• ETRI Journal
• /
• v.44 no.1
• /
• pp.62-72
• /
• 2022

In this study, we propose a multi-GPU-based 8KVR stitching system that operates in real time on both local and cloud machine environments. The proposed system first obtains multiple 4 K video inputs, decodes them, and generates a stitched 8KVR video stream in real time. The generated 8KVR video stream can be downloaded and rendered omnidirectionally in player apps on smartphones, tablets, and head-mounted displays. To speed up processing, we adopt group-of-pictures-based distributed decoding/encoding and buffering with the NV12 format, along with multi-GPU-based parallel processing. Furthermore, we develop several algorithms such as equirectangular projection-based color correction, real-time CG overlay, and object motion-based seam estimation and correction, to improve the stitching quality. From experiments in both local and cloud machine environments, we confirm the feasibility of the proposed 8KVR stitching system with stitching speed of up to 83.7 fps for six-channel and 62.7 fps for eight-channel inputs. In addition, in an 8KVR live streaming test on the 5G MEC/cloud, the proposed system achieves stable performances with 8 K@30 fps in both indoor and outdoor environments, even during motion.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.9
• /
• pp.2458-2482
• /
• 2023

Medical image segmentation techniques based on convolution neural networks indulge in feature extraction triggering redundancy of parameters and unsatisfactory target localization, which outcomes in less accurate segmentation results to assist doctors in diagnosis. In this paper, we propose a multi-level semantic-rich encoding-decoding network, which consists of a Pooling-Conv-Former (PCFormer) module and a Cbam-Dilated-Transformer (CDT) module. In the PCFormer module, it is used to tackle the issue of parameter explosion in the conservative transformer and to compensate for the feature loss in the down-sampling process. In the CDT module, the Cbam attention module is adopted to highlight the feature regions by blending the intersection of attention mechanisms implicitly, and the Dilated convolution-Concat (DCC) module is designed as a parallel concatenation of multiple atrous convolution blocks to display the expanded perceptual field explicitly. In addition, MultiHead Attention-DwConv-Transformer (MDTransformer) module is utilized to evidently distinguish the target region from the background region. Extensive experiments on medical image segmentation from Glas, SIIM-ACR, ISIC and LGG demonstrated that our proposed network outperforms existing advanced methods in terms of both objective evaluation and subjective visual performance.

• Radiation Oncology Journal
• /
• v.17 no.4
• /
• pp.321-328
• /
• 1999

Purpose : Total body irradiation (TBI) or whole body irradiation is used to acquire immune suppression, to treat malignant lymphoma and leukemia, and as an conditioning regimen for bone marrow transplantation. For these purposes, many methods were developed to obtain homogenous dose distribution. The objective of this study was to analyze and confirm the accuracy and the homogeneity of the treatment setup, the parallel opposed lateral technique, currently used in Seoul National University Hospital. Materials and Metheods : Surface dose data, measured with a thermoluminescent dosimeter, of 8 patients among 10 patients, who were given total body irradiation with the parallel opposed lateral technique between September 1996 to August 1998, at Seoul National University Hospital were analyzed. Surface doses were measured at the head, neck, axilla, thigh, and ankle level. Surface and midline doses were measured with similar set-up and technique in the Humanoid phantom. Results : Measured surface doses relative to prescribed dose for the head, neck, axilla, thight, and ankle leve were $91.3{\pm}7.8,{\;}98.3{\pm}7.5,{\;}95.1{\pm}6.3,{\;}98.3{\pm}5.5$, and $95.3{\pm} 6.3\%$, respectively. The midline doses of the head, neck, axilla, thigh, and ankle level estimated from the surface-to-midline ratios in the Humanoid Phantom were $103.4{\pm}9.0,{\;}107.8{\pm}10.5,{\;}91.1{\pm}6.1,{\pm} 93.8{\pm}4.5,{\;}and{\;}104.5{\pm}9.3\%$, respectively. Measured surface doses and estimated midline doses ranged from $-8.9\%$ to $+7.8\%$. Midline doses at the neck and the axilia level deviated more than $5\%$ from the prescribed doses. The difference of the estimated midline doses at the neck and the axilla level and the actual doses were attributed to the thickness differences between the Humanoid phantom and the patients. Conclusion Distribution of the midline doses as well as the suface doses were measured to be within $-8.7\~{\pm}7.8\%$ range. Actual dose distribution in the patient is expected to be better than the measured dose range mainly attributed to thickness difference between the patient and the Humanoid phantom.

• The Korean Journal of Nuclear Medicine
• /
• v.35 no.6
• /
• pp.378-388
• /
• 2001

Purpose: In this study, we developed a new method for the determination of renal depth with anterior and posterior renal scintigrams in a dual-head gamma camera, considering the attenuation factor $e^{-{\mu}x}$ of the conjugate-view method. Material and Method: We developed abdomen and kidney phantoms to perform experiments using Technetium-99m dimercaptosuccinic acid ($^{99m}Tc$-DMSA). The phantom images were obtained by dual-head gamma camera equipped with low-energy, high-resolution, parallel-hole collimators (ICONf, Siemens). The equation was derived from the linear integration of omission ${\gamma}$-ray considering attenuation from the posterior abdomen to the anterior abdomen phantom surface. The program for measurement was developed by Microsoft Visual C++ 6.0. Results : Renal depths of the phantoms were derived from the derived equations and compared with the exact geometrical values. Differences between the measured and the calculated values were the range of 0.1 to 0.7 cm ($0.029{\pm}0.15cm,\;mean{\pm}S.D.$). Conclusion: The present study showed that the use of the derived equations for renal depth measurements, combined with quantitative planar imaging using dual-head gamma camera, could provide more accurate results for individual variation than the conventional method.

• Investigative Magnetic Resonance Imaging
• /
• v.21 no.2
• /
• pp.65-70
• /
• 2017

Purpose: To optimize the saturation time and maximizing the pH-weighted difference between the normal and ischemic brain regions, on 3-tesla amide proton transfer (APT) imaging using an in vivo rat model. Materials and Methods: Three male Wistar rats underwent middle cerebral artery occlusion, and were examined in a 3-tesla magnetic resonance imaging (MRI) scanner. APT imaging acquisition was performed with 3-dimensional turbo spin-echo imaging, using a 32-channel head coil and 2-channel parallel radiofrequency transmission. An off-resonance radiofrequency pulse was applied with a Sinc-Gauss pulse at a $B_{1,rms}$ amplitude of $1.2{\mu}T$ using a 2-channel parallel transmission. Saturation times of 3, 4, or 5 s were tested. The APT effect was quantified using the magnetization-transfer-ratio asymmetry at 3.5 ppm with respect to the water resonance (APT-weighted signal), and compared with the normal and ischemic regions. The result was then applied to an acute stroke patient to evaluate feasibility. Results: Visual detection of ischemic regions was achieved with the 3-, 4-, and 5-s protocols. Among the different saturation times at $1.2{\mu}T$ power, 4 s showed the maximum difference between the ischemic and normal regions (-0.95%, P = 0.029). The APTw signal difference for 3 and 5 s was -0.9% and -0.7%, respectively. The 4-s saturation time protocol also successfully depicted the pH-weighted differences in an acute stroke patient. Conclusion: For 3-tesla turbo spin-echo APT imaging, the maximal pH-weighted difference achieved when using the $1.2{\mu}T$ power, was with the 4 s saturation time. This protocol will be helpful to depict pH-weighted difference in stroke patients in clinical settings.

• Korean Journal of Fisheries and Aquatic Sciences
• /
• v.28 no.1
• /
• pp.7-14
• /
• 1995

As a part of the multidisciplinary oceanographic study for the productivity enhancement in Deukryang Bay, temperature and salinity were observed from 1992 through 1993. From the results, only the data in summer of two years are compared. Owing to the contrary meteorolgical conditions in both summers both of temperature and salinity had the patterns of horizontal distributions quite different from each other. In 1992 with low precipitation, there was a tendency of temperature increase and salinity decrease from the bay mouth towards the bay head. In 1993 when the air temperature was abnormally low, isotherms and isotherms tended to be parallel to the local u3s of the bay where the warmer and less saline water distributed along the western coast. Reduced solar radiation and increase in the relative importance of the distribution of properties along the current that was parallel to the axis of the bay could be responsible for this result. Vertical structures of both temperature and salinity were dependent on the stirring effect of tidal current. Stratification was destroyed during the spring tide while it was formed during the neap tide.

• Journal of the Korean Society of Radiology
• /
• v.13 no.5
• /
• pp.699-704
• /
• 2019

It is important to minimize the exposure dose during an examination and obtain good quality images at the same time. This study compared the beam harding effect according to the baseline superior orbito meatal line(SOML), orbito meatal line(OML), inferior orbito metal line(OML) and measured the exposure dose of the lens, especially in brain CT examinations, which generally apply to head diease patients. The beam harding effect assessment of each image along the baseline was performed quantitatively using the Image J program, and the exposure dose of the lens was detected by OSLDs and compared. As a result, As a result, when the SOML was used as the reference line, the dose of the lens was decreased by 85.08% at 80 kV and by 79.7% at 80 kV, compared to when IOML was used as the baseline. If the gantry angle at brain CT was parallel scan to SOML, there were no significant differences in the exposure to the lens and between the OML and IOML. Therefore, this study has shown that it is efficient to have a parallel scan on SOML as a protocol during Brain CT examinations.

• The Journal of the Acoustical Society of Korea
• /
• v.41 no.3
• /
• pp.359-366
• /
• 2022

In this paper, we present the Korean menu-ordering Sentence Text-to-Speech (TTS) system using conformer-based FastSpeech2. Conformer is the convolution-augmented transformer, which was originally proposed in Speech Recognition. Combining two different structures, the Conformer extracts better local and global features. It comprises two half Feed Forward module at the front and the end, sandwiching the Multi-Head Self-Attention module and Convolution module. We introduce the Conformer in Korean TTS, as we know it works well in Korean Speech Recognition. For comparison between transformer-based TTS model and Conformer-based one, we train FastSpeech2 and Conformer-based FastSpeech2. We collected a phoneme-balanced data set and used this for training our models. This corpus comprises not only general conversation, but also menu-ordering conversation consisting mainly of loanwords. This data set is the solution to the current Korean TTS model's degradation in loanwords. As a result of generating a synthesized sound using ParallelWave Gan, the Conformer-based FastSpeech2 achieved superior performance of MOS 4.04. We confirm that the model performance improved when the same structure was changed from transformer to Conformer in the Korean TTS.

• Development and Reproduction
• /
• v.2 no.1
• /
• pp.89-99
• /
• 1998

This study was carried out to investigate the morphological changes of gonadotropes in pituitary gland and spermatogenic cells in testis, obtained from 150 of 3-year-old immature and mature male rainbow trout (Oncorhynchus mykiss) during the reproductive cycles from March to February in the following year. In the maturation cycle of the pituitary gonadotropes of cultured rainbow trout, three periods can be distinguished i.e. a period of resting(March-August), a period of full spermatogenesis (September-November), and a period of breeding (December-February). The ultrastructures of the gonadotropes largely parallel the cyclical changes in the tests. The seminiferous tubules contain all spermatogenetic stages and sperm cells in a period of early maturation. At first, the size of the nucleus and cytoplasm decrease gradually at every stages from spermatogonia to spermatids. In the secondary spermatocytes, the small mitochondria are located over the outer cytoplam. In spermatids, the cytoplasmic masses move toward the posterior part of the nucleus. In spermatids, the two large mitochondria are located over the cytoplasm. In spermatids, the cytoplasmic masses move towark the posterior part of the nucleus. In spermatids, the two large mitochondria are located over the cytoplasm and begin to elongate. In spermatozoa, the surface of the nucleus devreases in volume. Examination by TEM shows that the nuclear envelope and plasma membrane are slightlywrinkled and closely adhered to the nucleus of spermatozoa. Two oval mitochondria are quite separated and the flagellum is inserted into the base of the spermatozoa head.The axoneme in this fish has the typical pattern such as nine peripheral doublets and a central doublet(9+2). there are remarkable individual differences in the size and morphology of spermatozoa head as observed by transmission and scanning electron microscopy.

• Korean Journal of Head & Neck Oncology
• /
• v.19 no.1
• /
• pp.34-40
• /
• 2003

Objective: To analyze the treatment-related parameters after the radiotherapy of T1N0 squamous cell carcinoma of the glottic larynx. Materials and Methods: Between October 1989 and August 2000, 54 patients with histologically proven T1N0 squamous cell carcinoma of the glottic larynx who received definitive radiation therapy in Department of Radiation Oncology, Asan Medical Center were analyzed. They were all males with age ranged from 31 to 80 years (median 61 years). 1997 AJCC stages were 31 T1a, 23 T1b. Patients were treated with 4-MV X-rays with a parallel-opposed two-field technique. Ten patients received 66.0-68.4Gy at 1.2Gy per fraction twice daily, 21 patients received 64.8-66.6Gy at 1.8Gy per fraction once daily, and 23 patients received 66.0Gy at 2.0Gy per fraction once daily. Follow-up period was 16-119 months (median 56 months). Results: 5-year overall survival and local control rates for patients with T1 lesions were 87.0% and 88.5%, respectively. 5-year local control with larynx preservation rate was 90.5%. Host and tumor-related prognostic factors including age, stage, anterior commissure involvement and tumor bulk proved not to be significant. Only shorter overall treatment time among treatment-related factors had correlation with imporved local control. Conclusion: Comparable high local control rate with organ preservation was achieved with primary radiation therapy and salvage surgery. Shortening of overall treatment time is related to improved local control rate. To determine the optimal fractionation scheme, randomized trial is mandatory.