• The Journal of the Acoustical Society of Korea
• /
• v.39 no.1
• /
• pp.24-31
• /
• 2020

In this paper, the deep convolutional neural network for sound event detection is studied. Especially, the end-to-end neural network, which generates the detection results from the input audio waveform, is studied for weakly supervised problem that includes weakly-labeled and unlabeled dataset. The proposed system is based on the network structure that consists of deeply-stacked 1-dimensional convolutional neural networks, and enhanced by the skip connection and gating mechanism. Additionally, the proposed system is enhanced by the sound event detection and post processings, and the training step using the mean-teacher model is added to deal with the weakly supervised data. The proposed system was evaluated by the Detection and Classification of Acoustic Scenes and Events (DCASE) 2019 Task 4 dataset, and the result shows that the proposed system has F₁-scores of 54 % (segment-based) and 32 % (event-based).

• Progress in Medical Physics
• /
• v.19 no.1
• /
• pp.14-20
• /
• 2008

Accounting for tumor motion in treatment planning and delivery is one of the most recent and significant challenges facing radiotherapy. The purpose of this study was to investigate the correlation and clarified the relationship between the motion of an external marker using the Real-Time Position Management (RPM) System and an internal organ motion signal obtained fluoroscope. We enrolled 10 patients with locally advanced lung cancer and liver cancer, retrospectively. The external marker was a plastic box, which is part of the RPM used to track the patient's respiration. We investigated the quantitatively correlation between the motions of an external marker with RPM and internal motion with fluoroscope. The internal fiducial motion is predominant in the caraniocaudal direction, with a range of $1.3{\sim}3.5cm$ with fluoroscopic unit. The external fiducial motion is predominant in the caraniocaudal direction, with a range of $0.43{\sim}2.19cm$ with RPM gating. The two measurements ratio is from 1.31 to 5.56. When the regularization guided standard deviation is from 0.08 to 0.87, mean 0.204 cm, except only for patients #3 separated by a mean 0.13 cm, maximum of 0.23 cm. This result is a good correlation between internal tumor motion imaged by fluoroscopic unit and external marker motion with RPM during expiration within 0.23 cm. We have demonstrated that gating may be best performed but special attention should be paid to gating for patients whose fiducials do not move in synchrony, because targeting on the correct phase difference alone would not guarantee that the entire tumor volume is within the treatment field.

• Journal of Acupuncture Research
• /
• v.20 no.5
• /
• pp.172-186
• /
• 2003

Acupuncture therapy has demonstrated efficacy in several clinical areas, and of these areas the understanding of pain has progressed immensely in the last two decades. The underlying mechanisms of acupuncture in general and the analgesic effect in particular are still not clearly delineated. The leading hypothesis include the effects of local stimulation, neuronal gating, release of endogenous opiates, and the placebo effect. Accumulating evidence suggests that the central nervous system(CNS) is essential for the processing of these effects, via its modulation of the autonomic nervous system, neuro-immune system, and hormonal regulation. These processes tap into basic survival mechanisms. As such, understanding the effects of acupuncture within a neuroscience-based framework becomes vital. We propose a model which incorporates the stress-induced hypothalamus-pituitary-adrenal axis(HPA-axis) model of Akil et al., the cholinergic anti-inflamatory observations of Tracey et al., and Petrovic et al.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.7 no.2
• /
• pp.205-213
• /
• 2002

This paper deals with the implementation of three-phase VSI systems which can control the power regenerated from DC bus line to AC supply. The overall system consists of the line-to-line voltage and line current sensors, an actual power calculator using d-q transformation method, a complex power controller with PI control scheme, a gating signal generator for modified ｑ-conduction mode, a DPLL for frequency followup, and Power circuits. Control board is constructed by using a 32-bit DSP TMS32C32, two EFLDs , six ADCs, and a DAC. To verify the performance of the proposed system, we designed and constructed the propotype with the power rating of 5kVA at AC 220V. Experimental results show that the regenerated active power is well controlled to its command vague and the regenerated reactive power still remained at nearly zero through all operating modes.

• JSTS:Journal of Semiconductor Technology and Science
• /
• v.16 no.6
• /
• pp.745-753
• /
• 2016

This paper presents a fully synthesizable low power interconnect bus for millimeter-scale wireless sensor nodes. A segmented ring bus topology minimizes the required chip real estate with low input/output pad count for ultra-small form factors. By avoiding the conventional open drain-based solution, the bus can be fully synthesizable. Low power is achieved by obviating a need for local oscillators in member nodes. Also, aggressive power gating allows low-power standby mode with only 53 gates powered on. An integrated wakeup scheme is compatible with a power management unit that has nW standby mode. A 3-module system including the bus is fabricated in a 180 nm process. The entire system consumes 8 nW in standby mode, and the bus achieves 17.5 pJ/bit/chip.

• Investigative Magnetic Resonance Imaging
• /
• v.19 no.1
• /
• pp.1-9
• /
• 2015

Purpose: Several morphometric studies have been performed to investigate brain abnormalities in congenitally deaf people. But no report exists concerning structural brain abnormalities in congenitally deaf adolescents. We evaluated the regional volume changes in gray matter (GM) using voxel-based morphometry (VBM) in congenitally deaf adolescents. Materials and Methods: A VBM8 methodology was applied to the T1-weighted magnetic resonance imaging (MRI) scans of eight congenitally deaf adolescents (mean age, 15.6 years) and nine adolescents with normal hearing. All MRI scans were normalized to a template and then segmented, modulated, and smoothed. Smoothed GM data were tested statistically using analysis of covariance (controlled for age, gender, and intracranial cavity volume). Results: The mean values of age, gender, total volumes of GM, and total intracranial volume did not differ between the two groups. In the auditory centers, the left anterior Heschl's gyrus and both inferior colliculi showed decreased regional GM volume in the congenitally deaf adolescents. The GM volumes of the lingual gyri, nuclei accumbens, and left posterior thalamic reticular nucleus in the midbrain were also decreased. Conclusions: The results of the present study suggest that early deprivation of auditory stimulation in congenitally deaf adolescents might have caused significant underdevelopment of the auditory cortex (left Heschl's gyrus), subcortical auditory structures (inferior colliculi), auditory gain controllers (nucleus accumbens and thalamic reticular nucleus), and multisensory integration areas (inferior colliculi and lingual gyri). These defects might be related to the absence of general auditory perception, the auditory gating system of thalamocortical transmission, and failure in the maturation of the auditory-to-limbic connection and the auditorysomatosensory-visual interconnection.

• Molecules and Cells
• /
• v.39 no.4
• /
• pp.322-329
• /
• 2016

Voltage-gated $Ca^{2+}$ ($Ca_V$) channels are dynamically modulated by Gprotein-coupled receptors (GPCR). The $M_1$ muscarinic receptor stimulation is known to enhance $Ca_V2.3$ channel gating through the activation of protein kinase C (PKC). Here, we found that $M_1$ receptors also inhibit $Ca_V2.3$ currents when the channels are fully activated by PKC. In whole-cell configuration, the application of phorbol 12-myristate 13-acetate (PMA), a PKC activator, potentiated $Ca_V2.3$ currents by ~two-fold. After the PMA-induced potentiation, stimulation of $M_1$ receptors decreased the $Ca_V2.3$ currents by $52{\pm}8%$. We examined whether the depletion of phosphatidylinositol 4,5-bisphosphate ($PI(4,5)P_2$) is responsible for the muscarinic suppression of $Ca_V2.3$ currents by using two methods: the Danio rerio voltage-sensing phosphatase (Dr-VSP) system and the rapamycin-induced translocatable pseudojanin (PJ) system. First, dephosphorylation of $PI(4,5)P_2$ to phosphatidylinositol 4-phosphate (PI(4)P) by Dr-VSP significantly suppressed $Ca_V2.3$ currents, by $53{\pm}3%$. Next, dephosphorylation of both PI(4)P and $PI(4,5)P_2$ to PI by PJ translocation further decreased the current by up to $66{\pm}3%$. The results suggest that $Ca_V2.3$ currents are modulated by the $M_1$ receptor in a dual mode-that is, potentiation through the activation of PKC and suppression by the depletion of membrane $PI(4,5)P_2$. Our results also suggest that there is rapid turnover between PI(4)P and $PI(4,5)P_2$ in the plasma membrane.

• The Journal of Korean Society for Radiation Therapy
• /
• v.26 no.2
• /
• pp.297-303
• /
• 2014

Purpose : The purpose of this study is reproducibility evaluation of deep inspiration breath-hold(DIBH) technique by respiration data and heart position analysis in radiation therapy for Left Breast cancer patients. Materials and Methods : Free breathing(FB) Computed Tomography(CT) images and DIBH CT images of three left breast cancer patients were used to evaluate the heart volume and dose during treatment planing system( Eclipse version 10.0, Varian, USA ). The signal of RPM (Real-time Position Management) Respiratory Gating System (version 1.7.5, Varian, USA) was used to evaluate respiration stability of DIBH during breast radiation therapy. The images for measurement of heart position were acquired by the Electronic portal imaging device(EPID) cine acquisition mode. The distance of heart at the three measuring points(A, B, C) on each image was measured by Offline Review (ARIA 10, Varian, USA). Results : Significant differences were found between the FB and DIBH plans for mean heart dose (6.82 vs. 1.91 Gy), heart $V_{30}$ (68.57 vs. $8.26cm^3$), $V_{20}$ (76.43 vs. $11.34cm^3$). The standard deviation of DIBH signal of each patient was ${\pm}0.07cm$, ${\pm}0.04cm$, ${\pm}0.13cm$, respectively. The Maximum and Minimum heart distance on EPID images were measured as 0.32 cm and 0.00 cm. Conclusion : Consequently, using the DIBH technique with radiation therapy for left breast cancer patients is very useful to establish the treatment plan and to reduce the heart dose. In addition, it is beneficial to using the Cine acquisition mode of EPID for the reproducibility evaluation of DIBH.

• Progress in Medical Physics
• /
• v.24 no.3
• /
• pp.191-197
• /
• 2013

Upon radiation treatment, it is the important factor to monitor the patient's motion during radiation irradiated, since it can determine whether the treatment is successful. Thus, we have developed the system in which the patient's motion is monitored in real time and moving treatment position can be automatically corrected during radiation irradiation. We have developed the patient's position monitoring system in which the patient's position is three dimensionally identified by using two CCD cameras which are orthogonal located around the isocenter. This system uses the image pattern matching technique using a normalized cross-correlation method. We have developed the system in which trigger signal for beam on and off is generated by quantitatively analyzing the changes in a treatment position through delivery of the images taken from CCD cameras to the computer and the motor of moving couch can be controlled. This system was able to automatically correct a patient's position with the resolution of 0.5 mm or less.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.12 no.1
• /
• pp.33-38
• /
• 2008

Purpose: The Gated cardiac blood pool scan is non-invasive method that a quantitative evaluation of left ventricular function. Also this scan have shown the value of radionuclide ejection fraction measurements during the course of chemotherapy as a predictor of cardiac toxicity. Therefore a reliable method of monitoring its cardiotoxic effects is necessary. the purpose of this study is to minimize the overestimate of left ventricular ejection fraction (LVEF) by modified body position to reduce the influence of scattered rays from surrounding organs of the heart in the background region of interest. Materials and Methods: Gated cardiac blood pool scan using in vivo $^{99m}Tc$-red blood cell (RBC) was carried out in 20 patients (mean $44.8{\pm}8.6$ yr) with chemotherapy for a breast carcinoma. Data acquisition requires about 600 seconds and 24 frames of one heart cycle by the multigated acquisition mode, Synchronization deteriorates toward the end of the cycle and with the distance from the trigger signal (R-wave) by ECG gating. Gated cardiac blood pool scan was studied with conventional method (supine position and the detector head in $30-45^{\circ}$ left anterior oblique position and caudal $10-20^{\circ}$ tilt) and compared with modified method (left lateral flexion position with 360 mL of drinking water). LVEF analysis was performed by using the automatically computer mode. Results: The ROI counts of modified scan method were lower than LV conventional method ($1429{\pm}251$ versus $1853{\pm}243$, <0.01). And LVEF of modified method was also decrease compared with conventional method ($58.3{\pm}5.6%$ versus $65.3{\pm}6.1%$, <0.01). Imaging analysis indicated that stomach was expanded because of water and spleen position was changed to lateral inferior compared with conventional method. Conclusion: This study shows that the modified method in MUGA reduce the influence of scattered rays from surrounding organs. Because after change the body position to left lateral flexion and drinking water, the location of spleen, left lobe of liver and stomach had changed and they could escaped from background ROI. Therefore, modified method could help to minimize the overestimate LVEF (%).