• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• 제28권1호
• /
• pp.1-21
• /
• 2024

This review comprehensively explores the evolution, theoretical underpinnings, variations, and applications of diffusion models. Originating as a generative framework, diffusion models have rapidly ascended to the forefront of machine learning research, owing to their exceptional capability, stability, and versatility. We dissect the core principles driving diffusion processes, elucidating their mathematical foundations and the mechanisms by which they iteratively refine noise into structured data. We highlight pivotal advancements and the integration of auxiliary techniques that have significantly enhanced their efficiency and stability. Variants such as bridges that broaden the applicability of diffusion models to wider domains are introduced. We put special emphasis on the ability of diffusion models as a crucial foundation model, with modalities ranging from image, 3D assets, and video. The role of diffusion models as a general foundation model leads to its versatility in many of the downstream tasks such as solving inverse problems and image editing. Through this review, we aim to provide a thorough and accessible compendium for both newcomers and seasoned researchers in the field.

• The Journal of Korean Physical Therapy
• /
• 제32권6호
• /
• pp.388-390
• /
• 2020

Objectives: The nigrostriatal tract (NST) connect from the substantia nigra pars compacta to the striatum. A few previous studies have reported on the NST in the Parkinson's disease using a proboblistic tractography method. However, no study has been conducted for identification of the NST using streamline DTT technique. In the current study, we used streamline DTI technique to investigate the reconstruction method and characteristics of the NST in normal subjects. Methods: Eleven healthy subjects were recruited in this study. The NST from the substantia nigra of the midbrain and the striatum of basal ganglia was reconstructed using DTI data. Fractional anisotropy, apparent diffusion coefficient (ADC) values and fiber numbers of the NST were measured. Results: In all subjects, the NST between the substantia nigra of the midbrain and the striatum. Mean values for FA, ADC, and tract volume were 0.460, 0.818, and 154.3 in the right NST, and 0.485, 0.818, and 176.3 in the left NST respectively. Conclusions: we reconstructed the NRT from the substantia nigra of the midbrain and the striatum of the basal ganglia using streamline tractography method. We believe that the findings and the proposed streamline reconstruction method of this study would be useful in future researches on the NST of the human brain.

• Investigative Magnetic Resonance Imaging
• /
• 제4권1호
• /
• pp.34-41
• /
• 2000

목적: 두개강내 종양의 감별진단에 있어서 확산강조자기공명영상(DWi: diffusion weighted MR imaging)의 임상적 유용성을 알아 보고자 하였다. 대상 및 방법: 19예의 두개강내 종양(전이암 10예, 고등급 교종 4예, 저등급 성상세포종 4예, 핍지교종 1예)을 대상으로 1.5T 장치를 이용하여 통상적인 자기공명영상과 EPI기법을 사용한 DWI(TR/TE=6500/107, b value 1000)를 얻었다. DWI에서 종%$\varepsilon$을 고형성분, 괴사나 낭성 부위, 주위 부종으로 나누어 신호강도(뇌설과 뇌실질을 기준으로 5등급으로 나눔)와 벙변과 반 대쪽 정상 뇌설질의 상대적 신호강도비(SIR: signal intensity ratio)를 구하였다. 이렇게 얻어진 종양간의 신호강도와 신호강도비의 차이를 독립표본 T 검정을 이용하여 비교 분석하였다. 결과: DWI에서 고형성분의 경우 전이암과 고등급 교종은 전예에서 뇌실질보다 고신호강도를 보 였으며, 저등급 성상세포종과 핍지교종은 뇌실질과 등신호 또는 약간 높은 신호강도를 보였다. 각각의 고형성분에서 평균 신호강도비는 전이암 1.52, 고등급 교종 1.48, 저등급 성상세포종 1.16, 핍지교종 1.31로 측정되어서 전반적으로 악성도가 증가할수록 높은 신호강도비를 보였다(P(0.05). 종양 주위 부종은 전이암 10예와 고등급 교종 4예에서 관찰되었으며 이중 전이암 7예, 고등급 교종 2예에서는 뇌실질과 유사한 신호강도를 보인 반면 나머지 5예들에서는 뇌실 질보다 높은 신호강도로 관찰되었다. 부종에서의 평균 신호강도비는 전이암이 1.14, 고등급 교종 1.31로 전이암에서 낮게 관찰되었지만 통계적 의의는 없었다(p >0.05). 괴사나 낭성부위의 평균 신호강도비는 0.63이였다. 조영증강되지 않았던 6예의 고형성분중 3예는 주위 부종보다 고신호강도를보여 종양의 경계가통상적인 자기공명영상에 비해 우수하였다. 결론: DWI에서 뇌종양의 고형성분의 신호강도는 종양의 악성도가 높을수록 고신호강도를 보였고 종양 주위 부종은 DWI기법에 기인된 고유효과인 "T2 shine through effect"와 혈관성 부종의 정도 간의 우열에 따라 다양한 소견을 보였으며, 또한 DWI에서 종양과 주위 부종과의 관계를 좀더 명확하게 구분 할 수 있었다.

• 대한디지털의료영상학회논문지
• /
• 제11권2호
• /
• pp.51-57
• /
• 2009

The purpose of this study would evaluate if having clinical effects on diffusion image with quantitative analysis through ADC values of brain's normal tissue and lesions before and after contrast injections using a 3.0T. From November in 2007 until December in 2008, a total of 32 patient was performed on 3.0T(Signa Excite, GE Medical System, USA) with the normal or lesions in the patient who requests diffusion weighted image with 8channel head coil. The pulse sequence was used with spin echo EPI(TR: 10000msec, TE: 72.2 msec, Matrix: 128*128, FOV: 240 mm, NEX: 1, diffusion direction: 3, b-value: 1000). Measurement results of ADC values on lesions, CSF, white matter, gray matter, lesions after contrast injection were measured less 75% than before contrast injection, infarction: 100%, CSF: 78%(high), white matter: 71.4%(low), gray matter: 50%(high, low). The results of paired t-test on the deference of ADC values which statically is significant in three(lesions, CSF, white matter)regions except for white matter(p<0.05). Quantitative analysis of lesions, CSF, white matter, gray matter have difference on all regions. ADC values were low in lesions and white matter, normal CSF after contrast injection commonly is high than before contrast injection, ADC values which white matter were high and low (50:50) after contrast injection. 3.0T diffusion weighted image clinically supposed that performing DWI examination after contrast injection was not desirable because of having effects on brain tissue.

• 한국의학물리학회지:의학물리
• /
• 제30권1호
• /
• pp.22-31
• /
• 2019

Purpose: To develop a new diffusion-based functional MRI (fMRI) sequence to generate apparent diffusion coefficient (ADC) maps in single excitation and evaluate the contribution of b0 signal on neuronal changes. Materials and Methods: A diffusion-based fMRI sequence was designed with single measurement that can acquire images of three directions at a time, obtaining $b=0s/mm^2$ during the first baseline condition (b0_b), followed by 107 diffusion-weighted imaging (DWI) with $b=600s/mm^2$ during the baseline and visual stimulation conditions, and another $b=0s/mm^2$ during the last activation condition (b0_a). ADC was mapped in three different ways: 1) using b0_b (ADC_b) for all time points, 2) using b0_a (ADC_a) for all time points, and 3) using b0_b and b0_a (ADC_ba) for baseline and stimulation scans, respectively. The fMRI studies were conducted on the brains of 16 young healthy volunteers using visual stimulations in a 3T MRI system. In addition, the blood oxygen level dependent (BOLD) fMRI was also acquired to compare it with diffusion-based fMRI. A sample t-test was used to investigate the voxel-wise average between the subjects. Results: The BOLD data consisted of only activated voxels. However, ADC_ba data was observed in both deactivated and activated voxels. There were no statistically significant activated or deactivated voxels for DWI, ADC_b, and ADC_a. Conclusions: With the new sequence, neuronal activations can be mapped with visual stimulation as compared to the baseline condition in several areas in the brain. We showed that ADC should be mapped using both DWI and b0 images acquired with the same conditions.

• Korean Journal of Radiology
• /
• 제22권7호
• /
• pp.1142-1151
• /
• 2021

Objective: To evaluate microstructural damage in high myopia (HM) patients using 3T diffusion kurtosis imaging (DKI). Materials and Methods: This prospective study included 30 HM patients and 33 age- and sex-matched healthy controls (HCs) with DKI. Kurtosis parameters including kurtosis fractional anisotropy (FA), mean kurtosis (MK), axial kurtosis (AK), and radial kurtosis (RK) as well as diffusion metrics including FA, mean diffusivity, axial diffusivity (AD), and radial diffusivity derived from DKI were obtained. Group differences in these metrics were compared using tract-based spatial statistics. Partial correlation analysis was used to evaluate correlations between microstructural changes and disease duration. Results: Compared to HCs, HM patients showed significantly reduced AK, RK, MK, and FA and significantly increased AD, predominately in the bilateral corticospinal tract, right inferior longitudinal fasciculus, superior longitudinal fasciculus, inferior fronto-occipital fasciculus, and left thalamus (all p < 0.05, threshold-free cluster enhancement corrected). In addition, DKI-derived kurtosis parameters (AK, RK, and MK) had negative correlations (r = -0.448 to -0.376, all p < 0.05) and diffusion parameter (AD) had positive correlations (r = 0.372 to 0.409, all p < 0.05) with disease duration. Conclusion: HM patients showed microstructural alterations in the brain regions responsible for motor conduction and vision-related functions. DKI is useful for detecting white matter abnormalities in HM patients, which might be helpful for exploring and monitoring the pathogenesis of the disease.

• 대한방사선기술학회지:방사선기술과학
• /
• 제32권3호
• /
• pp.313-323
• /
• 2009

확산강조영상 (diffusion weighted image, DWI)은 급성 뇌경색, 뇌종양, 뇌백질 질환, 뇌 막질의 확산 정도 등 여러 뇌질환의 진단을 획기적으로 향상시켰으며 그 활용도가 증가하고 있다. 본 연구는 $10{\sim}60$대 환자들의 뇌를 대상으로 두 기법간의 신호강도, 현성확산계수의 평균치를 측정하였다. 그 결과, 확산강조영상에서의 신호강도 평균값은 편도체부 (amygdala)가 가장 높고, 뇌척수액(cerebrospinal fluid)에서 가장 낮았다. 현성확산계수의 평균값은 뇌척수액이 높고, 교뇌 (pons)가 낮게 측정되었다. 확산강조 신호강도와 현성확산계수의 평균값은 $T_2^*$-DW-EPI 기법이 FLAIR-DW-EPI 기법보다 높고, b-value의 변화에 따른 평균값은 두 기법의 b-value에 모두 반비례하였다. 또한 뇌경색환자의 뇌의 시간 경과에 따른 분석결과, 초급성뇌경색 환자의 일반적인 MR 영상에서는 병변부분이 명확하지 않았으나 확산강조영상에서는 고신호강도로 나타났다. 출혈성 뇌경색, 급성 뇌경색 등 여러질환별로 분석한 결과 그 두 기법의 특성에 따라 신호강도의 값이 차이가 클수록 현성확산계수는 낮게 나타났다. 결론적으로 뇌 질환이 자주 발생되는 부위와 뇌 질환의 확산강조 신호강도 및 현성확산계수 값은 b-value의 변환과 영상기법에 따라 각각 다르게 나타났다. 이러한 정량적인 결과를 바탕으로 보다 안정적인 기법과 적절한 b-value 값을 이용하여 검사를 한다면 여러 뇌의 질환 및 병변 등을 발견, 판독하는 것뿐만 아니라 정상부위나 질환에 따른 기법별 신호의 인지를 통한 정확한 질병 진단과 치료에 중요한 의미가 있다고 사료된다.

• Investigative Magnetic Resonance Imaging
• /
• 제11권2호
• /
• pp.110-118
• /
• 2007

목적: 확산텐서자기공명영상(DT-MRI: Diffusion Tensor Image)을 이용하여 확산의 주축을 구성하는 세 성분에 대한 고유치 (eigefvalue)의 영상을 구현해 보고자 하였다. 대상 및 방법: 고유치 영상을 구현하기 위해서 3.0 테슬러 MRI(Magnetic Resonance Imaging)를 이용하여 확산텐서영상을 얻었으며, Moore-Penrose pseudo-inverse 방법과 SVD(single value decomposition) 방법을 이용하여 확산 주축을 계산하였다. 이 과정을 픽셀단위로 반복적으로 계산하여 새로운 확산 주축 영상들을 만들었으며, 이 확산 주축 영상들과 분할 비등방성 영상의 관계를 조사하였다. 결과: 확산텐서영상 기법으로 얻어진 확산텐서영상을 이용하여, 세 방향의 확산 주축에 대한 고유치 영상을 구성하였으며, 고유치 영상들과 분할 비등방성 영상을 함께 분석함으로써, 뇌의 해부학적 구조물에 따른 분할 비등방성 값의 차이를 확인할 수 있었다. 또한, 확산 주축에 대한 고유치의 변화에 대한 컴퓨터 모의실험에서, 변화하는 고유치에 따른 분할 비등방성 값의 변동 추이를 알아볼 수 있었다. 그리고 확산 주축의 크기가 비등방성을 좌우하는 것이 아니라, 세 확산 주축의 조합으로 비등방성의 정도를 표현한다는 것을 확인할 수 있었다. 확산 주축 방향의 고유치들을 분리하여 영상화 함으로써, 뇌의 병변에 의한 비등방성의 변화의 원인이 확산 주축의 어떠한 변화에 의해 발생하는 것인지 확인할 수 있을 것으로 기대된다.

• 한국컴퓨터정보학회:학술대회논문집
• /
• 한국컴퓨터정보학회 2021년도 제64차 하계학술대회논문집 29권2호
• /
• pp.701-702
• /
• 2021

The purpose of the study was to identify fiber changes induced by electrical stimulation of a certain neural substrate in the rat brain. In the stimulation group, the peripheral nerve was injured and the brain area associated to inhibit sensory information was electrically stimulated. There were sham and sham stimulation groups as controls. Then high-field diffusion tensor imaging (DTI) was acquired. 35 features were taken from the DTI measures from 7 different brain pathways. To compare the efficacy of the classification for 3 animal groups, the linear regression analysis (LDA) and the machine learning technique (MLP) were applied. It was found that the testing accuracy by MLP was about 77%, but that of accuracy by LDA was much higher than MLP. In conclusion, machine learning algorithm could be used to identify and predict the changes of the brain white matter in some situations. The limits of this study will be discussed.

• Investigative Magnetic Resonance Imaging
• /
• 제25권1호
• /
• pp.23-34
• /
• 2021

Purpose: Differentiating between glioblastoma and solitary metastasis is very important for the planning of further workup and treatment. We assessed the ability of various morphological parameters using conventional MRI and diffusion-based techniques to distinguish between glioblastomas and solitary metastases in tumoral and peritumoral regions. Materials and Methods: We included 38 patients with solitary brain tumors (21 glioblastomas, 17 solitary metastases). To find out if there were differences in the morphologic parameters of enhancing tumors, we analyzed their shape, margins, and enhancement patterns on postcontrast T1-weighted images. During analyses of peritumoral regions, we assessed the extent of peritumoral non-enhancing lesion on T2- and postcontrast T1-weighted images. We also aimed to detect peritumoral neoplastic cell infiltration by visual assessment of T2-weighted and diffusion-based images, including DWI, ADC maps, and exponential DWI, and evaluated which sequence depicted peritumoral neoplastic cell infiltration most clearly. Results: The shapes, margins, and enhancement patterns of tumors all significantly differentiated glioblastomas from metastases. Glioblastomas had an irregular shape, ill-defined margins, and a heterogeneous enhancement pattern; on the other hand, metastases had an ovoid or round shape, well-defined margins, and homogeneous enhancement. Metastases had significantly more extensive peritumoral T2 high signal intensity than glioblastomas had. In visual assessment of peritumoral neoplastic cell infiltration using T2-weighted and diffusion-based images, all sequences differed significantly between the two groups. Exponential DWI had the highest sensitivity for the diagnosis of both glioblastoma (100%) and metastasis (70.6%). A combination of exponential DWI and ADC maps was optimal for the depiction of peritumoral neoplastic cell infiltration in glioblastoma. Conclusion: In the differentiation of glioblastoma from solitary metastatic lesions, visual morphologic assessment of tumoral and peritumoral regions using conventional MRI and diffusion-based techniques can also offer diagnostic information.