1), 4006 (2014)
Afshar, P., Mohammadi, A., Plataniotis, K.N., Oikonomou, A., Benali, H.: From handcrafted to deep-learning-based cancer radiomics: challenges and opportunities. IEEE Signal Process. Maga. 36(4), 132–160 (2019)
Antropova, N., Huynh, B.Q., Giger, M.L.: A deep feature fusion methodology for breast cancer diagnosis demonstrated on three imaging modality datasets. Med. Phys. 44(10), 5162–5171 (2017)
Chen, S., et al.: Automatic scoring of multiple semantic attributes with multi-task feature leverage: a study on pulmonary nodules in CT images. IEEE Trans. Med. Imaging 36(3), 802–814 (2016)
Huynh, B.Q., Li, H., Giger, M.L.: Digital mammographic tumor classification using transfer learning from deep convolutional neural networks. J. Med. Imaging 3(3), 034501–034501 (2016)
Isensee, F., et al.: nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nat. Methods 18(2), 203–211 (2021)
Khoury, T., Asombang, A.W., Berzin, T.M., Cohen, J., Pleskow, D.K., Mizrahi, M.: The clinical implications of fatty pancreas: a concise review. Digest. Dis. Sci. 62, 2658–2667 (2017)
Kim, H., Mnih, A.: Disentangling by factorising. In: International Conference on Machine Learning, pp. 2649–2658. PMLR (2018)
Kingma, D.P., Welling, M.: Auto-encoding variational bayes. In: 2nd International Conference on Learning Representations, ICLR (2014)
Kumar, D., Wong, A., Clausi, D.A.: Lung nodule classification using deep features in CT images. In: Conference on Computer and Robot Vision, pp. 133–138. IEEE (2015)
Kumar, V., et al.: Radiomics: the process and the challenges. Magn. Reson. Imaging 30(9), 1234–1248 (2012)
Lambin, P., et al.: Radiomics: extracting more information from medical images using advanced feature analysis. Eur. J. Cancer 48(4), 441–446 (2012)
Lao, J., et al.: A deep learning-based radiomics model for prediction of survival in glioblastoma multiforme. Sci. Rep. 7(1), 10353 (2017)
Litjens, G., et al.: A survey on deep learning in medical image analysis. Med. Image Anal. 42, 60–88 (2017)
Liu, F., Xie, L., Xia, Y., Fishman, E., Yuille, A.: Joint shape representation and classification for detecting PDAC. In: Suk, H.-I., Liu, M., Yan, P., Lian, C. (eds.) MLMI 2019. LNCS, vol. 11861, pp. 212–220. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-32692-0_25
Liu, S., Xie, Y., Jirapatnakul, A., Reeves, A.P.: Pulmonary nodule classification in lung cancer screening with three-dimensional convolutional neural networks. J. Med. Imaging 4(4), 041308–041308 (2017)
Majumder, S., Philip, N.A., Takahashi, N., Levy, M.J., Singh, V.P., Chari, S.T.: Fatty pancreas: should we be concerned? Pancreas 46(10), 1251 (2017)
Matsuda, Y.: Age-related morphological changes in the pancreas and their association with pancreatic carcinogenesis. Pathol. Int. 69(8), 450–462 (2019)
Miura, S., et al.: Focal parenchymal atrophy and fat replacement are clues for early diagnosis of pancreatic cancer with abnormalities of the main pancreatic duct. Tohoku J. Exp. Med. 252(1), 63–71 (2020)
Paul, R., et al.: Deep feature transfer learning in combination with traditional features predicts survival among patients with lung adenocarcinoma. Tomography 2(4), 388–395 (2016)
Ravì, D., et al.: Deep learning for health informatics. J. Biomed. Health Inf. 21(1), 4–21 (2016)
Ronneberger, O., Fischer, P., Brox, T.: U-net: convolutional networks for biomedical image segmentation. In: Navab, N., Hornegger, J., Wells, W.M., Frangi, A.F. (eds.) MICCAI 2015. LNCS, vol. 9351, pp. 234–241. Springer, Cham (2015). https://doi.org/10.1007/978-3-319-24574-4_28
Shafiee, M.J., Chung, A.G., Khalvati, F., Haider, M.A., Wong, A.: Discovery radiomics via evolutionary deep radiomic sequencer discovery for pathologically proven lung cancer detection. J. Med. Imaging 4(4), 041305–041305 (2017)
Van Griethuysen, J.J., et al.: Computational radiomics system to decode the radiographic phenotype. Cancer Res. 77(21), e104–e107 (2017)
Vétil, R., et al.: Learning shape distributions from large databases of healthy organs: applications to zero-shot and few-shot abnormal pancreas detection. In: International Conference on Medical Image Computing and Computer Assisted Intervention, Part II, pp. 464–473. Springer, Heidelberg (2022). https://doi.org/10.1007/978-3-031-16434-7_45
Yamao, K., et al.: Partial pancreatic parenchymal atrophy is a new specific finding to diagnose small pancreatic cancer (\(\le \) 10 mm) including carcinoma in situ: comparison with localized benign main pancreatic duct stenosis patients. Diagnostics 10(7), 445 (2020)
Acknowledgments
This work was partly funded by a CIFRE grant from ANRT # 2020/1448.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendix
Appendix
1.1 Estimating the Mutual Information
The Mutual Information (MI) is estimated following the density-ratio trick [8] which requires to train a discriminator \(\mathcal {D}_{\lambda }\) predicting whether concatenated radiomics vectors [h, d] come from q(h, d) or q(h)q(d). Samples for training \(\mathcal {D}_{\lambda }\) are obtained following the procedure shown in Fig. 4. In practice, \(\mathcal {D}_{\lambda }\) is modeled as a 2-layer Multi Layer Perceptron with ReLu activation, which is trained by minimizing a binary cross-entropy (BCE) loss term. Once the discriminator is trained, the MI between HCR and DLR features can be approximated as follows:
1.2 Influence of the Hyperparameter \(\kappa \)
The final loss function for training our model is:
where \(\kappa \) is a hyperparameter weighting the importance of the the mutual information in the total loss function. Table 4 reports prediction results obtained with different values of \(\kappa \). According to these results, \(\kappa \) was set to 1 in all our experiments.
1.3 HCR Features Extraction
32 HCR features were extracted using the pyradiomics library [24]:
-
14 shape features describing the size and shape of the pancreas
-
Mesh Volume
-
Voxel Volume
-
Surface Area
-
Surface Area to Volume ratio
-
Sphericity
-
Maximum 3D diameter
-
Maximum 2D diameter in the axial plane
-
Maximum 2D diameter in the coronal plane
-
Maximum 2D diameter in the sagittal plane
-
Major Axis Length
-
Minor Axis Length
-
Least Axis Length
-
Elongation
-
Flatness
-
-
18 first-order intensity features describing the intensities distribution within the organ
-
Energy
-
Total Energy
-
Entropy
-
Minimum
-
\(10^{th}\) percentile
-
\(90^{th}\) percentile
-
Maximum
-
Mean
-
Median
-
Interquartile Range
-
Range
-
Mean Absolute Deviation
-
Robust Mean Absolute Deviation
-
Root Mean Squared
-
Skewness
-
Kurtosis
-
Variance
-
Uniformity
-
More details about each feature can be found on the online documentation.
1.4 Model Architecture
As detailed in Fig. 5, the proposed variational autoencoder (VAE) followed a 3D encoder-decoder architecture. The network topology (number of convolutions per block, filter sizes) was chosen based on the nnU-Net self-configuring procedure [6], resulting in 1, 110, 240 trainable parameters. The VAE was trained on 1000 epochs with a batch size of size 32. Every 5 epochs, the VAE was frozen and the discriminator \(\mathcal {D}_{\lambda }\) was trained for 150 epochs with a batch size equal to the total training dataset. The VAE and \(\mathcal {D}_{\lambda }\) were optimized using two independent Adam optimizers with a learning rate of \(10^{-3}\).
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Vétil, R. et al. (2023). Non-redundant Combination of Hand-Crafted and Deep Learning Radiomics: Application to the Early Detection of Pancreatic Cancer. In: Ali, S., van der Sommen, F., van Eijnatten, M., Papież, B.W., Jin, Y., Kolenbrander, I. (eds) Cancer Prevention Through Early Detection. CaPTion 2023. Lecture Notes in Computer Science, vol 14295. Springer, Cham. https://doi.org/10.1007/978-3-031-45350-2_6
Download citation
DOI: https://doi.org/10.1007/978-3-031-45350-2_6
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-45349-6
Online ISBN: 978-3-031-45350-2
eBook Packages: Computer ScienceComputer Science (R0)