论文部分内容阅读
目的:建立一个超高精度的喉部切片图像数据集。方法:将新鲜人喉标本以改良川本法进行包埋,采用LeicaCM3600XP冷冻切片机以10μm层厚进行连续切片,并用Nikon D750单反相机(2400万像素)对每个断面信息进行图像采集,分别以JPG和RAW格式储存。随后参照规则断面法,用Photoshop7软件裁剪所得图像进行尺寸校正和配准,以TIFF格式保存。结果:总计获得6 656张24位RGB原始图像,数据量为JPG格式97.5GB,RAW格式174GB,配准后图像最终分辨率为40μm/像素,数据总量107GB。结论:联合改良川本法、高分辨率单反相机和规则断面法,能够获得超高精度切片图像数据集,为类似于喉的人体局部解剖精细数据集的建立提供了基础。
Objective: To create an ultra-high-precision laryngeal slice image dataset. Methods: Fresh human laryngeal specimens were embedded by modified Kawamoto’s method. The slices were continuously sliced with a thickness of 10 μm using a Leica CM3600XP cryostat and the information of each section was collected with a Nikon D750 SLR camera (24 million pixels) And RAW format storage. Then refer to the rules section method, with Photoshop7 software tailored images for size correction and registration, to TIFF format. Results: A total of 6 656 RGB 24-bit raw images were obtained. The amount of data was 97.5 GB in JPG format and 174 GB in RAW format. The final resolution of the image after registration was 40 μm / pixel and the total amount of data was 107 GB. Conclusion: The combination of improved Kawamoto-Koji method, high-resolution SLR camera and regular cross-section method can obtain ultra-high resolution slice image datasets, which provides a foundation for the establishment of anatomical finer datasets similar to larynx.