• The KIPS Transactions:PartB
• /
• v.15B no.6
• /
• pp.507-516
• /
• 2008

Cardiac disease is the leading cause of death in the world. Quantification of cardiac function is performed by manually calculating blood volume and ejection fraction in routine clinical practice, but it requires high computational costs. In this study, an automatic left ventricle (LV) segmentation algorithm using short-axis cine cardiac MRI is presented. We compensate coil sensitivity of magnitude images depending on coil location, classify edge information after extracting edges, and segment LV by applying region-growing segmentation. We design a weighting function for intensity signal and calculate a blood volume of LV considering partial voxel effects. Using cardiac cine SSFP of 38 subjects with Cornell University IRB approval, we compared our algorithm to manual contour tracing and MASS software. Without partial volume effects, we achieved segmentation accuracy of $3.3mL{\pm}5.8$ (standard deviation) and $3.2mL{\pm}4.3$ in diastolic and systolic phases, respectively. With partial volume effects, the accuracy was $19.1mL{\pm}8.8$ and $10.3mL{\pm}6.1$ in diastolic and systolic phases, respectively. Also in ejection fraction, the accuracy was $-1.3%{\pm}2.6$ and $-2.1%{\pm}2.4$ without and with partial volume effects, respectively. Results support that the proposed algorithm is exact and useful for clinical practice.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.24 no.10B
• /
• pp.1986-1994
• /
• 1999

In this paper, a semi-automatic approach is adopted to extract a semantic object from real-world video sequences human-aided segmentation for the first frame and automatic tracking for the remaining frames. The proposed algorithm has a hierarchical structure using watershed algorithm. Each hierarchy consists of 3 basic steps: First, seeds are extracted from the simplified current frame. Second, region growing bv a modified watershed algorithm is performed to get over-segmented regions. Finally, the segmented regions are classified into 3 categories, i.e., inside, outside or uncertain regions according to region probability values, which are acquired by the probability map calculated from an estimated motion-vector field. Then, for the remaining uncertain regions, the above 3 steps are repeated at lower hierarchies with less simplified frames until every region is classified into a certain region. The proposed algorithm provides prospective results in studio-quality sequences such as 'Claire', 'Miss America', 'Akiyo', and 'Mother and daughter'.

• Proceedings of the Korean Information Science Society Conference
• /
• 1998.10c
• /
• pp.538-540
• /
• 1998

본 논문은 고개를 상/하로 끄덕이거나 좌/우로 가로 저어서 긍정과 부정을 구별하기 위한 것이다. 다시 말해서, 마우스나 키보드 대신에 머리의 움직임을 사용해서 '예/아니오'를 인식한다. 본 논문에서는 정규화된 칼라 공간(chromatic color space)과 조도(illumination)를 이용하여 얼굴 영역을 찾고 분할하는 자동 얼굴 영역 찾기와 영상차의 위치 비교와 움직임 량을 이용하여 우선 순위를 갖는 단순한 방향성을 구별하는 자동 의사 인식의 두 단계로 구성되어 있다. 이러한 단순한 방향성의 조합으로 '예/아니오'를 구분한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2011.04a
• /
• pp.426-429
• /
• 2011

영상처리를 이용한 영상간의 유사도 비교 기법은 영상의 검색 및 영상의 자동 인식 등을 위한 연구로 최근 각광받고 있다. 최근 영상 처리 기법은 화소의 질적 향상 및 처리시간 최적화, 효율적인 특정 요소의 추출 등 다양한 방법으로 시도되고 있다. 특히, 영상의 유사도 비교는 유사 영상 검색과 같은 경우에 많이 쓰인다. 영상의 유사도를 비교하기 위한 기법으로는 영상 데이터의 특징에 따라 대상 영역을 여러 영역으로 나누는 영역분할 기법과 군집화, 퍼지, 유전자 알고리즘 등이 있다. 본 논문에서는 영상을 HSV 색공간으로 변환한 후 색상 값에 대하여 전역 정렬 기법을 사용하는 유사도 측정 방법을 제시한다. 전역 정렬 기법은 유전자 서열 비교 기법 중 하나로서 두 유전체의 유사도를 측정하는데 사용된다. 유사도 측정 효율을 높이기 위해 색상 값을 8단계로 양자화하여 영상의 서열을 생성하였다. 실험결과 제시한 방법을 영상 회전이나 대칭, 글자 삽입 등의 간단한 연산에 크게 영향을 받지 않는 것으로 드러났다.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.47 no.6
• /
• pp.75-81
• /
• 2010

The segmentation of cartilage is crucial for the diagnose and treatment of osteoarthritis (OA), and has mostly been done manually by an expert, requiring a considerable amount of time and effort due to the thin shape and vague boundaries of the cartilage in MR (magnetic resonance) images. In this paper, we propose a fully automatic method to segment cartilage in a knee joint on MR images. The proposed method is based on a small number of manually segmented images as the training set and comprised of an initial per patch segmentation process and a global refinement process on the cumulative per patch results. Each patch for per patch segmentation is positioned by classifying the bone-cartilage interface on the pre-segmented bone surface. Next, the shape and intensity priors are constructed for each patch based on information extracted from reference patches in the training set. The ratio of influence between the shape and intensity priors is adaptively determined per patch. Each patch is segmented by graph cuts, where energy is defined based on constructed priors. Finally, global refinement is conducted on the global cartilage using the results of per patch segmentation as the shape prior. Experimental evaluation shows that the proposed framework provide accurate and clinically useful segmentation results.

• Proceedings of the Korean Information Science Society Conference
• /
• 2008.06a
• /
• pp.281-282
• /
• 2008

• Proceedings of the Korean Information Science Society Conference
• /
• 2010.06a
• /
• pp.244-245
• /
• 2010

• Database Research
• /
• v.34 no.3
• /
• pp.124-136
• /
• 2018

There have been active research activities to use neural networks to analyze OCT images and make medical decisions. One requirement for these approaches to be promising solutions is that the trained network must be generalized to new devices without a substantial loss of performance. In this paper, we use a deep convolutional neural network to distinguish AMD from normal patients. The network was trained using a data set generated from an OCT device. We observed a significant performance degradation when it was applied to a new data set obtained from a different OCT device. To overcome this performance degradation, we propose an image normalization method which performs segmentation of OCT images to identify the retina area and aligns images so that the retina region lies horizontally in the image. We experimentally evaluated the performance of the proposed method. The experiment confirmed a significant performance improvement of our approach.

• Proceedings of the IEEK Conference
• /
• 2001.06d
• /
• pp.97-100
• /
• 2001

In this paper, We propose a new semiautomatic segmentation method using spatio-temporal similarity. In the proposed scheme, segmentation is performed using gradual region merging and hi-direction at spatio-temporal refinement. Simulation results show the efficiency of the proposed method in semantic object extraction.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.36S no.2
• /
• pp.57-62
• /
• 1999

Estimation of the point spread function (PSF) is one of the main research topic of image processing, because it determines the performance of the auto-focusing system. In this paper, a new algorithm for PSF estimation is proposed, and its application to image restoration is also presented. The procedure for complete realization of the auto-focusing system consists of two steps: PSF estimation based on edge classification, and image restoration using the estimated PSF. More specifically, we divide imput image into multiple small image or block, estimate unit step response and average them on the blocks which contain edge, and estimate 2-dimensional isotropic PSF from the 1 dimensional step response. Finally we obtain in-focused image by using image restoration based on the estimated PSF.