Search

Low contrast to noise ratio (CNR) of unenhanced computed tomography (CT) is sometimes hard to visualize by the clinical practice. In order to assist the clinical diagnosis, a computer aided for unenhanced CT image processing is introduced in detection of hepatocellular carcinoma (HCC). This study utilized the stochastic resonance (SR) filter by adjusting localized threshold range with adding random noise for enhancing the region of interest (ROI). The quantitative measurement by using the measure of enhancement or measure of improvement (EME) is applied on the series of original and enhanced images. The value of mean and standard deviation of EME values is 2.652 ± 2.167 for the original images and 6.260 ± 1.206 for enhanced images. Then k-mean clustering method played the role based on the cluster analysis with the nearest mean for the local segmentation. The diagnostic check for determining the number of clusters on each enhanced images is important for getting a better result. In fact, K = 10 is more appropriate for the data sets of enhanced images. Finally, the image fusion process is involved two sets of data, enhanced and post-processed of enhanced and clustering information, to provide relevant information. Using the T = 0.45 as the threshold value applied on clustering and enhanced images eliminates the stronger intensity of pixels. Though those processes, the unenhanced information could be extracted out as the reference information for the clinical diagnosis. HCC was well isolated on processed images. Our results demonstrated the utilization of the computer aided for image processing of CT images might help to detect the HCC.

Contrast-enhanced computed tomography (CT) played an important role in detection of hepatocellular carcinoma (HCC). However, it's contraindication to use contrast media in patients with some comorbidities. In this study, an approach for enhancement of CT images of HCC utilizing stochastic resonance (SR), a transform based method, is introduced. A series of CT images of HCC were obtained. The non-contrast-enhanced images were processed by SR. According to the results of applications of SR, it significantly increases the signal-to-noise ratio and enhances the degree of order in a system with appropriate tuning of the noise intensity and adjusting of the threshold. In this application of CT image enhancement, SR uses the ubiquitous random perturbations of noise with appropriate intensity to emphasize the gray- level range of interest. HCC was well visualized on SR enhanced images. Furthermore, a measure of image enhancement related with concepts of the Webers Low of the human visual system was applied to quantify the enhancement contributed by SR. The values of measure of enhancement or measure of improvement (EME) for the original CT images performed mean and standard deviation of 2.615 + 2.147, and 6.257 + 1.189 for the enhanced CT images.