IJIGSP Vol. 17, No. 3, 8 Jun. 2025
Cover page and Table of Contents: PDF (size: 1159KB)
PDF (1159KB), PP.123-145
Views: 0 Downloads: 0
Binary Semantic Segmentation, Transfer Learning, Dataset Distance, Optimal Transport, Gaussian Mixture Models, Clustering
This work is devoted to developing a novel transfer learning approach for solving binary semantic segmentation problems that often arise on short samples in the medical (segmentation of nodules in lungs, tumors, polyps, etc.) and other domains. The goal is to optimally select the most suitable dataset from a different subject area with similar feature space and distribution to the target data. Examples show that a severe disadvantage of transfer learning is the difficulty of selecting an initial training sample for pre-training a neural network. In this paper, we propose metrics for calculating the distance between binary segmentation datasets, allowing us to select the optimal initial training set for transfer learning. These metrics are based on the geometric distances estimation of the dataset using optimal transport, Wasserstein distance for Gaussian mixture models, clustering, and their hybrids. Experiments on datasets of medical segmentation Decathlon, LIDC, and a private dataset of tuberculomas in the lungs are presented, proving a statistically strict correlation of these metrics with a relative increase in segmentation accuracy during transfer learning.
Victor Sineglazov, Kirill Riazanovskiy, Olexander Klanovets, "Binary Segmentation Dataset Distances for Transfer Learning", International Journal of Image, Graphics and Signal Processing(IJIGSP), Vol.17, No.3, pp. 123-145, 2025. DOI:10.5815/ijigsp.2025.03.07
[1]Yan Jiang, Feng Gao, and Guoyan Xu. “Computer vision-based multiple-lane detection on straight road and in a curve”. In: 2010 International Conference on Image Analysis and Signal Processing. Zhejiang, 2010, pp. 114–117. DOI: 10.1109/IASP.2010.5476151.
[2]O. N. Lungu, L. M. Chabala, and S. Chizumba. “Satellite-Based Crop Monitoring and Yield Estimation—A Re- view”. In: Journal of Agricultural Science 13.1 (Dec. 2024), pp. 180–194. DOI: 10.5539/jas.v13n1p180.
[3]P. A. Burrough, R. A. Mcdonnell, and C. D. Lloyd. Principles of Geographical Information Systems. 330 pp. Oxford University Press, 2015.
[4]N. Funk et al. “Multi-Resolution 3D Mapping With Explicit Free Space Representation for Fast and Accurate Mobile Robot Motion Planning”. In: IEEE Robotics and Automation Letters 6.2 (Apr. 2021), pp. 3553–3560. DOI: 10.1109/LRA.2021.3061989.
[5]F.J. Yanovsky et al. “Doppler-Polarimetric Weather Radar: Returns from Wide Spread Precipitation”. In: Telecom- munications and Radio Engineering 66.8 (2007). English translation of Elektrosvyaz and Radiotekhnika, pp. 715– 727. DOI: 10.1615/TelecomRadEng.v66.i8.20.
[6]F.J. Yanovsky. “Evolution and Prospects of Airborne Weather Radar Functionality and Technology”. In: 18th Inter- national Conference on Applied Electromagnetics and Communications, ICECom 2005. Dubrovnik, Croatia, 2005, pp. 349–352. DOI: 10.1109/ICECOM.2005.204987.
[7]F.J. Yanovsky, H.W.J. Russchenberg, and C.M.H. Unal. “Retrieval of information about turbulence in rain by using Doppler-polarimetric radar”. In: IEEE Transactions on Microwave Theory and Techniques 53.2 (2005), pp. 444–449. DOI: 10.1109/TMTT.2004.840772.
[8]F.J. Yanovsky. “Doppler-Polarimetric Approach for Supercooled Water Detection in Clouds and Precipitation by Airborne Weather Radar”. In: International Radar Symposium, IRS 2004. Warsaw, Poland, 2004, pp. 93–100.
[9]V. I. Pokrovsky, V. V. Belkin, and F.J. Yanovsky. “Airborne Weather Radar for Windshear Detection”. In: 18th International Conference on Applied Electromagnetics and Communications. Dubrovnik, Croatia, 2005, pp. 1–3. DOI: 10.1109/ICECOM.2005.204989.
[10]F. Yanovsky. “Methods and means of remote definition of clouds’ electrical structure”. In: Physics and Chemistry of the Earth 22.3–4 (1997), pp. 241–245. DOI: 10.1016/S0079-1946(97)00139-0.
[11]Y. Averyanova, A. Averjanov, and F. Yanovsky. “Doppler polarization radar methods for meteorological applica- tions”. In: IEEE Aerospace and Electronic Systems Magazine 29.7 (2014), pp. 64–73. DOI: 10.1109/MAES. 2014.130143.
[12]I. Catapano et al. “Contactless Ground Penetrating Radar Imaging: State of the art, challenges, and microwave tomography-based data processing”. In: IEEE Geoscience and Remote Sensing Magazine 10.1 (Mar. 2022), pp. 251–273. DOI: 10.1109/MGRS.2021.3082170.
[13]F.J. Yanovsky, V.E. Ivashchuk, and V.P. Prokhorenko. “Through-the-wall surveillance technologies”. In: 6th In- ternational Conference on Ultrawideband and Ultrashort Impulse Signals. Sevastopol, Ukraine, 2012, pp. 30–33. DOI: 10.1109/UWBUSIS.2012.6379723.
[14]F. J. Yanovsky and R. B. Sinitsyn. “Ultrawideband Signal Processing Algorithms for Radars and Sodars”. In: 2006 3rd International Conference on Ultrawideband and Ultrashort Impulse Signals. 2006, pp. 66–71. DOI: 10. 1109/UWBUS.2006.307160.
[15]Yabo Fu et al. “A review of deep learning based methods for medical image multi-organ segmentation”. In: Physica Medica 85 (2021), pp. 107–122. ISSN: 1120-1797. DOI: 10.1016/j.ejmp.2021.05.003.
[16]Xiangbin Liu et al. “A Review of Deep-Learning-Based Medical Image Segmentation Methods”. In: Sustainability 13.3 (Jan. 2021), p. 1224. ISSN: 2071-1050. DOI: 10.3390/su13031224.
[17]Victor Sineglazov et al. “Intelligent tuberculosis activity assessment system based on an ensemble of neural net- works”. In: Computers in Biology and Medicine 147 (2022), p. 105800. ISSN: 0010-4825. DOI: https://doi. org/10.1016/j.compbiomed.2022.105800.
[18]V. Sineglazov, K. Riazanovskiy, and O. Klanovets. “A Three-Stage 2D–3D Convolutional Network Ensemble for Segmenting Malignant Brain Tumors on MRI Images”. In: Cybernetics and Systems Analysis 59.2 (Mar. 2023), pp. 199–211. DOI: 10.1007/s10559-023-00555-5.
[19]Victor Sineglazov, Olexander Klanovets, and Kirill Riazanovskiy. “Transitive Transfer Learning for Lungs CT Segmentation”. In: 2022 IEEE 3rd International Conference on System Analysis Intelligent Computing (SAIC). 2022, pp. 1–5. DOI: 10.1109/SAIC57818.2022.9923023.
[20]David Alvarez-Melis and Nicolo` Fusi. “Geometric Dataset Distances via Optimal Transport”. In: Proceedings of the 34th International Conference on Neural Information Processing Systems. NIPS’20. Vancouver, BC, Canada: Curran Associates Inc., 2020. ISBN: 9781713829546. DOI: 10.48550/arXiv.2002.02923.
[21]Julie Delon and Agne`s Desolneux. “A Wasserstein-Type Distance in the Space of Gaussian Mixture Models”. In: SIAM Journal on Imaging Sciences 13.2 (2020), pp. 936–970. DOI: 10.1137/19M1301047.
[22]Eliseu Guimara˜es et al. “Transfer learning for Twitter sentiment analysis: Choosing an effective source dataset”. In: Anais do VIII Symposium on Knowledge Discovery, Mining and Learning. Evento Online: SBC, 2020, pp. 161–168. DOI: 10.5753/kdmile.2020.11972.
[23]V. Cheplygina. “Cats or CAT scans: transfer learning from natural or medical image source datasets?” In: ArXivabs/1810.05444 (2018).
[24]Alessandro Achille et al. “Task2Vec: Task Embedding for Meta-Learning”. In: 2019 IEEE/CVF International Con- ference on Computer Vision (ICCV). 2019, pp. 6429–6438. DOI: 10.1109/ICCV.2019.00653.
[25]Yossi Rubner, Carlo Tomasi, and Leonidas J. Guibas. “The Earth Mover’s Distance as a Metric for Image Retrieval”. In: International Journal of Computer Vision 40 (2000), pp. 99–121. DOI: 10.1023/A:1026543900054.
[26]Marco Cuturi. “Sinkhorn Distances: Lightspeed Computation of Optimal Transport”. In: Proceedings of the 26th International Conference on Neural Information Processing Systems - Volume 2. NIPS’13. Lake Tahoe, Nevada: Curran Associates Inc., 2013, pp. 2292–2300. DOI: 10.48550/arXiv.1306.0895.
[27]bi Hui et al. “Accurate image segmentation using Gaussian mixture model with saliency map”. In: Pattern Analysis and Applications 21 (Aug. 2018). DOI: 10.1007/s10044-017-0672-1.
[28]Farhan Riaz et al. “Gaussian Mixture Model Based Probabilistic Modeling of Images for Medical Image Segmen- tation”. In: IEEE Access 8 (2020), pp. 16846–16856. DOI: 10.1109/ACCESS.2020.2967676.
[29]Xue Shi, Yu Li, and Quanhua Zhao. “Flexible Hierarchical Gaussian Mixture Model for High-Resolution Remote Sensing Image Segmentation”. In: Remote Sensing 12.7 (2020). ISSN: 2072-4292. DOI: 10.3390/rs12071219.
[30]Xianghong Lin, Xiaofei Yang, and Ying Li. “A Deep Clustering Algorithm based on Gaussian Mixture Model”. In: Journal of Physics: Conference Series 1302 (Aug. 2019), p. 032012. DOI: 10.1088/1742-6596/1302/ 3/032012.
[31]Jeff A. Bilmes. “A gentle tutorial of the EM algorithm and its application to parameter estimation for Gaussian mixture and hidden Markov models”. In: CTIT technical reports series (1998).
[32]Scott Shaobing Chen and P.S. Gopalakrishnan. “Clustering via the Bayesian information criterion with applications in speech recognition”. In: Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP ’98 (Cat. No.98CH36181). Vol. 2. 1998, pp. 645–648. DOI: 10.1109/ICASSP.1998. 675347.
[33]Peter J. Rousseeuw. “Silhouettes: A graphical aid to the interpretation and validation of cluster analysis”. In: Jour- nal of Computational and Applied Mathematics 20 (1987), pp. 53–65. ISSN: 0377-0427. DOI: https://doi. org/10.1016/0377-0427(87)90125-7.
[34]M. Hassaballah and Ali Ismail Awad. “Detection and Description of Image Features: An Introduction”. In: Image Feature Detectors and Descriptors : Foundations and Applications. Ed. by Ali Ismail Awad and Mahmoud Hassa- ballah. Cham: Springer International Publishing, 2016, pp. 1–8. ISBN: 978-3-319-28854-3. DOI: 10.1007/978- 3-319-28854-3_1.
[35]Fan Xu et al. “Comparison of Image Feature Detection Algorithms”. In: 2022 9th International Conference on Dependable Systems and Their Applications (DSA). 2022, pp. 723–731. DOI: 10.1109/DSA56465.2022. 00103.
[36]David Arthur and Sergei Vassilvitskii. “K-Means++: The Advantages of Careful Seeding”. In: Proceedings of the Eighteenth Annual ACM-SIAM Symposium on Discrete Algorithms. SODA ’07. New Orleans, Louisiana: So- ciety for Industrial and Applied Mathematics, 2007, pp. 1027–1035. ISBN: 9780898716245. DOI: 10 . 1145 / 1283383.1283494.
[37]Robert Tibshirani, Guenther Walther, and Trevor Hastie. “Estimating the Number of Clusters in a Data Set Via the Gap Statistic”. In: Journal of the Royal Statistical Society Series B: Statistical Methodology 63.2 (Jan. 2002), pp. 411–423. ISSN: 1369-7412. DOI: 10.1111/1467-9868.00293. eprint: https://academic.oup. com/jrsssb/article-pdf/63/2/411/49590410/jrsssb\_63\_2\_411.pdf.
[38]Michela Antonelli et al. “The Medical Segmentation Decathlon”. In: Nature Communications 13 (2022). DOI: 10.1038/s41467-022-30695-9.
[39]Samuel Armato III et al. “The Lung Image Database Consortium (LIDC) and Image Database Resource Initiative (IDRI): A completed reference database of lung nodules on CT scans”. In: Medical Physics 38 (Jan. 2011), pp. 915– 931. DOI: 10.1118/1.3528204.
[40]Liang-Chieh Chen et al. “Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmenta- tion”. In: Computer Vision – ECCV 2018: 15th European Conference, Munich, Germany, September 8–14, 2018, Proceedings, Part VII. Munich, Germany: Springer-Verlag, 2018, pp. 833–851. ISBN: 978-3-030-01233-5. DOI: 10.1007/978-3-030-01234-2_49.
[41]Manzil Zaheer et al. “Deep Sets”. In: Proceedings of the 31st International Conference on Neural Information Processing Systems (Dec. 2017), pp. 3394–3404. DOI: 10.5555/3294996.3295098.
[42]Amirata Ghorbani et al. “Data Shapley: Equitable Valuation of Data for Machine Learning”. In: Proceedings of the 36th International Conference on Machine Learning, vol. 97 (Jun. 2019), pp. 2242–2251. DOI: 10.48550/arXiv.1904.02868.
[43]A. Tran et al. “Transferability and Hardness of Supervised Classification Tasks”. In: 2019 IEEE/CVF International Conference on Computer Vision (ICCV) (Oct. 2019), pp. 1395–1405. DOI: 10.1109/ICCV.2019.00148.