內視鏡手術情境下的語意分割－以資料增強達到利用少量資 料訓練深層神經網路

Semantic Segmentation in Endoscopy Surgery: Using Data Augmentation to Train Deep Neural Net with Few Data

蔣承劭(Cheng-Shao Chiang)

指導教授 ： 施吉昇

國立臺灣大學/電機資訊學院/資訊工程學研究所/碩士(2019年)

https://doi.org/10.6342/NTU201903885

摘要

隨著內視鏡手術的普及，愈來愈多研究著重在透過影像來輔助醫師進行手術。許多研究都是建立在常見的電腦視覺問題，例如物件辨識、同時定位及建立地圖等。本篇論文提出了一個資料增強的方法，來解決只擁有少量資料，如何訓練神經網路的問題。本篇論文中的應用場景為內視鏡手術情境下的語意分割，語意分割可以讓我們知道影像中出現了哪些器官等等的資訊，以利後續的其他應用。雖然語意分割已經有了大量的研究，但是這些現有的方法都需要大量的訓練資料，但是關於內視鏡手術的資料十分稀少以至於現有的演算法被侷限。實驗結果證明提出的資料增強方法可以有效的增加器官的辨識率。

關鍵字

資料增強 ； 生成對抗網路 ； 電腦輔助內視鏡手術 ； 類神經網路 ； 語意分割 ； 深度學習

並列摘要

As the computer-aided surgery getting popular, more and more research has been conducted to help surgeons operate. Most of the research are focusing on common tasks with respect to computer vision and trying to provide surgeons with more information by analyzing the images captured, whereas in this thesis, we aim at the semantic segmentation in the endoscopy surgery scenario because semantic segmentation is the first step for a computer to grasp what shows up in the vision of an endoscope. Although semantic segmentation is a popular research topic, most of the current algorithm focus on road’s scene, which needs myriads of training data. Since the data endoscopy surgery scene is relatively scarce, the performance of existing algorithms is thus rather limited.Therefore, we tried to solve the problem of training a semantic segmentation network with few data in this work. We propose a data augmentation method that can synthesize new training data. The experiment results show that our method can improve the performance in recognizing anatomical objects effectively.

並列關鍵字

Data Augmentation ； Semantic Segmentation ； Generative Adversarial Network ； Computer-Assisted Endoscopy Surgery ； Neural Network ； Deep Learning

參考文獻

[1] H.-A. Chiang, C.-S. Chiang, and C.-S. Shih, “Using collaged data augmentation to train deep neural net with few data,” in International Conference on Medical Imaging with Deep Learning – Extended Abstract Track, London, United Kingdom, 08–10 Jul 2019. [Online]. Available: https://openreview.net/forum?id=Hkg8UwMRKE

[2] L.-C. Chen, Y. Zhu, G. Papandreou, F. Schroff, and H. Adam, “Encoder-decoder with atrous separable convolution for semantic image segmentation,” in The European Conference on Computer Vision (ECCV), September 2018.

[3] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” CoRR, vol. abs/1512.03385, 2015. [Online]. Available: http://arxiv.org/abs/1512.03385

[4] M. Heusel, H. Ramsauer, T. Unterthiner, B. Nessler, G. Klambauer, and S. Hochreiter, “GANs trained by a two time-scale update rule converge to a nash equilibrium,” CoRR, vol. abs/1706.08500, 2017. [Online]. Available: http://arxiv.org/abs/1706.08500

[5] J. Song, J. Wang, L. Zhao, S. Huang, and G. Dissanayake, “MIS-SLAM: real-time large scale dense deformable SLAM system in minimal invasive surgery based on heterogeneous computing,” CoRR, vol. abs/1803.02009, 2018. [Online]. Available: http://arxiv.org/abs/1803.02009