• Journal of the Korean Society of Radiology
• /
• v.17 no.5
• /
• pp.693-699
• /
• 2023

Ultrasound is widely used in the medical field for non-destructive and non-invasive disease diagnosis. In order to improve the disease diagnosis accuracy of diagnostic medical images, improving spatial resolution is a very important factor. In this study, we aim to model the super resolution convolutional neural network (SRCNN) algorithm in ultrasound images and analyze its applicability in the medical diagnostic field. The study was conducted as an experimental study using Field II simulation and open source clinical liver hemangioma ultrasound imaging. The proposed SRCNN algorithm was modeled so that end-to-end learning can be applied from low resolution (LR) to high resolution. As a result of the simulation, we confirmed that the full width at half maximum in the phantom image using a Field II program was improved by 41.01% compared to LR when SRCNN was used. In addition, the peak to signal to noise ratio (PSNR) and structural similarity index (SSIM) evaluation results showed that SRCNN had the excellent value in both simulated and real liver hemangioma ultrasound images. In conclusion, the applicability of SRCNN to ultrasound images has been proven, and we expected that proposed algorithm can be used in various diagnostic medical fields.

• 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 Astronomical Society
• /
• v.46 no.5
• /
• pp.191-200
• /
• 2013

Full-disk solar images are provided by many solar telescopes around the world. However, the observed images show Non-Radial Variation (NRV) over the disk. In this paper, we propose algorithms for detecting distortions and restoring these images. For detecting NRV, the cross-correlation coefficients matrix of radial profiles is calculated and the minimum value in the matrix is defined as the Index of Non-radial Variation (INV). This index has been utilized to evaluate the H images of GONG, and systemic variations of different instruments are obtained. For obtaining the NRV's image, a Multi-level Morphological Filter (MMF) is designed to eliminate structures produced by solar activities over the solar surface. Comparing with the median filter, the proposed filter is a better choice. The experimental results show that the effect of our automatic detection and restoration methods is significant for getting a flat and high contrast full-disk image. For investigating the effect of our method on solar features, structural similarity (SSIM) index is utilized. The high SSIM indices (close to 1) of solar features show that the details of the structures remain after NRV restoring.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.12
• /
• pp.5998-6016
• /
• 2017

In thin-film transistor liquid-crystal displays (TFT-LCDs), which are most commonly used in mobile devices, the backlight accounts for about 70% of the power consumption. Therefore, most low-power-related studies focus on realizing power savings through backlight dimming. Image compensation is performed to mitigate the visual distortion caused by the backlight dimming. Therefore, popular techniques include pixel compensation for brightness recovery and contrast enhancement, such as histogram equalization. However, existing pixel compensation techniques often have limitations with respect to blur owing to the pixel saturation phenomenon, or because contrast enhancement cannot adequately satisfy the human visual system (HVS). To overcome these, in this study, we propose a novel dimming technique to achieve both power saving and HVS-awareness by combining the pixel compensation and histogram specifications, which convert the original cumulative density function (CDF) by designing and using the desired CDF of an image. Because the process of obtaining the desired CDF is customized to consider image characteristics, histogram specification is found to achieve better HVS-awareness than histogram equalization. For the experiments, we employ the LIVE image database, and we use the structural similarity (SSIM) index to measure the degree of visual satisfaction. The experimental results show that the proposed technique achieves up to 15.9% increase in the SSIM index compared with existing dimming techniques that use pixel compensation and histogram equalization in the case of the same low-power ratio. Further, the results indicate that it achieves improved HVS-awareness and increased power saving concurrently compared with previous techniques.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.22 no.3
• /
• pp.425-433
• /
• 2018

In this paper, we propose a super-resolution reconstruction algorithm for plenoptic images based on the reliability of disparity. The subperture image generated by the Flanoptic camera image is used for disparity estimation and reconstruction of super-resolution image based on TV_L1 algorithm. In particular, the proposed image reconstruction method is effective in the boundary region where disparity may be relatively inaccurate. The determination of reliability of disparity vector is based on the upper, lower, left and right positional relationship of the sub-aperture image. In our method, the unreliable vectors are excluded in reconstruction. The performance of the proposed method was evaluated by comparing to a bicubic interpolation method, a conventional disparity based method and dictionary based method. The experimental results show that the proposed method provides the best performance in terms of PSNR(Peak Signal to noise ratio), SSIM(Structural Similarity).

• Journal of radiological science and technology
• /
• v.44 no.6
• /
• pp.655-661
• /
• 2021

Positron emission tomography (PET) images is affected by acquisition time, short acquisition times results in low gamma counts leading to degradation of image quality by statistical noise. Noise2Void(N2V) is self supervised denoising model that is convolutional neural network (CNN) based deep learning. The purpose of this study is to evaluate denoising performance of N2V for PET image with a short acquisition time. The phantom was scanned as a list mode for 10 min using Biograph mCT40 of PET/CT (Siemens Healthcare, Erlangen, Germany). We compared PET images using NEMA image-quality phantom for standard acquisition time (10 min), short acquisition time (2min) and simulated PET image (S2 min). To evaluate performance of N2V, the peak signal to noise ratio (PSNR), normalized root mean square error (NRMSE), structural similarity index (SSIM) and radio-activity recovery coefficient (RC) were used. The PSNR, NRMSE and SSIM for 2 min and S2 min PET images compared to 10min PET image were 30.983, 33.936, 9.954, 7.609 and 0.916, 0.934 respectively. The RC for spheres with S2 min PET image also met European Association of Nuclear Medicine Research Ltd. (EARL) FDG PET accreditation program. We confirmed generated S2 min PET image from N2V deep learning showed improvement results compared to 2 min PET image and The PET images on visual analysis were also comparable between 10 min and S2 min PET images. In conclusion, noisy PET image by means of short acquisition time using N2V denoising network model can be improved image quality without underestimation of radioactivity.

• Korean Journal of Materials Research
• /
• v.32 no.8
• /
• pp.345-353
• /
• 2022

In this study, using deep learning, super-resolution images of transmission electron microscope (TEM) images were generated for nanomaterial analysis. 1169 paired images with 256 × 256 pixels (high resolution: HR) from TEM measurements and 32 × 32 pixels (low resolution: LR) produced using the python module openCV were trained with deep learning models. The TEM images were related to DyVO₄ nanomaterials synthesized by hydrothermal methods. Mean-absolute-error (MAE), peak-signal-to-noise-ratio (PSNR), and structural similarity (SSIM) were used as metrics to evaluate the performance of the models. First, a super-resolution image (SR) was obtained using the traditional interpolation method used in computer vision. In the SR image at low magnification, the shape of the nanomaterial improved. However, the SR images at medium and high magnification failed to show the characteristics of the lattice of the nanomaterials. Second, to obtain a SR image, the deep learning model includes a residual network which reduces the loss of spatial information in the convolutional process of obtaining a feature map. In the process of optimizing the deep learning model, it was confirmed that the performance of the model improved as the number of data increased. In addition, by optimizing the deep learning model using the loss function, including MAE and SSIM at the same time, improved results of the nanomaterial lattice in SR images were achieved at medium and high magnifications. The final proposed deep learning model used four residual blocks to obtain the characteristic map of the low-resolution image, and the super-resolution image was completed using Upsampling2D and the residual block three times.

• Imaging Science in Dentistry
• /
• v.52 no.4
• /
• pp.393-398
• /
• 2022

Purpose: This study proposed a generative adversarial network (GAN) model for T2-weighted image (WI) synthesis from proton density (PD)-WI in a temporomandibular joint(TMJ) magnetic resonance imaging (MRI) protocol. Materials and Methods: From January to November 2019, MRI scans for TMJ were reviewed and 308 imaging sets were collected. For training, 277 pairs of PD- and T2-WI sagittal TMJ images were used. Transfer learning of the pix2pix GAN model was utilized to generate T2-WI from PD-WI. Model performance was evaluated with the structural similarity index map (SSIM) and peak signal-to-noise ratio (PSNR) indices for 31 predicted T2-WI (pT2). The disc position was clinically diagnosed as anterior disc displacement with or without reduction, and joint effusion as present or absent. The true T2-WI-based diagnosis was regarded as the gold standard, to which pT2-based diagnoses were compared using Cohen's ĸ coefficient. Results: The mean SSIM and PSNR values were 0.4781(±0.0522) and 21.30(±1.51) dB, respectively. The pT2 protocol showed almost perfect agreement(ĸ=0.81) with the gold standard for disc position. The number of discordant cases was higher for normal disc position (17%) than for anterior displacement with reduction (2%) or without reduction (10%). The effusion diagnosis also showed almost perfect agreement(ĸ=0.88), with higher concordance for the presence (85%) than for the absence (77%) of effusion. Conclusion: The application of pT2 images for a TMJ MRI protocol useful for diagnosis, although the image quality of pT2 was not fully satisfactory. Further research is expected to enhance pT2 quality.

• Journal of the Korean Society of Radiology
• /
• v.17 no.7
• /
• pp.1041-1047
• /
• 2023

Temporal bone CT, which is a high-resolution CT, uses a high tube voltage and a thin section thickness, so the scan dose is higher than that of adjacent areas. Accordingly, we applied changes to the reconstruction algorithm among the test conditions to find an algorithm with excellent sensitivity to lesions while reducing the test dose, and investigated its significance and the possibility of providing basic clinical data. As a result, when the tube voltage was lowered to 100 kVp and applied, the dose was reduced by about 35.6%, and when the definition algorithm was applied to the raw data acquired at 100 kVp, the SNR and CNR were excellent, and a statistically significant difference was shown when compared to other algorithms(p<0.05). And as a result of comparing structural similarity, the SSIM index was analyzed as 0.776, 0.813, and 0.741 for each ROI. Therefore, we believe that applying algorithm changes to temporal bone CT scans can partially reduce the dose generated from CT scans and are very meaningful in terms of basic clinical data.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.3
• /
• pp.881-895
• /
• 2023

Modern image inpainting techniques based on deep learning have achieved remarkable performance, and more and more people are working on repairing more complex and larger missing areas, although this is still challenging, especially for facial image inpainting. For a face image with a huge missing area, there are very few valid pixels available; however, people have an ability to imagine the complete picture in their mind according to their subjective will. It is important to simulate this capability while maintaining the identity features of the face as much as possible. To achieve this goal, we propose a three-stage network model, which we refer to as the identity and structure feature refinement network (ISFRNet). ISFRNet is based on 1) a pre-trained pSp-styleGAN model that generates an extremely realistic face image with rich structural features; 2) a shallow structured network with a small receptive field; and 3) a modified U-net with two encoders and a decoder, which has a large receptive field. We choose structural similarity index (SSIM), peak signal-to-noise ratio (PSNR), L1 Loss and learned perceptual image patch similarity (LPIPS) to evaluate our model. When the missing region is 20%-40%, the above four metric scores of our model are 28.12, 0.942, 0.015 and 0.090, respectively. When the lost area is between 40% and 60%, the metric scores are 23.31, 0.840, 0.053 and 0.177, respectively. Our inpainting network not only guarantees excellent face identity feature recovery but also exhibits state-of-the-art performance compared to other multi-stage refinement models.