• Proceedings of the Korea Information Processing Society Conference
• /
• 2009.11a
• /
• pp.343-344
• /
• 2009

The aim in this paper is to show how to extract scalp of a series of brain MR images by using region growing segmentation algorithm. Most researches are all forces on the segmentation of skull, gray matter, white matter and CSF. Prior to the segmentation of these inner objects in brain, we segmented the scalp and the brain from the MR images. The scalp mask makes us to quickly exclude background pixels with intensities similar those of the skull, while the brain mask obtained from our brain surface. We make use of connected threshold method (CTM) and confidence connected method (CCM). Both of them are two implementations of region growing in Insight Toolkit (ITK). By using these two methods, the results are displayed contrast in the form of 2D and 3D scalp images.

• Journal of Korea Multimedia Society
• /
• v.19 no.4
• /
• pp.736-746
• /
• 2016

In this paper, we propose an automatic prostate cancer detection method using position, signal intensity and texture feature based on SVM in multi-parametric MR images. First, to align the prostate on DWI and ADC map to T2wMR, the transformation parameters of DWI are estimated by normalized mutual information-based rigid registration. Then, to normalize the signal intensity range among inter-patient images, histogram stretching is performed. Second, to detect prostate cancer areas in T2wMR, SVM classification with position, signal intensity and texture features was performed on T2wMR, DWI and ADC map. Our feature classification using multi-parametric MR imaging can improve the prostate cancer detection rate on T2wMR.

• Journal of Multimedia Information System
• /
• v.8 no.2
• /
• pp.79-84
• /
• 2021

Detecting brain tumors of different sizes is a challenging task. This study aimed to identify brain tumors using detection algorithms. Most studies in this area use segmentation; however, we utilized detection owing to its advantages. Data were obtained from 64 patients and 11,200 MR images. The deep learning model used was RetinaNet, which is based on ResNet152. The model learned three different types of pre-processing images: normal, general histogram equalization, and contrast-limited adaptive histogram equalization (CLAHE). The three types of images were compared to determine the pre-processing technique that exhibits the best performance in the deep learning algorithms. During pre-processing, we converted the MR images from DICOM to JPG format. Additionally, we regulated the window level and width. The model compared the pre-processed images to determine which images showed adequate performance; CLAHE showed the best performance, with a sensitivity of 81.79%. The RetinaNet model for detecting brain tumors through deep learning algorithms demonstrated satisfactory performance in finding lesions. In future, we plan to develop a new model for improving the detection performance using well-processed data. This study lays the groundwork for future detection technologies that can help doctors find lesions more easily in clinical tasks.

• Investigative Magnetic Resonance Imaging
• /
• v.11 no.2
• /
• pp.127-132
• /
• 2007

Sinonasal neuroendocrine carcinoma is a rare disease, and reports focusing on the MR imaging findings of sinonasal neuroendocrine carcinoma are extremely rare. Threrefore we intend to report 2 cases of histologically confirmed neuroendocrine carcinoma. A 62-year-old man and a 74-year-old man are both presented with nasal bleeding. Computed tomography(CT) images of the 2 patients showed large, ill-defined masses in sinonasal cavities with adjacent bony destructions. MR images showed masses with isosignal intensity on Tl-weighted images and mixed iso- and high signal intensity on T2-weighted images. Post-contrast MR images showed heterogenous enhancement of masses with necrosis. Adjacent bony destructions were also noted on MR images. In both cases, peritumoral cystic lesions or mucoceles with high signal intensity on T1-weighted images were noted in sphenoid sinus. Both of the CT and MR imaging findings of the 2 patients were nonspecific which are usually seen in malignant tumor. But further study is needed for the significance of the peritumoral cystic areas adjacent the tumors.

• Investigative Magnetic Resonance Imaging
• /
• v.15 no.2
• /
• pp.102-109
• /
• 2011

Purpose : To retrospectively evaluate the usefulness of diffusion-weighted images, ADC maps and conventional MR images for determination of brain tumor consistency. Materials and Methods : Twenty-three patients with brain tumor underwent MR examinations with T1, T2 and diffusion-weighted images. Regions of interest (ROIs) were drawn in the tumors, and the measured signal intensities (SI) were normalized with the contralateral side. We evaluated the correlation between SI ratios from various images and tumor consistency assessed at surgery. In three patients with both cystic and solid components, each component was evaluated independently. Qualitatively observed SIs were also correlated with tumor consistency. Results : Statistical analysis revealed significant correlation between tumor consistency and ADC ratio (r = -0.586, p = 0.002), SI ratios on T2-weighted images (r = -0.497, p = 0.010), and observed SIs on T2-weighted images (r = -0.461, p = 0.018). The relative ratio of ADC value correlated with tumor consistency most strongly. Conclusion : The measured ratio of ADC, SI ratio and observed SI grade on T2-weighted images can provide valuable information about the consistency of brain tumor.

• The Transactions of the Korea Information Processing Society
• /
• v.7 no.1
• /
• pp.286-295
• /
• 2000

We present a novel statistically adaptive method using the Minimum Description Length(MDL) principle for unsupervised segmentation of magnetic resonance(MR) images. In the method, Markov random filed(MRF) modeling of tissue region accounts for random noise. Intensity measurements on the local region defined by a window are modeled by a finite Gaussian mixture, which accounts for image inhomogeneities. The segmentation algorithm is based on an iterative conditional modes(ICM) algorithm, approximately finds maximum ${\alpha}$ posteriori(MAP) estimation, and estimates model parameters on the local region. The size of the window for parameter estimation and segmentation is estimated from the image using the MDL principle. In the experiments, the technique well reflected image characteristic of the local region and showed better results than conventional methods in segmentation of MR images with inhomogeneities, especially.

• Journal of Biosystems Engineering
• /
• v.28 no.3
• /
• pp.253-260
• /
• 2003

In this study, the feasibility of magnetic resonance techniques for nondestructive internal quality evaluation of Korean red ginseng was examined. Relaxation time constants were measured using various grades of red ginsengs. Solid state magnetic resonance imaging technique was applied to image dried red ginsengs which have low moisture contents (about 13%). A 7 tesla magnetic resonance imaging system operating at a proton resonant frequency of 300 ㎒ was used for acquiring MR images of dried Korean red ginseng. The comparison test of cross cut digital images and magnetic resonance images of heaven grade, good grade with cavity inside, and good grade with white part inside red ginseng suggested the feasibility of the internal quality evaluation of Korean red ginsengs using MRI techniques. A good grade red ginseng included abnormal tissues such as cavities or white parts inside was observed by the signal intensity of MR image based on magnetic resonance properties of proton nucleus. Analysis on an one dimensional profile of acquired MR image of Korean red ginseng showed easy discrimination of normal and abnormal tissues. MR techniques suggested ways to detect internal defects of red ginsengs effectively.

• Proceedings of the Korea Multimedia Society Conference
• /
• 2002.11b
• /
• pp.97-100
• /
• 2002

뇌 MR 영상의 분석을 통해 질환을 자동적으로 진단하고 판별을 하기 위한 전처리 과정으로 정상인의 MR 영상 모델과 현재 고려되어지는 대상 영상과의 비교 작업이 요구된다. 이를 통해 보다 정확한 질병에 대한 근거를 제시함으로서 진단이 가능하게 된다. 이러한 비교 작업을 위해 우선적으로 해결해야 하는 것이 현재 대상 영상이 정상인의 MR 영상 시리즈 중 어느 위치의 영상과 일치하는 지를 판별해야 한다. 실질적으로 뇌 MR 시리즈는 영상의 특징에 따라 크게 몇 개의 그룹으로 분류된다. 그루핑 결과 뇌척수액이 존재하는 그룹은 또 다시 4 종류의 세부분류로 나누어지는데, 본 논문에서는 이 뇌 척수액의 모양에 따라 분류하는 알고리즘을 소개한다.

• Journal of Biomedical Engineering Research
• /
• v.21 no.3 s.61
• /
• pp.247-254
• /
• 2000

We propose an efficient MR image compression technique using fractal coding in wavelet transform domain. In the Proposed method , we construct significant coefficient trees with the absolute values of discrete wavelet transform coefficients and then perform the fractal coding with the information of significant coefficients having high energy. For MR images, most Pixels including background have very low gray level values, the number of significant coefficients is small. so we can expect high compression rate. In addition. since this method uses the fractal coding in wavelet transform domain, blocking artifact is reduced prominently and edges sensitive to human visual system are well preserved. As a result of computer simulation, we obtained the reconstructed images with better quality than those by JPEG at the low bit rates below 0.33[bpp].

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.46 no.6
• /
• pp.1-6
• /
• 2009

We propose an intensity-based image registration method for medical images. The proposed registration is performed by the use of a new measure based on the entropy of conditional probabilities. To achieve the registration, we define a modified conditional entropy (MCE) computed from the joint histograms for the area intensities of two given images. And we conduct experiments with our method as well as existing methods based on the sum of squared differences (SSD), normalized correlation coefficient (NCC), normalized mutual information (NMI) criteria. We evaluate the precision of SSD-, NCC-, MI- and MCE-based measurements by comparing the registration obtained from the same modality magnetic resonance (MR) images and the different modality transformed MR/transformed CT images. The experimental results show that the proposed method is faster and more accurate than other optimization methods.