• Journal of the Korea Society of Computer and Information
• /
• v.26 no.7
• /
• pp.37-44
• /
• 2021

Automatic classification of brain MRI images play an important role in early diagnosis of brain tumors. In this work, we present a deep learning-based brain tumor classification model in MRI images using ensemble of deep features. In our proposed framework, three different deep features from brain MR image are extracted using three different pre-trained models. After that, the extracted deep features are fed to the classification module. In the classification module, the three different deep features are first fed into the fully-connected layers individually to reduce the dimension of the features. After that, the output features from the fully-connected layers are concatenated and fed into the fully-connected layer to predict the final output. To evaluate our proposed model, we use openly accessible brain MRI dataset from web. Experimental results show that our proposed model outperforms other machine learning-based models.

• Journal of the Korean Society of Radiology
• /
• v.15 no.4
• /
• pp.565-573
• /
• 2021

The spatial distribution of electromagnetic fields in the human body can be adjusted by using high dielectric materials. This method has a complementary compared to other methods. However, it can be used as a powerful dielectric shimming tool in certain applications. It can be manufactured in a geometrically free shape and a pad manufactured according to the purpose can be applied without any change of the system. Especially in ultrahigh magnetic field (UHF) MRI, the clinical high dielectric pad used to increase the intensity of the transmit (B1+) and receive (B1-) fields, which has low sensitivity due to the high operating frequency, has great potential. In addition, there are few studies applied to UHF MRI. Therefore, in this study, a high dielectric material pad made of calcium titanate suspension was developed in the laboratory. And the signal increase of clinically useful images was confirmed in various protocols of UHF 7T MRI.

• The Transactions of the Korea Information Processing Society
• /
• v.5 no.10
• /
• pp.2704-2712
• /
• 1998

In this paper, a MRI image compression technique, which allocates bits effectively by using lossless coding for region having important infommtion to decide disease and lossy coding for the rest, is proposed. In the proposed method, for MHI images needed to rccognize disk disease, we recognizc region having important objects by using the characteristics of c1isease. As the recognized region is imrxlrtant to decide whether disease exists or not, it is compressed by lossless coding and the rest is compressed by lossy coding, Also for the region compressed by lossy coding, we can obtain fine reconstructed images without blocking effect by adopting fractal coding in wavelet transform domain.

• Korean Journal of Digital Imaging in Medicine
• /
• v.13 no.2
• /
• pp.55-58
• /
• 2011

It is important to study using experimental animals for research about stroke. Magnetic Resonance Imaging(MRI) is avaluable diagnosticmethods for stroke diagnosis. The purpose of this research is to know the Magnetic Resonance Imaging(MRI) and histopathological characteristics findings after induction of photothrombotic cerebral infarction in rat brain. Male Sprague-Dawley rats were anesthetized, Rose Bengal dye(20 mg/kg) was intravenously injected. The right sensonrimotor cortex of rat brain was exposed to cold light of 7 mm diameter at a position of 1 mm anterior and 3.5 mm lateral to bregma for 20 min. The post-infarction effects were monitored by T1 weighted and T2 weighted images of brain MRI. Histopathological changes were observed after Hematoxylin & Eosin staining. The lesion appeared clearly high signal intensity area on T2 weighted images(the major axis $7.04{\pm}0.11$ mm, the minor axis $3.08{\pm}0.04$ mm) and also H&E staining was same result. In conclusion, MRI was avaluable diagnostic methods for diagnosis and serial changes of stroke.

• Proceedings of the Korea Society for Simulation Conference
• /
• 2004.05a
• /
• pp.118-122
• /
• 2004

본 논문에서는 인체의 머리 부분을 촬영한 의료 영상에서 뇌 영역만을 분할하는 방법에 대해 제시하고자 한다. 뇌의 해부학적 구조 및 기능적 이상 부위를 파악할 경우에 영상 내에 함께 보여지는 두개골과 뇌척수액 등을 제외한 대뇌피질 영역을 분할하면 보다 효과적인 정보 분석 및 진단이 가능하게 된다. 본 시스템에서는 3단계 알고리즘을 제시한다. 첫 번째 단계에서는 영상 내에 존재하는 잡음을 제거하기 위한 필터링이고, 두 번째 단계에서는 필터링된 결과에 대한 영상분할을 수행하는 것이다 이 때 정확한 결과 도출을 위하여 사용자의 인터렉션이 들어가게 된다. 세번째 단계에서는 형태학적 방법을 이용하여 분할 결과를 보완한다. 본 연구를 위한 실험에는 자기 공명 촬영 영상(MRI: Magnetic Resonance Imaging), 단일 광전자 방출 단층 촬영영상(SPECT: Single Photon Emission Computed Tomography), 양전자 방출 단층 촬영영상(PET: Positron Emission Tomography) 등을 사용하였다. 본 시스템에서는 다양한 모달리티의 뇌 영상에서 대뇌피질 부분을 정확하게 영상 분할함으로써 뇌의 구조적 이상을 판단하기 위한 해부학적 정보 분석을 가능케 하고 있다. 뿐만 아니라 뇌 질환에 대한 정확한 진단 시뮬레이션도 가능하게 하고자 한다.

• Journal of radiological science and technology
• /
• v.39 no.3
• /
• pp.345-352
• /
• 2016

The purpose of this study was to needed basis of measure MRI CAD development for signal to noise ratio (SNR) by pulse sequence analysis from region of interest (ROI) in brain magnetic resonance imaging (MRI) contrast. We examined images of brain MRI contrast enhancement of 117 patients, from January 2005 to December 2015 in a University-affiliated hospital, Seoul, Korea. Diagnosed as one of two brain diseases such as meningioma and cysts SNR for each patient's image of brain MRI were calculated by using Image J. Differences of SNR among two brain diseases were tested by SPSS Statistics21 ANOVA test for there was statistical significance (p < 0.05). We have analysis socio-demographical variables, SNR according to sequence disease, 95% confidence according to SNR of sequence and difference in a mean of SNR. Meningioma results, with the quality of distributions in the order of T1CE, T2 and T1, FLAIR. Cysts results, with the quality of distributions in the order of T2 and T1, T1CE and FLAIR. SNR of MRI sequences of the brain would be useful to classify disease. Therefore, this study will contribute to evaluate brain diseases, and be a fundamental to enhancing the accuracy of CAD development.

• Investigative Magnetic Resonance Imaging
• /
• v.18 no.1
• /
• pp.25-33
• /
• 2014

Purpose : To investigate whether quantitative parameters derived from Diffusion-weighted magnetic resonance imaging (DW-MRI) correlate with those of Dynamic contrast-enhanced MRI (DCE-MRI). Materials and Methods: Thirteen patients with pathologically or clinically proven bony metastasis who had undergone MRI prior to treatment were included. The voxel size was $1.367{\times}1.367{\times}5mm$. A dominant tumor was selected and the apparent diffusion coefficient (ADC) value and DCE-MRI parameters were obtained by matching voxels. DCE-MRI data were analyzed yielding estimates of $K^{trans}$ (volume transfer constant) and $v_e$. (extravascular extracellular volume fraction). Statistical analysis of ADC, $K^{trans}$, and $v_e$ value was conducted using Pearson correlation analyses. Results: Fifteen lesions in pelvic bones were evaluated. Of these, 11 showed a statistically significant correlation (P<0.05) between ADC and $K^{trans}$. The ADC and $K^{trans}$ were inversely related in 7 lesions and positively related in 4 lesions. This did not depend on the primary cancer or site of metastasis. The ADC and $v_e$ of 9 lesions correlated significantly. Of these, 4 lesions were inversely related and 5 lesions were positively related. Conclusion: Unlike our theoretic hypothesis, there was no consistent correlation between ADC values and $K^{trans}$ or between ADC values and $v_e$ in metastatic bone tumors.

• Science of Emotion and Sensibility
• /
• v.20 no.4
• /
• pp.15-30
• /
• 2017

The purpose of this study was to verify the basic emotion theory based on the emotion-related research using functional brain imaging technology. For this purpose, a meta-analysis on the functional magnetic resonance imaging (fMRI) studies was performed. Six individual emotions-joy, happiness, fear, anger, disgust, sadness-were selected. In order to collect the fMRI data of individual emotions, we searched the electronic journals such as Medline, PsychInfo, PubMed for the past 10 years. fMRI experiment data aimed at healthy subjects for 6 emotions were collected, and only studies reported in Talairach or MNI standard coordinate system were included. In order to eliminate the difference between Talairach and MNI coordinate systems, we analyzed fMRI data based on the Talairach coordinate system. A meta-analysis using GingerALE 2.3 program adopting the activation likelihood estimates (ALE) techniques was performed. In this study, we confirmed that the individual emotions are associated with consistent and distinguishable regional brain responses within the framework of the basic emotion theory. The conclusion of this study of the brain areas associated with each individual emotional reaction was substantially consistent with the results of existing review articles. Finally, the limitations of this study and some suggestions for the future research were presented.

• Journal of Yeungnam Medical Science
• /
• v.18 no.1
• /
• pp.30-38
• /
• 2001

Background: The purpose of this study was to evaluate the usefulness of diffusion-weighted magnetic resonance imaging for monitoring the response to radiation therapy in metastatic bone marrow of the spines. Materials and Methods: Twenty-one patients with metastatic bone marrow of the spines were examined with MRI. Diffusion-weighted and spin-echo MRI were performed in 10 patients before and after radiation therapy with or without systemic chemotherapy, and performed in 11 patients after radiation therapy alone. Follow up spin-echo and diffusion-weighted MRI were obtained at 1 to 6 months after radiation therapy according to patients' condition. The diffusion-weighted imaging sequence was based on reversed fast imaging with steady-state precession (PSIF). Signal intensity changes of the metastatic bone marrows before and after radiation therapy on conventional spin-echo sequence MRI and diffusion-weighted MRI were evaluated. Bone marrow contrast ratios and signal-to-noise ratios before and after radiation therapy of diffusion- weighted MRI were analyzed. Results: All metastatic bone marrow of the spinal bodies were hyperintense to normal bone marrow of the spinal bodies on pretreatment diffusion-weighted MRI and positive bone marrow contrast ratios(p<0.001), and hypointense to normal spinal bodies on posttreatment diffusion-weighted MRI and negative bone marrow contrast ratios(p<0.001). The signal to noise ratios after treatment decreased comparing with those of pretreatment. Decreased signal intensity of the metastatic bone marrows on diffusion-weighted MRI began to be observed at average more than one month after the initiation of the radiation therapy. Conclusion: These results suggest that diffusion-weighted MRI would be an excellent method for monitoring the response to therapy of metastatic bone marrow of the spinal bodies, however, must be investigated in a larger series of patients with longer follow up period.

• Investigative Magnetic Resonance Imaging
• /
• v.9 no.1
• /
• pp.30-35
• /
• 2005

Purpose : The purpose of this study is to observe the blood oxygen level dependent (BOLD) contrast changes due to the reaction of human brain at a gustatory sense in response to a salty-taste stimulation. Materials and Methods : Twelve healthy, non-smoking, right-handed male subjects (mean age: 25.6, range: 23-28 years) participated in this salty-taste stimulus functional magnetic resonance (fMRI) study. MRI scans were performed with 1.57 GE Signa, using a multi-slice GE-EPI sequence according to a blood-oxy-gen-level dependent (BOLD) experiment paradigm. Scan parameters included matrix size $128\times128$, FOV 250 mm, TR 5000 msec, TE 60 msec, TH/GAP 5/2 mm. Sequential data acquisitions were carried out for 42 measurements with a repetition time of 5 sec for each taste-stimulus experiments. Analysis of fMRI data was carried out using SPM99 implemented in Matlab. NaCl solution $(3\%)$ was used as a salty stimulus. The task paradigm consisted of alternating rest-stimulus cycles (30-second rest, 15-second stimulus) for 210 seconds. During the stimulus period, NaCl-solution was presented to the subject's mouth through plastic tubes as a bolus of delivered every 5 sec using -processor controlled auto-syringe pump. Results : Insula, frontal opercular taste cortex, amygdala and orbitofrontal cortex (OFC) were activated by a salty-taste stimulation $(NaCl,\;3\%)$ in the fMRI experiments. And dosolateral prefrontal cortex (DLPFC) was also significantly responded to salty-taste stimuli. Activation areas of the right side hemisphere were more superior to the left side hemisphere. Conclusion : The results of this study well correspond to the fact that both insula, amygdala, OFC, DLPFC areas are established as taste cortical areas by neuronal recordings in primates. Authors found that laboratory-developed auto-syringe pump is suitable for gustatory fMRI study. Further research in this field will accelerate to inquire into the mechanism of higher order gustatory process.