Tingelhoff K, Moral A I, Kunkel M E, Rilk M, Wagner I, Eichhorn K G, Wahl F M, Bootz F
Clinic and Policlinic of Otolaryngology/Ear, Nose and Throat Surgery, University of Bonn, Germany.
Annu Int Conf IEEE Eng Med Biol Soc. 2007;2007:5505-8. doi: 10.1109/IEMBS.2007.4353592.
Segmentation of medical image data is getting more and more important over the last years. The results are used for diagnosis, surgical planning or workspace definition of robot-assisted systems. The purpose of this paper is to find out whether manual or semi-automatic segmentation is adequate for ENT surgical workflow or whether fully automatic segmentation of paranasal sinuses and nasal cavity is needed. We present a comparison of manual and semi-automatic segmentation of paranasal sinuses and the nasal cavity. Manual segmentation is performed by custom software whereas semi-automatic segmentation is realized by a commercial product (Amira). For this study we used a CT dataset of the paranasal sinuses which consists of 98 transversal slices, each 1.0 mm thick, with a resolution of 512 x 512 pixels. For the analysis of both segmentation procedures we used volume, extension (width, length and height), segmentation time and 3D-reconstruction. The segmentation time was reduced from 960 minutes with manual to 215 minutes with semi-automatic segmentation. We found highest variances segmenting nasal cavity. For the paranasal sinuses manual and semi-automatic volume differences are not significant. Dependent on the segmentation accuracy both approaches deliver useful results and could be used for e.g. robot-assisted systems. Nevertheless both procedures are not useful for everyday surgical workflow, because they take too much time. Fully automatic and reproducible segmentation algorithms are needed for segmentation of paranasal sinuses and nasal cavity.
在过去几年中，医学图像数据的分割变得越来越重要。其结果被用于诊断、手术规划或机器人辅助系统的工作空间定义。本文的目的是弄清楚手动或半自动分割是否适用于耳鼻喉科手术流程，或者是否需要对鼻窦和鼻腔进行全自动分割。我们展示了鼻窦和鼻腔的手动分割与半自动分割的比较。手动分割由定制软件执行，而半自动分割由一款商业产品（Amira）实现。在本研究中，我们使用了一个鼻窦CT数据集，它由98个横向切片组成，每个切片厚度为1.0毫米，分辨率为512×512像素。为了分析这两种分割方法，我们使用了体积、范围（宽度、长度和高度）、分割时间和三维重建。分割时间从手动分割的960分钟减少到半自动分割的215分钟。我们发现鼻腔分割的差异最大。对于鼻窦，手动和半自动分割的体积差异不显著。取决于分割精度，两种方法都能得出有用的结果，并且可用于例如机器人辅助系统。然而，这两种方法对于日常手术流程都没有用，因为它们花费的时间太多。鼻窦和鼻腔的分割需要全自动且可重复的分割算法。
