Department of Mathematics, Malviya National Institute of Technology Jaipur, Jaipur, India.
Department of CSE and Indian Institute of Information Technology Kota, Jaipur, India.
J Comput Biol. 2022 Jun;29(6):545-564. doi: 10.1089/cmb.2021.0267. Epub 2022 Mar 28.
For the past two decades, fractional-order derivatives have been used to model many systems in science and engineering with more accuracy than existing integer-order derivatives. Many of these applications have been employed in the image processing field. It is undeniable that an image enhancement algorithm is very much desirable for medical image analysis to diagnose various kinds of diseases more efficiently. These requirements demand that the image should be of high quality. Hence, accurate edge-detection and denoising models are required in medical image processing, improving, and enhancing the contrast of an image to attain a better texture and avoid noise. In this study, we employ and compare the conventional methods and recent and most popular fractional-order-based methods for medical image analysis texture enhancement. To make a fair comparison, the fractional-order operators are optimized for all images with gray wolf optimizer while considering the performance metric mean squared error. The results showed that fractional differential-based operators perform better than conventional integer-order operators for texture enhancement of medical images.
在过去的二十年中，分数阶导数已经被用于科学和工程领域中的许多系统建模，其准确性优于现有的整数阶导数。这些应用中的许多都已经应用于图像处理领域。不可否认的是，对于医学图像处理来说，一种非常需要的图像增强算法可以更有效地诊断各种疾病。这些需求要求图像具有高质量。因此，在医学图像处理、改进和增强图像对比度以获得更好的纹理和避免噪声时，需要精确的边缘检测和去噪模型。在这项研究中，我们使用和比较了传统方法和最新的、最流行的基于分数阶的方法来进行医学图像分析的纹理增强。为了进行公平的比较，使用灰狼优化器对所有图像进行了分数阶算子的优化，同时考虑了性能指标均方误差。结果表明，分数微分算子在医学图像的纹理增强方面比传统的整数阶算子表现更好。
