• 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.10 no.6
• /
• pp.419-425
• /
• 2016

The purpose of this study is to produce a shielding material to reduce a dose on the genital gland, one of the superficial organs, at a CT scan on the whole abdomen and hardly affect picture quality and examine its utility. This research made 22 mm silicone and 7.3 mm aluminum having the similar material quality and effect of previous bismuth. By using the non-shield, bismuth, 22 mm silicone, and 7.3 mm aluminum shielding materials, this author conducted a comparative experiment measuring the decay rate of the genital gland's exposure to radiation, change of the CT number and noise in the image, and the CT number, noise, and uniformity in the AAPM phantom. According to the results, exposure to radiation is reduced in bismuth as 29.96%, silicone 22 mm as 13.10%, and 7.3 mm aluminum as 18.27%. In bismuth, however, the image's CT number varies a lot, and uniformity is measured to be inappropriate in the AAPM phantom scan; therefore, it indicates great change in terms of picture quality in superficial organs like the genital gland. Concerning superficial organs like the genital gland, if 22 mm silicone and 7.3 mm aluminum are used as shielding materials, it will be helpful in reducing variation in picture quality and also decreasing radiation exposure to radiation.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2016.05a
• /
• pp.271-274
• /
• 2016

CT quality assurance imaging evaluation and enforcement as quantitative assessment by phantom image evaluation, assessment items include There are also contrasting the water attenuation coefficient, uniformity, noise, resolution, spatial resolution, 10mm slice thickness evaluation, contrast resolution, space for the resolution, the slice thickness evaluation, it is possible to estimate the error due to the evaluation by the subjective judgment of the tester, using a subjective error image processing program to be computed to minimize the objective evaluation. Basic recording conditions of the CT image quality control assessment is the same as special medical equipment quality control checks, the images were evaluated quantitatively using IMAGE J. For a CT attenuation coefficient, the uniformity, noise evaluation, were evaluated as CT quality control image the standard deviation of the measured value of the digital processing of image smaller and less noise uniform images than the, contrast and resolution assessment is the size of the diameter of a circle having a large the 1 inch, 0.75 inch, 0.5 inch quality if the diameter of the circle, was evaluated in the small circle in the near circle ellipse. Spatial resolution is evaluated by using a self-extracting features of an image processing program, all of the groups of members comprising the acceptance criteria to automatically extract, was evaluated to be very useful for the quantitative assessment. When CT image quality control assessment on the basis of the results such as the above, if using an image processing program to minimize the subjective judgment of the error evaluator and is determined more efficient than would be made quantitative evaluation.

• Journal of the Korean Society of Radiology
• /
• v.12 no.3
• /
• pp.403-407
• /
• 2018

Currently, the highest cancer death rate in Korea is lung cancer, which is a typical cancer that is difficult to detect early. Low-dose chest CT is being used for early detection, which has a greater lung cancer diagnosis rate of about three times than regular chest x-ray images. However, low-dose chest CT not only significantly reduces image resolution but also has a weak signal and is sensitive to noise. Also, air filled lungs are low-density organs and the presence of noise can significantly affect early diagnosis of cancer. This study used Visual C++ to set a circle inside a large circle with a density of 2.0, with a density of 1.0, which is the density of water, in which five small circle of mathematics have different densities. Gaussian noise was generated by 1%, 2%, 3%, and 4% respectively to determine the effect of noise on the mean value, the standard deviation value, and the relative noise ratio(SNR). In areas where the density difference between the large and small circles was greatest in the event of 1 % noise, the SNR in the area with the greatest variation in noise was 4.669, and in areas with the lowest density difference, the SNR was 1.183. In addition, the SNR values can be seen to be high if the same results are obtained for both positive and negative densities. Quality was also clearly visible when the density difference was large, and if the noise level was increased, the SNR was reduced to significantly affect the noise. Low-density organs or organs in areas of similar density to cancers, will have significant noise effects, and the effects of density differences on the probability of noise will affect diagnosis.

• The Journal of the Korea Contents Association
• /
• v.13 no.6
• /
• pp.331-338
• /
• 2013

CT for follow-up visits because of liver disease, body mass index (BMI) and kVp according to the change of the image quality and radiation dose to evaluate for changes. March 2010 to June 2011 at Pusan P University Hospital, abdominal CT scans a patient BMI (Body Mass Index. Less BMI) index was less than 25 in the treatment of subjects had a 48-person Noise and SNR at 100kVp abdominal image is lager than the 120kVp image. CTDI volume value at by the analysis of the radiation dose is 4.47mGy(100kVp) and 9.01mGy(120kVp). So CTDIvol in 100kVp is smaller than CTDIvol in 120kVp(decrease by 44.1%). And, effective dose is 7.1mSv(100kVp) and 12.51mSv(120kVp). So effective dose in 100kVp is smaller than effective dose in 120kVp(decrease by 43%). Evaluation of image quality is that Unacceptable 0 person, Suboptimal 0 person, Adequate 0 person, Good 1 person, Excellent 47 person. In case of repeatly patient, we examinate abdomianl CT scan by using low kVp and body mass index less than 25. We can has good quality image and benefit of low radiation dose.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2001.06a
• /
• pp.253-256
• /
• 2001

본 논문에서는 전처리 과정에서 잡음의 효과적 처리를 위해 기존의 필터 방식들이 가지는 단점인 경계 부분의 블러링 현상을 줄이고 정확한 에지 위치를 보존할 수 있는 비등방성 확산 필터를 사용하여 CT나 MRI 2차원 영상 슬라이스들을 만들어내고 이 슬라이스들을 3차원 데이터 셋으로 구성하여 3차원 공간의 볼륨 데이터로 시각적인 영상정보를 얻는데 있다.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.24 no.1
• /
• pp.15-19
• /
• 2020

Purpose Generative Adversarial Network(GAN) is one of deep learning technologies. This is a way to create a real fake image after learning the real image. In this study, after acquiring artificial intelligence images through GAN, We were compared and evaluated with real scan time images. We want to see if these technologies are potentially useful. Materials and Methods 30 patients who underwent ¹⁸F-FDG Brain PET/CT scanning at Severance Hospital, were acquired in 15-minute List mode and reconstructed into 1,2,3,4,5 and 15minute images, respectively. 25 out of 30 patients were used as learning images for learning of GAN and 5 patients used as verification images for confirming the learning model. The program was implemented using the Python and Tensorflow frameworks. After learning using the Pix2Pix model of GAN technology, this learning model generated artificial intelligence images. The artificial intelligence image generated in this way were evaluated as Mean Square Error(MSE), Peak Signal to Noise Ratio(PSNR), and Structural Similarity Index(SSIM) with real scan time image. Results The trained model was evaluated with the verification image. As a result, The 15-minute image created by the 5-minute image rather than 1-minute after the start of the scan showed a smaller MSE, and the PSNR and SSIM increased. Conclusion Through this study, it was confirmed that AI imaging technology is applicable. In the future, if these artificial intelligence imaging technologies are applied to nuclear medicine imaging, it will be possible to acquire images even with a short scan time, which can be expected to reduce artifacts caused by patient movement and increase the efficiency of the scanning room.

• Journal of the Korea Convergence Society
• /
• v.9 no.12
• /
• pp.107-113
• /
• 2018

The purpose of this study was to evaluate the dose and image quality according to the rotation of the X-axis direction in the abdominal CT scan, and to find ways to reduce the exposure dose. The phantom was scanned by rotating in the X-axis direction at 0, 5, 10, and 15 degrees, respectively. The CTDIvol value, HU, noise, and signal-to-noise ratio were measured at each rotation. ANOVA analysis was performed using the SPSSWIN (ver 19.0) program. The radiation exposure dose was 5.44mGy, 5.70mGy, 5.98mGy and 6.38mGy at 0, 5, 10 and 15 degrees, respectively. HU, noise, and signal-to-noise ratio were not statistically significant. In the CT scan, if the patient is located in the isocenter of the gantry aperture and there is no rotation in the X-axis direction, the exposure dose is reduced.

• Journal of radiological science and technology
• /
• v.38 no.2
• /
• pp.121-126
• /
• 2015

To image diagnosis in neurovascular diseases using Multi-Detector Computed Tomography(MDCT), injected the same contrast material when inspecting Brain Computed Tomography Angiography(BCTA) to examine radiation dose and Image quality on changing Cerebral Artery CT number by tube voltage. Executed an examination with same condition[Beam Collimation $128{\times}0.6mm$, Pitch 0.6, Rotation Time 0.5s, Slice Thickness 5.0mm, Increment 5.0mm, Delay Time 3.0sec, Care Dose 4D(Demension ; D)] except for tube voltage on 50 call patients for BCTA and divided them into two groups (25 people for a group, group A: 80, group B: 120kVp). From all the acquired images, set a ROI(Region of Interest) on four spots such as left cerebral artery, right cerebral artery, posterior cerebral artery and cerebral parenchyma to compare quantitative evaluation, qualitative evaluation and effective dose after measuring CT number value from Picture Archiving Communications System(PACS). Evaluating images with CT number acquired from BCTA examination, images with 80 kVp was 18% higher in Signal to Noise Ratio and 19% in Contrast to Noise Ratio than those with 120 kVp. It was seen that expose dose was decreased by over 50% with tube voltage 80 kVp than with 120 kVp. Group A (25 patients) was examination with tube voltage 80kVp while group B with 120 kVp to examine radiation dose and Image quality. It is considered effective to inspect with lower tube voltage than with conventional high kVp, which can reduce radiation dose without any affect on diagnosis.

• 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.