Histopatoloji Görüntülerde Derin Öğrenme Yöntemleri ile Çekirdek Segmentasyonu

Öz

Histopatoloji organlar, dokular ve hücreler üzerinde oluşan değişikliklerin mikroskop üzerinde incelenmesidir. İncelenmesi gereken dokular, mikro kesiciler tarafından incelenmeye uygun kalınlıkta kesilmektedir. Kesilen dokulara bazı boyama teknikleri uygulanmaktadır. Hematoksilin-Eozin(H&E) yöntemi, en yaygın kullanılan boyama tekniğidir. Hematoksilin, hücre çekirdeklerini mavi tonlarına, eozin ise sitozplazmaları pembe tonlarına boyamaktadır. Boyanan kesitler, uzman tarafından değerlendirilmektedir. Histopatoloji görüntüleri kanser hastalığının tespiti ve kanser durumunun derecelendirilmesi için oldukça önemli rol oynamaktadır. Bu görüntülerdeki çekirdeklerin daha kolay ve başarılı analiz edilmesi için literatürde birçok çalışma bulunmaktadır. Bu çalışmalar son zamanlarda derin öğrenmenin semantik segmentasyon alanına odaklanmış ve ilgili yöntemlerle umut verici sonuçlar elde edilmiştir. Her ne kadar çekirdek segmentasyon alanında derin öğrenme yöntemleriyle çeşitli çalışmalar gerçekleştirilmiş olsa da ilgili derin öğrenme mimarilerinin birbiriyle karşılaştırmalı analizini gerçekleştiren kapsamlı bir çalışma literatürde bulunmamaktadır. Bu çalışmada popüler derin öğrenme yöntemlerinden olan U-Net, SegNet, FCN ve DeepLabV3+ olmak üzere dört farklı mimari, çekirdek segmentasyonuna uygulanıp ve gerekli analizler yapılarak en fazla faydayı sağlayan mimariyi ortaya çıkartmak amaçlanmıştır. Çalışmada uygulanan genel çekirdek segmentasyonu metodolojisi üç aşamadan oluşmaktadır: 1) Görüntü verileri eğitilmeden önce CLAHE algoritması ile ön işlemden geçirilip daha kaliteli görüntüler elde edilmeye çalışılmıştır; 2) CLAHE algoritması ile ön işlemden geçirilen görüntü verileri kullanılarak ilgili derin öğrenme mimarisi ile eğitilmiştir; 3) Eğitilen model test görüntüleri üzerinde global doğruluk ve ortalama IOU gibi kriterler kullanılarak geçerleme işlemi gerçekleştirilmiştir. Deneysel çalışmalar için içerisinde birden fazla organın histopatoloji görüntülerini bulunduran MoNuSeg veri seti kullanılmıştır. Gerçekleştirilen deneysel çalışmalar sonucunda, DeepLabV3+ mimarisi diğer mimarilere oranla çok daha kısa sürede işlemini tamamlamış ve diğer mimarilerden gözle görülür ölçülde daha iyi performans elde etmiştir.

Anahtar Kelimeler

• Çekirdek Segmentasyonu
• DeepLabV3+
• U-Net
• SegNet
• FCN
• Histopatoloji Görüntüler

Nucleus Segmentation with Deep Learning Approaches on Histopathology Images

Abstract

Histopathology is the examination of changes on organs, tissues and cells on a microscope. Tissues to be examined are cut by micro-cutters in a suitable thickness for examination. Some painting techniques are applied to the cut tissues. The Hematoxylin-Eosin (H&E) method is the most commonly used staining technique. Hematoxylin stains cell nuclei in shades of blue, and eosin stains cytosplasms in shades of pink. Stained sections are evaluated by the expert. Histopathology images play an important role in detecting cancer disease and grading cancer status. There exist many studies in the literature to analyze nucleus in these images more easily and successfully. These studies have recently focused on the semantic segmentation area of deep learning and promising results have been obtained using the corresponding methods. Although various works have been conducted with deep learning methods in the field of nucleus segmentation, there is no comprehensive work in the literature that makes comparative analysis of related deep learning architectures. This study aims to apply four different well-known deep learning architectures, which are U-Net, SegNet, FCN, and DeepLabV3+, to nuclei segmentation and determine the most suitable one for this process. The general nucleus segmentation methodology applied in the study consists of three stages; 1) Before training the image data, it is pre-processed with the CLAHE algorithm to obtain better quality images; 2) The pre-processed image data using the CLAHE algorithm is trained with the relevant deep learning architecture; 3) The trained model is verified on test images using criteria such as global accuracy and average IoU. For experimental analysis, MoNuSeg data set containing histopathology images of more than one organ is used. According to the results, DeepLabV3+ architecture completes its operation in a much shorter time than other architectures and achieves a noticeably better performance than the rest of the architectures.

Keywords

• Nucleus segmentation
• DeepLabV3+
• U-Net
• SegNet
• FCN
• Histopathology Images

Kaynakça

1. Badrinarayanan, V., Kendall, A., & Cipolla, R. (2017). SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(12), 2481–2495. https://doi.org/10.1109/TPAMI.2016.2644615
2. Chen, L. C., Papandreou, G., Kokkinos, I., Murphy, K., & Yuille, A. L. (2015). DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs. IEEE Transactions on Pattern Analysis and Machine Intelligence, 40(4), 834–848. https://doi.org/10.1109/TPAMI.2017.2699184
3. Chen, L. C., Zhu, Y., Papandreou, G., Schroff, F., & Adam, H. (2018). Encoder-decoder with atrous separable convolution for semantic image segmentation. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 11211 LNCS, 833–851. https://doi.org/10.1007/978-3-030-01234-2_49
4. Derraz, F., Beladgham, M., & Khelif, M. (2004). Application of active contour models in medical image segmentation. International Conference on Information Technology: Coding Computing, ITCC, 2, 675–681. https://doi.org/10.1109/ITCC.2004.1286732
5. Garcia-Garcia, A., Orts-Escolano, S., Oprea, S., Villena-Martinez, V., & Garcia-Rodriguez, J. (2017). A Review on Deep Learning Techniques Applied to Semantic Segmentation. 1–23. http://arxiv.org/abs/1704.06857
6. Jia, Z., Huang, X., Chang, E. I. C., & Xu, Y. (2017). Constrained Deep Weak Supervision for Histopathology Image Segmentation. IEEE Transactions on Medical Imaging, 36(11), 2376–2388. https://doi.org/10.1109/TMI.2017.2724070
7. Kumar, N., Verma, R., Anand, D., Zhou, Y., Onder, O. F., Tsougenis, E., Chen, H., Heng, P. A., Li, J., Hu, Z., Wang, Y., Koohbanani, N. A., Jahanifar, M., Tajeddin, N. Z., Gooya, A., Rajpoot, N., Ren, X., Zhou, S., Wang, Q., … Sethi, A. (2020). A Multi-Organ Nucleus Segmentation Challenge. IEEE Transactions on Medical Imaging, 39(5), 1380–1391. https://doi.org/10.1109/TMI.2019.2947628
8. Kumar, N., Verma, R., Sharma, S., Bhargava, S., Vahadane, A., & Sethi, A. (2017). A Dataset and a Technique for Generalized Nuclear Segmentation for Computational Pathology. IEEE Transactions on Medical Imaging, 36(7), 1550–1560. https://doi.org/10.1109/TMI.2017.2677499

9. Livne, M., Rieger, J., Aydin, O. U., Taha, A. A., Akay, E. M., Kossen, T., Sobesky, J., Kelleher, J. D., Hildebrand, K., Frey, D., & Madai, V. I. (2019). A U-net deep learning framework for high performance vessel segmentation in patients with cerebrovascular disease. Frontiers in Neuroscience, 13(FEB), 1–13. https://doi.org/10.3389/fnins.2019.00097
10. Moeskops, P., Wolterink, J. M., van der Velden, B. H. M., Gilhuijs, K. G. A., Leiner, T., Viergever, M. A., & Išgum, I. (2016). Deep learning for multi-task medical image segmentation in multiple modalities. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 9901 LNCS(October), 478–486. https://doi.org/10.1007/978-3-319-46723-8_55
11. Park, S. H., Yun, I. D., & Lee, S. U. (1998). Color image segmentation based on 3-D clustering: Morphological approach. Pattern Recognition, 31(8), 1061–1076. https://doi.org/10.1016/S0031-3203(97)00116-7
12. Ray, S., & Turi, R. H. (1999). Determination of number of clusters in k-means clustering and application in colour image segmentation. Proceedings of the 4th International Conference on Advances in Pattern Recognition and Digital Techniques, 137–143.
13. Reza, A. M. (2004). Realization of the contrast limited adaptive histogram equalization (CLAHE) for real-time image enhancement. Journal of VLSI Signal Processing Systems for Signal, Image, and Video Technology, 38(1), 35–44. https://doi.org/10.1023/B:VLSI.0000028532.53893.82
14. Ronneberger, O., Fischer, P., & Brox, T. (2015). U-net: Convolutional networks for biomedical image segmentation. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 9351, 234–241. https://doi.org/10.1007/978-3-319-24574-4_28
15. Roth, H. R., Shen, C., Oda, H., Oda, M., Hayashi, Y., Misawa, K., & Mori, K. (2018). Deep learning and its application to medical image segmentation. 1–6. https://doi.org/10.11409/mit.36.63
16. Shelhamer, E., Long, J., & Darrell, T. (2017). Fully Convolutional Networks for Semantic Segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(4), 640–651. https://doi.org/10.1109/TPAMI.2016.2572683
17. Simonyan, K., & Zisserman, A. (2015). Very deep convolutional networks for large-scale image recognition. 3rd International Conference on Learning Representations, ICLR 2015 - Conference Track Proceedings, 1–14.
18. Tobias, O. J., & Seara, R. (2002). Image segmentation by histogram thresholding using fuzzy sets. IEEE Transactions on Image Processing, 11(12), 1457–1465. https://doi.org/10.1109/TIP.2002.806231
19. Wang, G., Li, W., Zuluaga, M. A., Pratt, R., Patel, P. A., Aertsen, M., Doel, T., David, A. L., Deprest, J., Ourselin, S., & Vercauteren, T. (2018). Interactive Medical Image Segmentation Using Deep Learning with Image-Specific Fine Tuning. IEEE Transactions on Medical Imaging, 37(7), 1562–1573. https://doi.org/10.1109/TMI.2018.2791721
20. Zhang, X., Shan, Y., Wei, W., & Zhu, Z. (2010). An image segmentation method based on improved watershed algorithm. Proceedings - 2010 International Conference on Computational and Information Sciences, ICCIS 2010, 1(4), 258–261. https://doi.org/10.1109/ICCIS.2010.69
21. Zhou, X., Yamada, K., Takayama, R., Zhou, X., Hara, T., Fujita, H., Wang, S., & Kojima, T. (2018). Performance evaluation of 2D and 3D deep learning approaches for automatic segmentation of multiple organs on CT images. 10575, 83. https://doi.org/10.1117/12.2295178