MSAL-MIR: Multi-Stage Adaptive Loss for Medical Image Retrieval

Tạp chí Khoa học và Công nghệ - Đại học Đà Nẵng - Trang 81-88 - 2025
Nguyen Van Hoang Phuc1, Le Quang Nhat1, Duong Manh Quan1, Hoang Phuong Le1, Nguyen Van Hieu1
1The University of Danang - University of Science and Technology, Vietnam

Tóm tắt

Efficient and accurate retrieval of medical images underpins timely diagnosis and informed clinical decisions. This work introduces a novel multi-stage training paradigm designed for medical image retrieval. In the first stage, a ConvNeXt model pretrained on ImageNet is fine-tuned using Focal Loss to address class imbalance. Building on this foundation, the feature space is refined with Triplet Margin Loss, where chosen sample triplets are used to enhance discriminative learning. Our approach further streamlines the retrieval process by applying Global Max Pooling, L2 normalization, and Principal Component Analysis (PCA) for dimensionality reduction, followed by integration with Facebook AI Similarity Search (FAISS) for efficient similarity search. Experiments on the ISIC 2017 and COVID-19 chest X-ray datasets demonstrate that the proposed method achieves significant improvements in evaluation metrics, including mean Average Precision at 5(mAP@5), Precision at 1 (P@1), and Precision at 5 (P@5)

Từ khóa


Tài liệu tham khảo

<li>K. Fukushima, “Neocognitron: A self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position”, <em>Biological Cybernetics</em>, vol. 36, pp. 193–202, 1980, doi: 10.1007/BF00344251.</li>

<li>Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, “Gradient-based learning applied to document recognition”, <em>Proceedings of the IEEE</em>, vol. 86, no. 11, pp. 2278–2324, 1998, doi: 10.1109/5.726791.</li>

<li>T. Brosch and R. Tam, for the Alzheimer’s Disease Neuroimaging Initiative, “Manifold learning of brain MRIs by deep learning”, in <em>Medical Image Computing and Computer-Assisted Intervention – MICCAI 2013</em>, Berlin, Germany: Springer, 2013, vol. 8150, pp. 629–636, doi: 10.1007/978-3-642-40763-5_78.</li>

<li>S. M. Plis <em>et al.</em>, “Deep learning for neuroimaging: a validation study”, <em>Frontiers in Neuroscience</em>, vol. 8, p. 229, 2014, doi: 10.3389/fnins.2014.00229.</li>

<li>D. Yang, S. Zhang, Z. Yan, C. Tan, K. Li, and D. Metaxas, “Automated anatomical landmark detection on distal femur surface using convolutional neural network”, in <em>Proc. IEEE 12th Int. Symp. Biomedical Imaging (ISBI)</em>, Brooklyn, NY, USA, 2015, pp. 17–21, doi: 10.1109/ISBI.2015.7163806.</li>

<li>Y. Anavi, I. Kogan, E. Gelbart, O. Geva, and H. Greenspan, “Visualizing and enhancing a deep learning framework using patients’ age and gender for chest x-ray image retrieval”, in <em>Proc. SPIE Medical Imaging: Computer-Aided Diagnosis</em>, 2016, vol. 9785, p. 978510, doi: 10.1117/12.2217587.</li>

<li>X. Liu, H. R. Tizhoosh, and J. Kofman, “Generating binary tags for fast medical image retrieval based on convolutional nets and Radon Transform”, in <em>Proc. Int. Joint Conf. Neural Netw. (IJCNN)</em>, Vancouver, Canada, 2016, pp. 2872–2878, doi: 10.1109/IJCNN.2016.7727562.</li>

<li>K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition”, <em>arXiv preprint</em> arXiv:1512.03385, 2016.</li>

<li>N. Tajbakhsh <em>et al.</em>, “Convolutional neural networks for medical image analysis: Full training or fine tuning?”, <em>IEEE Transactions on Medical Imaging</em>, vol. 35, no. 5, pp. 1299–1312, 2016.</li>

<li>G. Huang, Z. Liu, L. van der Maaten, and K. Q. Weinberger, “Densely connected convolutional networks”, <em>Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR)</em>, <em>arXiv preprint</em> arXiv:1608.06993, 2017.</li>

<li>H. C. Shin <em>et al.</em>, “Deep convolutional neural networks for computer-aided detection: CNN architectures, dataset characteristics and transfer learning”, <em>IEEE Transactions on Medical Imaging</em>, vol. 35, no. 5, pp. 1285–1298, 2016.</li>

<li>Z. Liu <em>et al.</em>, “A ConvNet for the 2020s”, <em>arXiv preprint</em> arXiv:2201.03545, 2022.</li>

<li>A. Esteva <em>et al.</em>, “Dermatologist-level classification of skin cancer with deep neural networks”, <em>Nature</em>, vol. 542, no. 7639, pp. 115–118, 2017.</li>

<li>A. Dosovitskiy <em>et al.</em>, “An image is worth 16x16 words: Transformers for image recognition at scale”, <em>arXiv preprint</em> arXiv:2010.11929, 2020.</li>

<li>F. Schroff, D. Kalenichenko, and J. Philbin, “FaceNet: A unified embedding for face recognition and clustering”, <em>arXiv preprint</em> arXiv:1503.03832, 2015.</li>

<li>R. Hadsell, S. Chopra, and Y. LeCun, “Dimensionality reduction by learning an invariant mapping”, in <em>Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR)</em>, 2006, pp. 1735–1742.</li>

<li>J. Wang, F. Zhou, S. Wen, X. Liu, and Y. Lin, “Deep ranking for image similarity learning”, <em>arXiv preprint</em> arXiv:1405.0301, 2014.</li>

<li>T.-Y. Lin, P. Goyal, R. Girshick, K. He, and P. Dollár, “Focal loss for dense object detection”, <em>arXiv preprint</em> arXiv:1708.02002, 2017.</li>

<li>E. Ustinova and V. Lempitsky, “Learning deep embeddings with histogram loss”, <em>arXiv preprint</em> arXiv:1611.00822, 2016.</li>

<li>X. Wang <em>et al.</em>, “Multi-similarity loss with general pair weighting for deep metric learning”, <em>arXiv preprint</em> arXiv:1904.06657, 2019.</li>

<li>Y. Wen, K. Zhang, Z. Li, and Y. Qiao, “A discriminative feature learning approach for deep face recognition”, in <em>Proc. Eur. Conf. Computer Vision (ECCV)</em>, 2016, pp. 499–515.</li>

<li>J. Johnson, M. Douze, and H. Jégou, “Billion-scale similarity search with GPUs”, <em>arXiv preprint</em> arXiv:1702.08734, 2017.</li>

<li>T.-Y. Lin <em>et al.</em>, “Focal loss for dense object detection”, <em>arXiv preprint</em> arXiv:1708.02002, 2017.</li>

<li>J. Deng <em>et al.</em>, “ArcFace: Additive angular margin loss for deep face recognition”, in <em>Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR)</em>, 2019.</li>

<li>B. Cao, A. Araujo, and J. Sim, “DELG: Deep local and global features for image retrieval”, <em>arXiv preprint</em> arXiv:2001.05027, 2020.</li>

<li>A. Dosovitskiy <em>et al.</em>, “An image is worth 16x16 words: Transformers for image recognition at scale”, <em>arXiv preprint</em> arXiv:2010.11929, 2020.</li>

<li>B. Hu, B. Vasu, and A. Hoogs, “X-MIR: Explainable medical image retrieval”, in <em>Proc. IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)</em>, 2022, pp. 440–450.</li>

Thông tin
Thông tin xuất bản

Tạp chí Khoa học và Công nghệ - Đại học Đà Nẵng

81-88

Thông tin tác giả