# Deep learning and its application to medical image segmentation

**Authors:** Holger R. Roth, Chen Shen, Hirohisa Oda, Masahiro Oda, Yuichiro Hayashi, Kazunari Misawa, Kensaku Mori  
**arXiv:** [1803.08691](https://arxiv.org/abs/1803.08691)  
**Date:** 2018-03-23  
**Summary page:** https://www.summarizepaper.com/en/paper/1803.08691v1/

## Key points

- Semantic segmentation in medical imaging is a crucial task due to diverse anatomy in patients
- Deep learning advancements have revolutionized image recognition and segmentation methods
- Fully convolutional architectures are efficient for segmenting 3D medical images
- A 3D fully convolutional network (FCN) has been developed for automatic semantic segmentation of 3D medical images
- Research by Holger R. Roth, Chen Shen, Hirohisa Oda, Masahiro Oda, Yuichiro Hayashi, Kazunari Misawa, and Kensaku Mori demonstrates the potential of deep learning in medical image analysis

## Summary

, , , , 

In the field of medical imaging, semantic segmentation is a crucial task that has been challenging due to the diverse anatomy present in different patients. Recent advancements in deep learning have revolutionized image recognition and semantic segmentation methods in computer vision, paving the way for significant improvements in automatic segmentation processes. Deep convolutional neural networks have emerged as powerful tools for processing medical images, with fully convolutional architectures proving particularly efficient for segmenting 3D medical images. A notable development in this area is the creation of a 3D fully convolutional network (FCN) designed specifically to process 3D images and generate automatic semantic segmentations. This model has been trained and evaluated using a clinical computed tomography (CT) dataset, showcasing state-of-the-art performance in multi-organ segmentation tasks. The research conducted by Holger R. Roth, Chen Shen, Hirohisa Oda, Masahiro Oda, Yuichiro Hayashi, Kazunari Misawa, and Kensaku Mori highlights the potential of deep learning techniques in enhancing medical image analysis. Their work not only demonstrates the effectiveness of deep learning approaches but also underscores their applicability to real-world medical scenarios. By leveraging hierarchical feature learning within deep learning frameworks, these researchers have successfully translated cutting-edge advancements from computer vision to the realm of medical imaging. The implications of their findings extend beyond academic research, offering promising prospects for improving diagnostic accuracy and efficiency in healthcare settings. This study on "Deep learning and its application to medical image segmentation" published in Medical Imaging Technology provides valuable insights into how advanced technologies can be harnessed to address complex challenges in medical imaging.