• Journal of the Korea Society of Computer and Information
• /
• v.29 no.5
• /
• pp.21-29
• /
• 2024

Reducing the radiation dose during CT scanning can lower the risk of radiation exposure, but not only does the image resolution significantly deteriorate, but the effectiveness of diagnosis is reduced due to the generation of noise. Therefore, noise removal from CT images is a very important and essential processing process in the image restoration. Until now, there are limitations in removing only the noise by separating the noise and the original signal in the image area. In this paper, we aim to effectively remove noise from CT images using the wavelet transform-based GAN model, that is, the WT-GAN model in the frequency domain. The GAN model used here generates images with noise removed through a U-Net structured generator and a PatchGAN structured discriminator. To evaluate the performance of the WT-GAN model proposed in this paper, experiments were conducted on CT images damaged by various noises, namely Gaussian noise, Poisson noise, and speckle noise. As a result of the performance experiment, the WT-GAN model is better than the traditional filter, that is, the BM3D filter, as well as the existing deep learning models, such as DnCNN, CDAE model, and U-Net GAN model, in qualitative and quantitative measures, that is, PSNR (Peak Signal-to-Noise Ratio) and SSIM (Structural Similarity Index Measure) showed excellent results.

• Journal of the Korean Society of Radiology
• /
• v.13 no.7
• /
• pp.941-946
• /
• 2019

Computed tomography (CT) has been increasing in frequency and indications for use in clinical diagnosis and treatment decisions. Multidetector CT has the advantage of shortening the inspection time and obtaining a high resolution image compared to a single detector CT, but has been pointed out the disadvantage of increasing the radiation exposure. In addition, when the low tube voltage is used to reduce the exposure dose in the CT, noise increases relatively. In the existing method, the method of finding the optimal image quality using the method of adjusting the parameters of the image reconstruction method is not a fundamental measure. In this study, we applied a double-tree complex wavelet algorithm and analyzed the results to maintain the normal signal and remove only noise. Experimental results show that the noise is reduced from 8.53 to 4.51 when using a complex oriented 2D method with 100kVp and 0.5sec rotation time. Through this study, it was possible to remove the noise and reduce the patient dose by using the optimal noise reduction algorithm. The results of this study can be used to reduce the exposure of patients due to the low dose of CT.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.18 no.2
• /
• pp.33-38
• /
• 2014

Purpose Auto exposure control (AEC) in SPECT/CT automatically controls the exposure dose (mA) according to patient's shape and size. The aim of this study was to evaluate the effect of AEC in SPECT/CT on exposure dose reduction and image quality. Materials and Methods The model of SPECT/CT used in this study was Discovery 670 (GE, USA), Smart mA for AEC; and $^{99m}Tc$ as a radioisotope. To compare SPECT and CT images by CT exposure dose variation, we used a standard technique set at 80, 100, 120, 140 kVp, 10, 30, 50, 100, 150, 200, 250 mA, and AEC at 80, 100, 120, 140 kVp, 10-250 mA. To evaluate resolution and contrast of SPECT images, triple line phantom and flangeless Esser PET phantom were used. For CT images, noise and uniformity were checked by anthropomrphic chest phantom. For dose evaluation to find DLP value, anthropomorphic chest phantom was used and the CT protocol of torso was applied by standard technique (120 kVp, 100 mA) and AEC (120 kVp, 10-250 mA). Results When standard and AEC were applied, the resolutions at SPECT images with attenuation correction (AC) were the same as FWHM by center 3.65 mm, left 3.48 mm, right 3.61 mm. Contrasts of standard and AEC showed no significant difference: standard 53.5, 29.8, 22.5, 15.8, 6.0, AEC 53.5, 29.6, 22.4, 15.7, 6.1 In CT images, noise values at standard and AEC were 15.4 and 18.5 respectively. The application of AEC increases noise but the value of coefficient variation were 33.8, 24.9 respectively, obtaining uniform noise image. The values of DLP at standard and AEC were 426.78 and 352.09 each, which shows that the application of AEC decreases exposure dose more than standard by approximately 18%. Conclusion The results of our study show that there was no difference of AC in SPECT images based on the CT exposure dose variation at SPECT/CT images. It was found that the increased CT exposure dose leads to the improvement of CT image quality but also increases the exposure dose. Thus, the use of AEC in SPECT/CT contributes to obtaining equal AC SPECT images, and uniform noise in CT images while reducing exposure dose.

• Journal of the Korean Society of Radiology
• /
• v.14 no.7
• /
• pp.981-989
• /
• 2020

In this study, a self-made customized phantom was used to quantitatively measure the change in CT number and noise according to the change of pitch. In order to acquire an image using the phantom, the inside of the phantom was filled with sterile distilled water. Inside the glass tube, a solution obtained by diluting the ratio of normal saline and contrast medium to 100%(NS), 400:1, 200:1, 100:1, 50:1, respectively, was placed and imaged. At this time, the pitch was divided into steps of 0, 0.35, 0.7, 1.05, and 1.4 for each dilution ratio of the solution and imaged, respectively. One-way ANOVA analysis were performed to verify whether the mean of the CT number and noise values measured in all ROIs by dilution ratio showed a significant difference according to the change in pitch. As a result of the experiment, there was no statistically significant difference in the change of the CT number according to the change in the pitch for each dilution ratio, but the noise value tended to increase with the increase of the pitch, and showed a statistically significant difference. In the spiral image acquisition of CT, noise can be changed to a significant level depending on the pitch. Therefore, it will be necessary to set the quality evaluation items and criteria for CT images using the spiral image acquisition method.

• Korean Journal of Digital Imaging in Medicine
• /
• v.9 no.1
• /
• pp.21-30
• /
• 2007

Our objective was to evaluate the CT attenuation coefficient and noise of spatial domain filtering as an alternative to additional image reconstruction using different kernels in abdominal CT. Derived from thin collimated source images was generated using abdomen B10 (very smooth), B20 (smooth), B30 (medium smooth), B40 (medium), B50 (medium sharp), B60 (sharp), B70 (very sharp) and B80 (ultra sharp) kernels. Quantitative CT coefficient and noise measurements provided comparable HU (hounsfield) units in this respect. CT attenuation coefficient (mean HU) values in the abdominal were 60.4$\sim$62.2 HU and noise (7.6$\sim$63.8 HU) in the liver parenchyma. In the stomach a mean (CT attenuation coefficient) of -2.2$\sim$0.8 HU and noise (10.1$\sim$82.4 HU) was measured. Image reconstructed with a convolution kernel led to an increase in noise, whereas the results for CT attenuation coefficient were comparable. Image medications of image sharpness and noise eliminate the need for reconstruction using different kernels in the future. CT images increase the diagnostic accuracy may be controlled by adjusting CT various kernels, which should be adjusted to take into account the kernels of the CT undergoing the examination.

• Journal of the Korean Society of Radiology
• /
• v.18 no.2
• /
• pp.93-99
• /
• 2024

The digital medical imaging, especially, computed tomography (CT), should necessarily be considered in terms of noise distribution caused by converting to X-ray photon to digital imaging signal. Recently, the denoising technique based on deep learning architecture is increasingly used in the medical imaging field. Here, we evaluated noise reduction effect according to various noise types based on the U-net deep learning model in the lung CT images. The input data for deep learning was generated by applying Gaussian noise, Poisson noise, salt and pepper noise and speckle noise from the ground truth (GT) image. In particular, two types of Gaussian noise input data were applied with standard deviation values of 30 and 50. There are applied hyper-parameters, which were Adam as optimizer function, 100 as epochs, and 0.0001 as learning rate, respectively. To analyze the quantitative values, the mean square error (MSE), the peak signal to noise ratio (PSNR) and coefficient of variation (COV) were calculated. According to the results, it was confirmed that the U-net model was effective for noise reduction all of the set conditions in this study. Especially, it showed the best performance in Gaussian noise.

• Journal of the Korean Society of Radiology
• /
• v.6 no.3
• /
• pp.217-226
• /
• 2012

This study examines image quality of medical image after compression using JPEG2000 for AAPM CT Performance Phantom in PACS. The compressed images of 15:1 showed change of 1.93% and 0.81% in the CT number of water and the slice thickness, respectively, compared to the original images. The variation of the uniformity did not give a correlation for each measured area. In noise measurements at compressions of 10:1 and 15:1, changes of 1.47% to 10.99% were observed, respectively. The noise showed incremation tendency as increasing over the compression ratio 15:1, and the noise of 81.68% was measured at a compression of 40:1. CT number, uniformity, slice thickness, spatial resolution and contrast resolution for the compressed images were slightly changed by increasing the compression ratio. However, the noise was seriously changed relatively at the compressed images. Thus the noise was a important factor to determine the compression ration. A compression ratio of 10:1 for the AAPM CT Performance Phantom image was appropriate and could be applied to diagnostic images.

• Journal of the Korean Society of Radiology
• /
• v.16 no.6
• /
• pp.771-778
• /
• 2022

The purpose of this study is to present a method of measuring noise by the percentage of effective line attenuation coefficient of water that can be used for quality control of CT image noise using AAPM CT performance phantom in clinical practice. In the CT images obtained by scanning the AAPM CT performance phantom with a 120 kVp CT X-ray beam, the mean CT number was measured for each pin and water in the CT number linearity insert part. The effective energy was determined as the photon energy with the largest correlation coefficient from the correlation coefficients of the linear regression analysis of the measured mean CT number for each pin and water and the linear attenuation coefficient for each photon energy. And for water and acrylic, the contrast scale was calculated as 0.000188 cm^-1 · HU^-1 from the measured mean CT number and effective line attenuation coefficient. Using the calculated contrast scale, the effective line attenuation coefficient of water, and the standard deviation measured in the water of the alignment pin part of the AAPM CT performance phantom, The noise measurement value by the percentage of effective line attenuation coefficient of water obtained 0.31 ~ 0.52% in the range of 100 ~ 300 mAs.

• Journal of radiological science and technology
• /
• v.39 no.3
• /
• pp.329-336
• /
• 2016

The purpose of this study was to evaluate the usefulness of low-dose CT (LDCT) for emphysema compared with high-resolution CT (HRCT). Measurements of radiation dose and noise were repeated 3 times in same exposure condition which was similar with obtaining HRCT and LDCT images. We analysed reading results of 146 subjects. Six images per participants selected for emphysema grading. Emphysema was graded for all 6 zones on the left and right sides of the lungs by the consensus reading of two chest radiologists using a 4-point scale. Between the HRCT and LDCT images, diagnostic differences and agreements for emphysema were analyzed by McNemar's and unweighted kappa tests, and radiation doses and noise by a Mann-Whitney U-test, using the SPSS 19.0 program. Radiation dose from HRCT was significantly higher than that of LDCT, but the noise was significantly lower in HRCT than in LDCT. Diagnostic agreement for emphysema between HRCT and LDCT images was excellent (k-value=0.88). Emphysema grading scores were not significantly different between HRCT and LDCT images for all six lung zones. Emphysema grading scores from LDCT images were significantly correlated with increased scores on HRCT images (r=0.599, p < 0.001). Considering the tradeoff between radiation dose and image noise, LDCT could be used as the gold standard method instead of HRCT for emphysema detection and grading.

• Journal of the Korean Society of Radiology
• /
• v.13 no.2
• /
• pp.159-168
• /
• 2019

We examined the effect of Siemens ADMIRE (Advanced Modeled Iterative Reconstruction) on image quality by measuring changes in HU, noise, and SNR of background air, fat, muscle, and background signals on a chest CT scan. Experimental results show that as the ADMIRE Strength increases, the noise decreases and the signal increases, consequently the signal-to-noise ratio increases. ADMIRE can reduce noise by 28 ~ 61% compared to FBP, which is a conventional image reconstruction algorithm, and improves SNR by 16 ~ 100%.