• Korean Journal of Radiology
• /
• v.20 no.8
• /
• pp.1275-1284
• /
• 2019

Objective: To develop algorithms using convolutional neural networks (CNNs) for automatic segmentation of acute ischemic lesions on diffusion-weighted imaging (DWI) and compare them with conventional algorithms, including a thresholding-based segmentation. Materials and Methods: Between September 2005 and August 2015, 429 patients presenting with acute cerebral ischemia (training:validation:test set = 246:89:94) were retrospectively enrolled in this study, which was performed under Institutional Review Board approval. Ground truth segmentations for acute ischemic lesions on DWI were manually drawn under the consensus of two expert radiologists. CNN algorithms were developed using two-dimensional U-Net with squeeze-and-excitation blocks (U-Net) and a DenseNet with squeeze-and-excitation blocks (DenseNet) with squeeze-and-excitation operations for automatic segmentation of acute ischemic lesions on DWI. The CNN algorithms were compared with conventional algorithms based on DWI and the apparent diffusion coefficient (ADC) signal intensity. The performances of the algorithms were assessed using the Dice index with 5-fold cross-validation. The Dice indices were analyzed according to infarct volumes (< 10 mL, ≥ 10 mL), number of infarcts (≤ 5, 6-10, ≥ 11), and b-value of 1000 (b1000) signal intensities (< 50, 50-100, > 100), time intervals to DWI, and DWI protocols. Results: The CNN algorithms were significantly superior to conventional algorithms (p < 0.001). Dice indices for the CNN algorithms were 0.85 for U-Net and DenseNet and 0.86 for an ensemble of U-Net and DenseNet, while the indices were 0.58 for ADC-b1000 and b1000-ADC and 0.52 for the commercial ADC algorithm. The Dice indices for small and large lesions, respectively, were 0.81 and 0.88 with U-Net, 0.80 and 0.88 with DenseNet, and 0.82 and 0.89 with the ensemble of U-Net and DenseNet. The CNN algorithms showed significant differences in Dice indices according to infarct volumes (p < 0.001). Conclusion: The CNN algorithm for automatic segmentation of acute ischemic lesions on DWI achieved Dice indices greater than or equal to 0.85 and showed superior performance to conventional algorithms.

• Asian Pacific Journal of Cancer Prevention
• /
• v.16 no.15
• /
• pp.6401-6406
• /
• 2015

Background: It is sometimes difficult to assess patients who have multiple hilar and mediastinal lymph nodes (MHMLN) with FDG accumulation in PET-CT. Since it is uncertain whether diffusion-weighted magnetic resonance imaging (DWI) is useful in the assessment of such patients, its diagnostic performance was assessed. Materials and Methods: Twenty-three patients who had three or more stations of hilar and mediastinal lymph nodes with SUVmax of 3 or more in PET-CT were included in this study. Results: For diagnosis of disease, there were 20 malignancies (lung cancers 17, malignant lymphomas 2 and metastatic lung tumor 1), and 3 benign cases (sarcoidosis 2 and benign disease 1). For diagnosis of lymph nodes, there were 7 malignancies (metastasis of lung cancer 7 and malignant lymphoma 1) and 16 benign lymphadenopathies (pneumoconiosis/silicosis 7, sarcoidosis 4, benign disease 4, and atypical lymphocyte infiltration 1). The ADC value ($1.57{\pm}0.29{\times}10^{-3}mm^2/sec$) of malignant MHMLN was significantly lower than that ($1.99{\pm}0.24{\times}10^{-3}mm^2/sec$) of benign MHMLN (P=0.0437). However, the SUVmax was not significantly higher ($10.0{\pm}7.34$ as compared to $6.38{\pm}4.31$) (P=0.15). The sensitivity (86%) by PET-CT was not significantly higher than that (71%) by DWI for malignant MHMLN (P=1.0). The specificity (100%) by DWI was significantly higher than that (31%) for benign MHMLN (P=0.0098). Furthermore, the accuracy (91%) with DWI was significantly higher than that (48%) with PET-CT for MHMLN (P=0.0129). Conclusions: Evaluation by DWI for patients with MHMLN with FDG accumulation is useful for distinguishing benign from malignant conditions.

• Journal of the Korean Society of Radiology
• /
• v.10 no.1
• /
• pp.7-13
• /
• 2016

The goal of this study is to investigate corpus callosum and both internal capsule changes with the internet addiction disorder compared to control group using MR diffusion tensor imaging. A total of 22 teenager volunteers who had 10 high-risk group with internet addiction and 12 normal control group were conducted for this study. Imaging was conducted on a 3 T using a EPI sequence. Image evaluation was analysed of the FA, ADC($10^{-3}mm^2/s$), length(mm). We did select ROI for image tracking on corpus callosum of 5 and including 2(internal capsule). The data from these ROIs were compared statistically among the groups using independent t-test, correlation coefficient. There were significant inter-group differences(p<0.05) among FA, ADC($10^{-3}mm^2/s$) and length(mm). And also significantly negative correlations were fond between FA values of corpus callosum and IAD scale(p=0.000). DTI was shown significant changes of FA and ADC, LNF values in IAD compared to control group. Therefore, our results may provided clinical information for brain wite matter functions.

• Investigative Magnetic Resonance Imaging
• /
• v.6 no.2
• /
• pp.166-172
• /
• 2002

Purpose : To evaluate the usefulness of diffusion-weighted imaging(DWI) and quantitative apparent diffusion coefficient (ADC) maps in the patients with spinal cord infarction. Materials and methods : We studied 6 patients presented symptoms with spinal cord infarction, retrospectively (3 men and 3 women). We obtained multi-shot echo planar-based, DWI using 1.5T MR scanner at 5.4 mean days after the onset of ischemic symptoms. In six patients, signal intensity was acquired at conventional b value $1000s/\textrm{mm}^2$). The ADC value for the normal spinal cord and for infarcted lesions was measured from the trace ADC maps by using regions of interest positioned over the spinal cord. We analyzed signal intensity of lesion on MRI and DWI, and compared with ADC values in infarcted lesions and normal site. Results : T1-weighted MR image showed isosignal intensity in four of six patients and low signal intensity in two of six. T2-weighted MR image demonstrated high signal intensity in all of six. All DWI were considered to be diagnostic. All of six depicted a bright signal intensity on DWI. ADC values of infarcted lesion were measured lower than that of normal spinal cord on ADC map. The differences in ADC values between infarcted and normal spinal cord were significantly different (p＜0.05). Conclusion : It is possible to obtain DWI and ADC map of the spinal cord and DWI may be useful in the early diagnosis and localization of lesion site in patients with spinal cord infarction.

• Korean Journal of Radiology
• /
• v.22 no.6
• /
• pp.922-930
• /
• 2021

Objective: It is uncertain why a b-value range of 1500-2000 s/mm² is optimal. This study was aimed at qualitatively and quantitatively analyzing the optimal b-value range of synthetic diffusion-weighted imaging (sDWI) for evaluating prostatic index lesions. Materials and Methods: This retrospective study included 92 patients who underwent DWI and targeted biopsy for magnetic resonance imaging (MRI)-suggested index lesions. We generated sDWI at a b-value range of 1000-3000 s/mm² using dedicated software and true DWI data at b-values of 0, 100, and 1000 s/mm². We hypothesized that lesion conspicuity would be best when the background (i.e., MRI-suggested benign prostatic [bP] and periprostatic [pP] regions) signal intensity (SI) is suppressed and becomes homogeneous. To prove this hypothesis, we performed both qualitative and quantitative analyses. For qualitative analysis, two independent readers analyzed the b-value showing the best visual conspicuity of an MRI-suggested index lesion. For quantitative analysis, the readers assessed the b-value showing the same bP and pP region SI. The 95% confidence interval (CI) or interquartile range of qualitatively and quantitatively selected optimal b-values was assessed, and the mean difference between qualitatively and quantitatively selected b-values was investigated. Results: The 95% CIs of optimal b-values from qualitative and quantitative analyses were 1761-1805 s/mm² and 1640-1771 s/mm² (median, 1790 s/mm² vs. 1705 s/mm²; p = 0.003) for reader 1, and 1835-1895 s/mm² and 1705-1841 s/mm² (median, 1872 s/mm² vs. 1763 s/mm²; p = 0.022) for reader 2, respectively. Interquartile ranges of qualitatively and quantitatively selected optimal b-values were 1735-1873 s/mm² and 1573-1867 s/mm² for reader 1, and 1775-1945 s/mm² and 1591-1955 s/mm² for reader 2, respectively. Bland-Altman plots consistently demonstrated a mean difference of less than 100 s/mm² between qualitatively and quantitatively selected optimal b-values. Conclusion: b-value range showing a homogeneous background signal may be optimal for evaluating prostatic index lesions on sDWI. Our qualitative and quantitative data consistently recommend b-values of 1500-2000 s/mm².

• Korean Journal of Radiology
• /
• v.22 no.10
• /
• pp.1680-1689
• /
• 2021

Objective: To investigate the diagnostic yield of diffusion-weighted imaging (DWI) in patients with transient global amnesia (TGA) and identify significant parameters affecting diagnostic yield. Materials and Methods: A systematic literature search of the MEDLINE and EMBASE databases was conducted to identify studies that assessed the diagnostic yield of DWI in patients with TGA. The pooled diagnostic yield of DWI in patients with TGA was calculated using the DerSimonian-Laird random-effects model. Subgroup analyses were also performed of slice thickness, magnetic field strength, and interval between symptom onset and DWI. Results: Twenty-two original articles (1732 patients) were included. The pooled incidence of right, left, and bilateral hippocampal lesions was 37% (95% confidence interval [CI], 30-44%), 42% (95% CI, 39-46%), and 25% (95% CI, 20-30%) of all lesions, respectively. The pooled diagnostic yield of DWI in patients with TGA was 39% (95% CI, 27-52%). The Higgins I² statistic showed significant heterogeneity (I² = 95%). DWI with a slice thickness ≤ 3 mm showed a higher diagnostic yield than DWI with a slice thickness > 3 mm (pooled diagnostic yield: 63% [95% CI, 53-72%] vs. 26% [95% CI, 16-40%], p < 0.01). DWI performed at an interval between 24 and 96 hours after symptom onset showed a higher diagnostic yield (68% [95% CI, 57-78%], p < 0.01) than DWI performed within 24 hours (16% [95% CI, 7-34%]) or later than 96 hours (15% [95% CI, 8-26%]). There was no difference in the diagnostic yield between DWI performed using 3T vs. 1.5T (pooled diagnostic yield, 31% [95% CI, 25-38%] vs. 24% [95% CI, 14-37%], p = 0.31). Conclusion: The pooled diagnostic yield of DWI in TGA patients was 39%. DWI obtained with a slice thickness ≤ 3 mm or an interval between symptom onset and DWI of > 24 to 96 hours could increase the diagnostic yield.

• Korean Journal of Radiology
• /
• v.23 no.11
• /
• pp.1078-1088
• /
• 2022

Objective: To develop and validate a model using radiomics features from apparent diffusion coefficient (ADC) map to diagnose local tumor recurrence in head and neck squamous cell carcinoma (HNSCC). Materials and Methods: This retrospective study included 285 patients (mean age ± standard deviation, 62 ± 12 years; 220 male, 77.2%), including 215 for training (n = 161) and internal validation (n = 54) and 70 others for external validation, with newly developed contrast-enhancing lesions at the primary cancer site on the surveillance MRI following definitive treatment of HNSCC between January 2014 and October 2019. Of the 215 and 70 patients, 127 and 34, respectively, had local tumor recurrence. Radiomics models using radiomics scores were created separately for T2-weighted imaging (T2WI), contrast-enhanced T1-weighted imaging (CE-T1WI), and ADC maps using non-zero coefficients from the least absolute shrinkage and selection operator in the training set. Receiver operating characteristic (ROC) analysis was used to evaluate the diagnostic performance of each radiomics score and known clinical parameter (age, sex, and clinical stage) in the internal and external validation sets. Results: Five radiomics features from T2WI, six from CE-T1WI, and nine from ADC maps were selected and used to develop the respective radiomics models. The area under ROC curve (AUROC) of ADC radiomics score was 0.76 (95% confidence interval [CI], 0.62-0.89) and 0.77 (95% CI, 0.65-0.88) in the internal and external validation sets, respectively. These were significantly higher than the AUROC values of T2WI (0.53 [95% CI, 0.40-0.67], p = 0.006), CE-T1WI (0.53 [95% CI, 0.40-0.67], p = 0.012), and clinical parameters (0.53 [95% CI, 0.39-0.67], p = 0.021) in the external validation set. Conclusion: The radiomics model using ADC maps exhibited higher diagnostic performance than those of the radiomics models using T2WI or CE-T1WI and clinical parameters in the diagnosis of local tumor recurrence in HNSCC following definitive treatment.

• Journal of Biomedical Engineering Research
• /
• v.37 no.2
• /
• pp.84-89
• /
• 2016

In this study, we proposed ameliorated method for region of interest (ROI) study to improve its accuracy using partial volume effect (PVE). PVE which arose in volumetric images when more than one tissue type occur in a voxel, could be used to reduce an amount of gray matter and cerebrospinal fluid within ROI of diffusion tensor image (DTI). In order to define ROIs, individual b0 image was spatially aligned to the JHU DTI-based atlas using linear and non-linear registration (http://cmrm.med.jhmi.edu/). Fractional anisotropy (FA) and mean diffusivity (MD) maps were estimated by fitting diffusion tensor model to each image voxel, and their mean values were computed within each ROI with PVE threshold. Participants of this study consisted of 20 healthy controls, 27 Alzheimer's disease and 27 normal-pressure hydrocephalus patients. The result showed that the mean FA and MD of each ROI were increased and decreased respectively, but standard deviation was significantly decreased when PVE was applied. In conclusion, the proposed method suggested that PVE was indispensable to improve an accuracy of DTI ROI study.

• The Journal of Korean Physical Therapy
• /
• v.32 no.6
• /
• 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.

• Health Policy and Management
• /
• v.18 no.4
• /
• pp.85-110
• /
• 2008

In this study, we explore the long-term demand forecasting of high-price medical equipments based on logistic and Bass diffusion model. We analyze the specific pattern of each equipment's diffusion curve by interpreting the parameter estimates of Bass diffusion model. Our findings are as follows. First, ultrasonic imaging system, CT are in the stage of maturity and so, the future demands of them are not too large. Second, medical image processing unit is between growth stage and maturity stage and so, the demand is expected to increase considerably for two or three years. Third, MRI is in the stage of take-off and Mammmography X-ray system is in the stage of maturity but, estimates of the potential number of adopters based on logistic model is considerably different to that based on Bass diffusion model. It means that additional data for these two equipments should be collected and analyzed to obtain the reliable estimates of their demands. Fourth, medical image processing unit have the largest q value. It means that the word-of-mouth effect is important in the diffusion of this equipment. Fifth, for MRI and Ultrasonic system, q/p values have the relatively large value. It means that collective power has an important role in adopting these two equipments.