Diffusion2D and Anchored Inference for Asymptotic Stabilization of Diffusion-Convolutional Neural Networks in Multidomain Medical Image Classification

Hanna Willa Dhany, Sutarman Sutarman, Poltak Sihombing, Mohammad Andri Budiman

Abstract


Medical image classification across heterogeneous domains remains challenging due to domain shift, spatial variability, and unstable inference behavior. This study proposes a diffusion-stabilized Diffusion-Convolutional Neural Network (DCNN) framework that integrates Diffusion2D and post-hoc Anchored Diffusion to improve inference stability, probabilistic consistency, and robustness in multidomain medical image classification. The main contribution of this work is the introduction of a two-stage stabilization mechanism in which Diffusion2D performs controlled intra-image diffusion on feature representations before graph construction, while Anchored Diffusion refines uncertain predictions in the logit space through a k-nearest neighbors graph without retraining. The framework was evaluated on heterogeneous medical imaging datasets consisting of brain MRI, leukemia microscopy, and COVID-19 chest radiographs. Experimental results show that the proposed approach maintained baseline classification performance with an accuracy of 64.70% while improving the Macro-F1 score from 0.7045 to 0.7061. The diffusion mechanism reduced the average Laplacian value from 0.864355 to 0.187525, corresponding to a 78.23% reduction in spatial gradient variability. Internal analysis further demonstrated stable diffusion coefficients with a mean value of 0.141734 and a standard deviation of 0.003757, indicating controlled diffusion behavior. Anchored Diffusion selectively refined uncertain predictions, affecting only 0.6% of evaluated samples while preserving overall decision consistency. Repeated inference experiments across 40 iterations also revealed highly stable confidence trajectories with no observable variance after diffusion stabilization. The novelty of this research lies in combining feature-level diffusion stabilization, post-hoc anchored inference, and asymptotic regularization within a unified DCNN framework, providing a theoretically grounded and uncertainty-aware approach for robust multidomain medical image classification.


Keywords


Diffusion-Convolutional Neural Networks; Medical Image Classification; Diffusion2D; Anchored Diffusion; Inference Stability; Probabilistic Calibration; Deep Learning; Multidomain Medical Imaging

Full Text:

PDF

References


. Ammar, S., Bouwmans, T., Zaghden, N., & Neji, M. (2020). Deep detector classifier (DeepDC) for moving objects segmentation and classification in video surveillance. IET Image Processing, 14(8), 1490–1501. https://doi.org/10.1049/iet-ipr.2019.1120

. Andreoletti, D., Troia, S., Musumeci, F., Giordano, S., Maier, G., & Tornatore, M. (2019). Network traffic prediction based on diffusion convolutional recurrent neural networks. IEEE International Conference on Communications (ICC), 1–7. IEEE. https://doi.org/10.1109/ICC.2019.8761314

. Atwood, J. C. (2017). Problems in graph-structured modeling and learning (Doctoral dissertation, University of Massachusetts Amherst). https://doi.org/10.7275/10010449.0

. Atwood, J., & Towsley, D. (2016). Diffusion-convolutional neural networks. In Advances in Neural Information Processing Systems (NeurIPS 2016) (pp. 1993–2001). Curran Associates, Inc. https://doi.org/10.48550/arXiv.1511.02136

. Azzouni, A., & Others. (2017). Neutm: A neural network-based framework for traffic matrix prediction in SDN. arXiv preprint arXiv:1710.06799.

. Camuto, A., Willetts, M., Şimşekli, U., Roberts, S., & Holmes, C. (2020). Explicit regularisation in Gaussian noise injections. Advances in Neural Information Processing Systems (NeurIPS), 33, 1662–1673.

. Cao, X., & Others. (2018). Interactive temporal recurrent convolution network for traffic prediction in data centers. IEEE Access, 6, 5276–5289. https://doi.org/10.1109/ACCESS.2017.2786995

. Chollet, F. (2017). Xception: Deep learning with depthwise separable convolutions. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 1251–1258.

. Dumoulin, V., & Visin, F. (2016). A guide to convolution arithmetic for deep learning. arXiv preprint arXiv:1603.07285.

. Fagerland, M. W., Lydersen, S., & Laake, P. (2013). The McNemar test for binary matched-pairs data: Mid-p and asymptotic are better than exact conditional. BMC Medical Research Methodology, 13, 91. https://doi.org/10.1186/1471-2288-13-91

. Fan, W., Ma, Y., Li, Q., He, Y., Zhao, E., Tang, J., & Yin, D. (2019). Graph neural networks for social recommendation. In Proceedings of the World Wide Web Conference (pp. 417–426). https://doi.org/10.1145/3308558.3313488

. Giri, E. P. (2021). Algoritma convolutional neural network: Konsep, penerapan, dan implementasi dengan contoh eksperimen. Departemen Ilmu Komputer, Institut Pertanian Bogor.

. Gonzalez, R. C., & Woods, R. E. (2018). Digital Image Processing (4th ed.). Pearson.

. Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press.

. Goodfellow, Ian; Bengio, Yoshua; Courville, Aaron (2016). "6.2.2.3 Softmax Units for Multinoulli Output Distributions". Deep Learning. MIT Press. pp. 180–184. ISBN 978-0-26203561-3.

. Grigorescu, S., Trasnea, B., Cocias, T., & Macesanu, G. (2020). A survey of deep learning techniques for autonomous driving. Journal of Field Robotics, 37(3), 362–386. https://doi.org/10.1002/rob.21918

. Haykin, S. (1999). Neural Networks: A Comprehensive Foundation (2nd ed.). Prentice Hall.

. Henaff, M., Bruna, J., & LeCun, Y. (2015). Deep convolutional networks on graph-structured data. arXiv preprint arXiv:1506.05163. https://arxiv.org/abs/1506.05163

. Huang, Y., Weng, Y., Yu, S., & Chen, X. (2019). Diffusion convolutional recurrent neural network with rank influence learning for traffic forecasting. In 2019 18th IEEE International Conference on Trust, Security and Privacy in Computing and Communications/13th IEEE International Conference on Big Data Science and Engineering (TrustCom/BigDataSE) (pp. 678–685). IEEE. https://doi.org/10.1109/TrustCom/BigDataSE.2019.00096

. Kipf, T. N., & Welling, M. (2016). Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907. https://arxiv.org/abs/1609.02907

. Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). ImageNet classification with deep convolutional neural networks. In Advances in Neural Information Processing Systems, 25 (NIPS 2012, pp. 1097–1105).

. Kurniawan, A. A., & Mustikasari, M. (2021). Implementasi deep learning menggunakan metode CNN dan LSTM untuk menentukan berita palsu dalam bahasa Indonesia. Jurnal Informatika Universitas Pamulang, 5(4), 544–552. Program Studi Teknik Informatika, Universitas Pamulang.

. Li, Y., Yu, R., Shahabi, C., & Liu, Y. (2018). Diffusion convolutional recurrent neural network: Data-driven traffic forecasting. In Proceedings of the NeurIPS 2017 Time Series Workshop. Long Beach, CA.

. Mandal, M. (2021, May 22). Convolutional neural networks (CNN). Analytics Vidhya. https://www.analyticsvidhya.com/blog/2021/05/convolutional-neural-networks-cnn/

. McNemar, Q. (1947). Note on the sampling error of the difference between correlated proportions or percentages. Psychometrika, 12(2), 153–157. https://doi.org/10.1007/BF02295996

. Naeini, M. P., Cooper, G. F., & Hauskrecht, M. (2015). “Obtaining Well Calibrated Probabilities Using Bayesian Binning.” Proceedings of the AAAI Conference on Artificial Intelligence (AAAI).

. Nielsen, M. A. (2015). Neural Networks and Deep Learning. Determination Press (online book).

. Ohue, M., Li, R., Yanagisawa, K., & Akiyama, Y. (2019). Molecular activity prediction using graph convolutional deep neural network considering distance on a molecular graph. arXiv preprint arXiv:1907.01103. https://arxiv.org/abs/1907.01103

. Raghav, P. (2018). Understanding of Convolutional Neural Network (CNN) — Deep Learning. Medium.

. Ronneberger, O., Fischer, P., & Brox, T. (2015). U-Net: Convolutional networks for biomedical image segmentation. In Proceedings of the International Conference on Medical Image Computing and Computer-Assisted Intervention (pp. 234–241). Springer. https://doi.org/10.1007/978-3-319-24574-4_28

. Rumelhart, D. E., Hinton, G. E., & Williams, R. J. (1986). Learning representations by back-propagating errors. Nature, 323(6088), 533–536.

. Saha, S. (2018). A Comprehensive Guide to Convolutional Neural Networks — the ELI5 way. Towards Data Science / Medium.

. Selvi, T. K., Thamilselvan, R., & Saranya, S. M. (2021). Diffusion convolution recurrent neural network – A comprehensive survey. IOP Conference Series: Materials Science and Engineering, 1055(1), 012119. IOP Publishing. https://doi.org/10.1088/1757-899X/1055/1/012119

. Soberanis-Mukul, R. D., Navab, N., & Albarqouni, S. (2020). Uncertainty-based graph convolutional networks for organ segmentation refinement. In Medical Imaging with Deep Learning (pp. 755–769).

. Wu, Z., Pan, S., Chen, F., Long, G., Zhang, C., & Yu, P. S. (2020). A comprehensive survey on graph neural networks. IEEE Transactions on Neural Networks and Learning Systems, 1–21. https://doi.org/10.1109/TNNLS.2020.2978386

. Xu, Z. (n.d.). Lecture 9: Numerical Partial Differential Equations (Part 2). University of Notre Dame.

. Yu, B., Yin, H., & Zhu, Z. (2017). Spatio-temporal graph convolutional neural network: A deep learning framework for traffic forecasting. arXiv preprint arXiv:1709.04875. https://arxiv.org/abs/1709.04875

. Zhang, S., Tong, H., Xu, J., & Maciejewski, R. (2019). Graph convolutional networks: A comprehensive review. Computational Social Networks, 6(1), 11. https://doi.org/10.1186/s40649-019-0069-y




DOI: https://doi.org/10.47738/jads.v7i3.1215

Refbacks

  • There are currently no refbacks.


Barcode

Journal of Applied Data Sciences

ISSN : 2723-6471 (Online)
Publisher : Bright Publisher
Website : http://bright-journal.org/JADS
Email : taqwa@amikompurwokerto.ac.id (principal contact)
    support@bright-journal.org (technical issues)

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0