Institute of Computer Science, Masaryk University, Botanická 68a, Brno, 602 00, Czech Republic.
J Digit Imaging. 2021 Feb;34(1):204-211. doi: 10.1007/s10278-020-00396-0. Epub 2021 Jan 7.
Medical image data plays a critical role in health care today. Whatever the context of image processing (research and educational activities, diagnostic or forensic purposes, etc.), they are supposed to be treated as highly sensitive. Unfortunately, currently available image-processing tools also enable very sophisticated malicious modification of their content.This paper is focused on the assessment of the effectiveness of selected so-called zero watermarking methods (ones that do not cause any modification of an image sample), in the protection of the integrity, and proving authorship, of these medical image studies. We have studied many zero watermarking methods and selected one representative from each type of known algorithms for comparison.We have conducted a series of simulations over a huge research database of anonymized medical image studies (patient examinations).
医疗图像数据在当今的医疗保健中起着至关重要的作用。无论图像处理的背景如何(研究和教育活动、诊断或法医学等),都应该将其视为高度敏感的信息。不幸的是,当前可用的图像处理工具也使得对其内容进行非常复杂的恶意修改成为可能。本文专注于评估选定的所谓零水印方法(即不会对图像样本进行任何修改的方法)在保护这些医学图像研究的完整性和证明作者身份方面的有效性。我们研究了许多零水印方法,并为比较选择了每种已知算法类型的一个代表性方法。我们对一个匿名医疗图像研究(患者检查)的大型研究数据库进行了一系列模拟。