• Investigative Magnetic Resonance Imaging
• /
• v.26 no.2
• /
• pp.104-116
• /
• 2022

The purpose of this study is to systematically determine an optimal percentile cut-off in histogram analysis for calculating the mean parameters obtained from a non-Gaussian continuous-time random-walk (CTRW) diffusion model for differentiating individual glioma grades. This retrospective study included 90 patients with histopathologically proven gliomas (42 grade II, 19 grade III, and 29 grade IV). We performed diffusion-weighted imaging using 17 b-values (0-4000 s/mm²) at 3T, and analyzed the images with the CTRW model to produce an anomalous diffusion coefficient (D_m) along with temporal (𝛼) and spatial (𝛽) diffusion heterogeneity parameters. Given the tumor ROIs, we created a histogram of each parameter; computed the P-values (using a Student's t-test) for the statistical differences in the mean D_m, 𝛼, or 𝛽 for differentiating grade II vs. grade III gliomas and grade III vs. grade IV gliomas at different percentiles (1% to 100%); and selected the highest percentile with P < 0.05 as the optimal percentile. We used the mean parameter values calculated from the optimal percentile cut-offs to do a receiver operating characteristic (ROC) analysis based on individual parameters or their combinations. We compared the results with those obtained by averaging data over the entire region of interest (i.e., 100th percentile). We found the optimal percentiles for D_m, 𝛼, and 𝛽 to be 68%, 75%, and 100% for differentiating grade II vs. III and 58%, 19%, and 100% for differentiating grade III vs. IV gliomas, respectively. The optimal percentile cut-offs outperformed the entire-ROI-based analysis in sensitivity (0.761 vs. 0.690), specificity (0.578 vs. 0.526), accuracy (0.704 vs. 0.639), and AUC (0.671 vs. 0.599) for grade II vs. III differentiations and in sensitivity (0.789 vs. 0.578) and AUC (0.637 vs. 0.620) for grade III vs. IV differentiations, respectively. Percentile-based histogram analysis, coupled with the multi-parametric approach enabled by the CTRW diffusion model using high b-values, can improve glioma grading.

• Imaging Science in Dentistry
• /
• v.52 no.3
• /
• pp.267-273
• /
• 2022

Purpose: The aim of this study was to assess the influence of kilovoltage- peak (kVp) and the metal artifact reduction (MAR) tool on the detection of buccal and lingual peri-implant dehiscence in the presence of titanium-zirconia (Ti-Zr) and zirconia (Zr) implants in cone-beam computed tomography (CBCT) images. Materials and Methods: Twenty implant sites were created in the posterior region of human mandibles, including control sites (without dehiscence) and experimental sites (with dehiscence). Individually, a Ti-Zr or Zr implant was placed in each implant site. CBCT scans were performed using a Picasso Trio device, with variation in the kVp setting (70 or 90 kVp) and whether the MAR tool was used. Three oral radiologists scored the detection of dehiscence using a 5-point scale. The area under the receiver operating characteristic (ROC) curve, sensitivity, and specificity were calculated and compared by multi-way analysis of variance (α=0.05). Results: The kVp, cortical plate involved (buccal or lingual cortices), and MAR did not influence any diagnostic values (P>0.05). The material of the implant did not influence the ROC curve values(P>0.05). In contrast, the sensitivity and specificity were statistically significantly influenced by the implant material (P<0.05) with Zr implants showing higher sensitivity values and lower specificity values than Ti-Zr implants. Conclusion: The detection of peri-implant dehiscence was not influenced by kVp, use of the MAR tool, or the cortical plate. Greater sensitivity and lower specificity were shown for the detection of peri-implant dehiscence in the presence of a Zr implant.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.12
• /
• pp.29-40
• /
• 2022

In this paper, we studied a system that detects and analyzes the pathological features of diabetic retinopathy using Mask R-CNN and a Random Forest classifier. Those are one of the deep learning techniques and automatically diagnoses diabetic retinopathy. Diabetic retinopathy can be diagnosed through fundus images taken with special equipment. Brightness, color tone, and contrast may vary depending on the device. Research and development of an automatic diagnosis system using artificial intelligence to help ophthalmologists make medical judgments possible. This system detects pathological features such as microvascular perfusion and retinal hemorrhage using the Mask R-CNN technique. It also diagnoses normal and abnormal conditions of the eye by using a Random Forest classifier after pre-processing. In order to improve the detection performance of the Mask R-CNN algorithm, image augmentation was performed and learning procedure was conducted. Dice similarity coefficients and mean accuracy were used as evaluation indicators to measure detection accuracy. The Faster R-CNN method was used as a control group, and the detection performance of the Mask R-CNN method through this study showed an average of 90% accuracy through Dice coefficients. In the case of mean accuracy it showed 91% accuracy. When diabetic retinopathy was diagnosed by learning a Random Forest classifier based on the detected pathological symptoms, the accuracy was 99%.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.6
• /
• pp.275-284
• /
• 2023

Virtual reality simulations are used for education and training in various fields, and are especially widely used in the medical field recently. The education/training simulator consists of tactile/force feedback generation and image/sound output hardware that provides a sense similar to a doctor's treatment of a real patient using real surgical tools, and software that produces realistic images and tactile feedback. Existing simulators are complicated and expensive because they have to use various types of hardware to simulate various surgical instruments used during surgery. In this paper, we propose a dental surgical simulation system using a force feedback device and a morphable haptic controller. Haptic hardware determines whether the surgical tool collides with the surgical site and provides a sense of resistance and vibration. In particular, haptic controllers that can be deformed, such as length changes and bending, can express various senses felt depending on the shape of various surgical tools. When the user manipulates the haptic feedback device, events such as movement of the haptic feedback device or button clicks are delivered to the simulation system, resulting in interaction between dental surgical tools and oral internal models, and thus haptic feedback is delivered to the haptic feedback device. Using these basic techniques, we provide a realistic training experience of impacted wisdom tooth extraction surgery, a representative dental surgery technique, in a virtual environment represented by sophisticated three-dimensional models.

• Journal of the Korean Society of Radiology
• /
• v.17 no.4
• /
• pp.551-556
• /
• 2023

Angiography equipment is used to evaluate and treat coronary artery disease. As a common feature of equipment, radiation is used, and function development for dose reduction is being carried out by each company. Therefore, the difference depending on whether DCR installed in angiography equipment is used is analyzed from a radiological point of view to prove the effect. Among 431 patients who underwent coronary artery intervention from March 2021 to February 2023, 250 patients with retrospective data were selected. And than among the 250 subjects obtained, 91 patients used the cardiovascular roadmap function during single-vessel intervention, and 159 patients did not use the roadmap. When DCR was used, total dose area product (34.57 uGy/m² : 69.15 uGy/m²), total air kerma dose (688.47 mGy : 1640.4 mGy), fluoroscopy dose (23.87 uGy/m² : 49.91 uGy/m²) and fluoroscopy time (723.55 s : 366.03 s), total number of images (17 : 26) showed lower values and were statistically significant than those not used. The use of DCR function in single vessel coronary intervention is thought to be radiologically safer as single vessel coronary intervention using dynamic cardiovascular DCR showed lower perspective time and perspective dose than procedures performed without the DCR.

• Korea Science and Art Forum
• /
• v.31
• /
• pp.171-180
• /
• 2017

The purpose of this study is to analyze the characteristics of color image for establishing the color environment contributing to the promotion of public health in the public health facilities and to utilize it as data of public health color plan and index development. For this purpose, the results of the previous precedent studies were integrated and public health facilities were classified into medical facilities (general hospitals), health facilities (public health centers), and sub - healing facilities (elderly care facilities). We visited 18 public health facilities in total, measured the environmental color of with a spectroscopic, compared the results and the precedent studies results, and identified color image characteristics and future supplement points. The results are as follows. First, the previous studies related to the environment color image vocabulary of the public health facilities, it prefer comfortable, bright and positive image. Second, as a result of direct measurement the environmental color of the public health facilities, it is found that most of them use the high brightness and low saturation color of Y series. Third, as a result of analyzing vocabulary of environmental color image of public health facilities, 'natural' image showed the highest frequency, and other images such as 'gentle' and 'decent' appeared. It was difficult to understand the characteristics of the color image vocabularies of public health facilities. This study is a convergence study of color science and environmental design, and it extends the scope of multidisciplinary research related to design and it will be helpful in environmental planning on user's emotion.

• Korea Science and Art Forum
• /
• v.35
• /
• pp.145-155
• /
• 2018

The goal of the study is to provide a new type of audio-visualization method through case analysis and work production based on Polysomnography(PSG) data that is difficult to interpret or not familiar to the public. Most art works are produced with conscious actions during waking hours. On the other hand, during sleep, we get into the world of unconsciousness. Therefore, through the experiment, want to discover if could get something new when we were in the subconscious state, and if so, wondered what kind of art could be made through it. The study method is to consider definition of sleep and sleep data first. The sleep data were classified into normal group and Narcolepsy, Insomnia, and sleep apnea by focusing on sleep disorder graphs that is measured by sleep polygraph. After that, I refined and converted the acquired biometric data into a text-based script. The degree of sleep in the text form of the script was rendered as a 3D animated image using Maya. In addition, the heart rate data script was transformed into a midi format, and the audition was implemented in the garage band. After Effects combines the image and sound to create four single channel images of 3 minutes and 20 seconds each. As a result of the research, I made an opportunity for anyone easy to understand the results, having difference with the normal data, through art instead of using difficult medical term. It also showed the possibility of artistic expression even when conscious actions did not occur. Through the results of this research, I expect the expansion and diversity of artistic audiovisual expression of biometric data.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.10
• /
• pp.67-76
• /
• 2023

Although deep learning models are making innovative achievements in the field of computer vision, the problem of vulnerability to adversarial examples continues to be raised. Adversarial examples are attack methods that inject fine noise into images to induce misclassification, which can pose a serious threat to the application of deep learning models in the real world. In this paper, we propose a model that detects adversarial examples using differences in predictive values between edge-learned classification models and underlying classification models. The simple process of extracting the edges of the objects and reflecting them in learning can increase the robustness of the classification model, and economical and efficient detection is possible by detecting adversarial examples through differences in predictions between models. In our experiments, the general model showed accuracy of {49.9%, 29.84%, 18.46%, 4.95%, 3.36%} for adversarial examples (eps={0.02, 0.05, 0.1, 0.2, 0.3}), whereas the Canny edge model showed accuracy of {82.58%, 65.96%, 46.71%, 24.94%, 13.41%} and other edge models showed a similar level of accuracy also, indicating that the edge model was more robust against adversarial examples. In addition, adversarial example detection using differences in predictions between models revealed detection rates of {85.47%, 84.64%, 91.44%, 95.47%, and 87.61%} for each epsilon-specific adversarial example. It is expected that this study will contribute to improving the reliability of deep learning models in related research and application industries such as medical, autonomous driving, security, and national defense.

• Journal of the Korean Society of Radiology
• /
• v.17 no.6
• /
• pp.865-872
• /
• 2023

The purpose of this study was to analyze fabrics suitable for use as examination gowns to determine whether examination gowns affect imaging during anterior to posterior chest examinations(Chest AP) on a digital X-ray system. Examination gowns in use at five medical centers in Seoul were collected and included modal, tencel, cotton, and rayon fabrics. The selection of fabrics was based on studies that reported fabrics with good tactile, absorbent, stretchable, and wrinkle resistance. Phantoms of five hospital gowns and four fabrics, arranged in overlapping layers from one to eight, were created and examined on a digital X-ray system in both Chest AP examination. The images examined were subjected to a first-step profile analysis, a second-step signal intensity averaging analysis, and a third-step microscopic analysis. The results showed that all nine materials had an increasing impact on the image as the number of layers of fabric increased, with the modal fabric having the least impact on the image in the first, second, and third analyses. In conclusion, as the resolution of digital x-ray systems increases, the impact of examination clothing on the image will increase, and research to find suitable materials for examination clothing will continue to be necessary.

• Journal of the Korean Society of Radiology
• /
• v.17 no.7
• /
• pp.1139-1147
• /
• 2023

The purpose of this paper is to present and evaluate the performance of a method for controlling the dose for optimal image acquisition while minimizing patient exposure by applying a small-sized Photomultiplier(SiPM) sensor inside a portable detector. Portable detectors have the advantage of being able to quickly access the patient's location for rapid diagnosis, but this mobility comes with the challenge of dose control. This paper presents a method to identify the dose that can have the DQE and optimal image quality of the detector through image evaluation based on IEC62220-1-1, an international standard for X-ray imaging devices, and to identify the optimal dose by matching the ADU of the image and the output of the SiPM Sensor. The Skull AP image was acquired by implementing the detector manufacturer's reference dose. The optimal dose was 342.8 µGy, and the optimal controlled dose was 148.3 µGy, which is 57 % of the manufacturer's reference dose. The Chest AP image was 81.9 µGy and the optimal controlled dose was 27.9 µGy, which is a high dose reduction effect of 66 %. In addition, the two images were analyzed by five radiologists and found to have no clinically significant difference in anatomical delineation.