Bum-Woo Park

Ulsan University Hospital, Urusan, Ulsan, South Korea

Are you Bum-Woo Park?

Claim your profile

Publications (17)25.61 Total impact

  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Lymph node (LN) status is an important parameter for determining the treatment strategy and for predicting the prognosis for patients with uterine cervical cancer. Computer-aided diagnosis (CAD) can be feasible for differentiating metastatic from non-metastatic lymph nodes in patients with uterine cervical cancer. To determine the usefulness of CAD that comprehensively evaluates MR images and clinical findings for detecting LN metastasis in uterine cervical cancer. In 680 LNs from 143 patients who underwent radical hysterectomy for uterine cervical cancer, the CAD system using the Bayesian classifier estimated the probability of metastasis based on MR findings and clinical findings. We compared the diagnostic accuracy for detecting metastatic LNs in the CAD and MR findings. Metastasis was diagnosed in 70 (12%) LNs from 34 (24%) patients. The area under ROC curves of CAD (0.924) was greater than those of the mean ADC (0.854), minimum ADC (0.849), maximum ADC (0.827), short-axis diameter (0.856) and long-axis diameter (0.753) (P < 0.05). The specificity and accuracy of the CAD (86%, 86%) were greater than those of the mean ADC (77%, 77%), maximum ADC (77%, 77%), minimum ADC (68%, 70%), and short-axis diameter (65%, 67%) (P < 0.05). CAD system can improve the diagnostic performance of MR for detecting metastatic LNs in uterine cervical cancer.
    Acta Radiologica 12/2011; 52(10):1175-83. · 1.33 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The purpose of this article is to assess the value of computer-aided diagnosis (CAD) for prostate cancer detection on dynamic contrast-enhanced MRI (DCE-MRI). DCE-MRI examinations of 42 patients with prostate cancer were used to generate perfusion parameters, including baseline and peak signal intensities, initial slope, maximum slope within the initial 50 seconds after the contrast injection (slope(50)), wash-in rate, washout rate, time to peak, percentage of relative enhancement, percentage enhancement ratio, time of arrival, efflux rate constant from the extravascular extracellular space to the blood plasma (k(ep)), first-order rate constant for eliminating gadopentetate dimeglumine from the blood plasma (k(el)), and constant depending on the properties of the tissue and represented by the size of the extravascular extracellular space (A(H)). CAD for cancer detection was established by comprehensive evaluation of parameters using a support vector machine. The diagnostic accuracy of single perfusion parameters was estimated using receiver operating characteristic analysis, which determined threshold and parametric maps for cancer detection. The diagnostic performance of CAD for cancer detection was compared with those of T2-weighted imaging (T2WI) and single perfusion parameter maps, using histologic results as the reference standard. The accuracy, sensitivity, and specificity of CAD were 83%, 77%, and 77%, respectively, in the entire prostate; 77%, 91%, and 64%, respectively, in the transitional zone; and 89%, 89%, and 89%, respectively, in the peripheral zone. Values for k(ep), k(el), initial slope, slope(50), wash-in rate, washout rate, and time to peak showed greater area under the curve values (0.803-0.888) than did the other parameters (0.545-0.665) (p < 0.01) and were compared with values for CAD. In the entire prostate, accuracy was greater for CAD than for all perfusion parameters or T2WI (63-77%); sensitivity was greater for CAD than for T2WI, initial slope, wash-in rate, slope(50), and washout rate (38-77%); and specificity was greater for CAD than for T2WI, k(ep), k(el), and time to peak (59-68%) (p < 0.05). CAD can improve the diagnostic performance of DCE-MRI in prostate cancer detection, which may vary according to zonal anatomy.
    American Journal of Roentgenology 11/2011; 197(5):1122-9. · 2.90 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: To evaluate the feasibility of flow-sensitive alternating inversion recovery (FAIR) for measuring blood flow in tumor models. In eight mice tumor models, FAIR and dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) was performed. The reliability for measuring blood flow on FAIR was evaluated using the coefficient of variation of blood flow on psoas muscle. Three regions of interest (ROIs) were drawn in the peripheral, intermediate, and central portions within each tumor. The location of ROI was the same on FAIR and DCE-MR images. The correlation between the blood flow on FAIR and perfusion-related parameters on DCE-MRI was evaluated using the Pearson correlation coefficient. The coefficient of variation for measuring blood flow was 9.8%. Blood flow on FAIR showed a strong correlation with Kep (r = 0.77), percent relative enhancement (r = 0.73), and percent enhancement ratio (r = 0.81). The mean values of blood flow (mL/100 g/min) (358 vs. 207), Kep (sec(-) (1)) (7.46 vs. 1.31), percent relative enhancement (179% vs. 134%), and percent enhancement ratio (42% vs. 26%) were greater in the peripheral portion than in the central portion (P < 0.01). As blood flow measurement on FAIR is reliable and closely related with that on DCE-MR, FAIR is feasible for measuring tumor blood flow.
    Journal of Magnetic Resonance Imaging 09/2010; 32(3):738-44. · 2.57 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The purpose of this study was to compare quantitative and visual assessments of regional heterogeneity of emphysema and to investigate the influence of regional heterogeneity on pulmonary function in smoking-related emphysema. We developed an automatic computerized algorithm to quantitatively assess heterogeneity in the upper-lower, anterior-posterior, and central-peripheral directions. The emphysema index was plotted with a linear function (emphysema index slopes: slope of emphysema index in upper-lower direction, slope of emphysema index in anterior-posterior direction, and slope of emphysema index in central-peripheral direction) for consecutive 1-pixel-thick slices using volumetric CT data of 59 patients (58 men and one woman; mean age, 65.7 years). Emphysema index was defined as the percentage area of lung with attenuation values below -950 HU. Visual assessment was performed using a 5-point scoring system. Quantitative and visual assessments were compared. Multiple linear regression was performed to evaluate the influence of emphysema index and emphysema index slopes on the pulmonary function test. Quantitative and visual assessments were significantly correlated in both upper-lower (r(2) = 0.40 and r(2) = 0.67 for observers 1 and 2, respectively) and central-peripheral (r(2) = 0.51 and r(2) = 0.47, respectively) directions. Multiple linear regression revealed that emphysema index, slope of emphysema index in upper-lower direction, and slope of emphysema index in anterior-posterior direction were independent determinants of forced expiratory volume in 1 second (FEV(1)) (r(2) = 0.30; p < 0.001). Emphysema index and slope of emphysema index in upper-lower direction were independent determinants of the ratio of FEV(1) to forced vital capacity (FEV(1)/FVC) (r(2) = 0.32; p < 0.001). In addition to higher emphysema index, lower and posterior lung dominance was associated with a decrease in FEV(1) and FEV(1)/FVC. Computerized, quantitative assessment using the emphysema index slope is comparable to visual assessment in the evaluation of regional heterogeneity of emphysema. In addition to the emphysema index, regional heterogeneity of smoking-related emphysema contributes to impairment of pulmonary function.
    American Journal of Roentgenology 03/2010; 194(3):W248-55. · 2.90 Impact Factor
  • Source
    Journal of Magnetic Resonance Imaging 05/2009; 29(5):1242. · 2.57 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The purpose of the study was to perform a node-by-node comparison of an ADC-based diagnosis and various size-based criteria on T2-weighted imaging (T2WI) with regard to their correlation with PET/CT findings in patients with uterine cervical cancer. In 163 patients with 339 pelvic lymph nodes (LNs) with short-axis diameter >5 mm, the minimum apparent diffusion coefficient (ADC), mean ADC, short- and long-axis diameters, and ratio of long- to short-axis diameters (L/S ratio) were compared in PET/CT-positive and -negative LNs. On PET/CT, 118 (35%) LNs in 58 patients were positive. The mean value of minimum and mean ADCs, short- and long-axis diameters, and L/S ratio were different in PET/CT-positive (0.6436 x 10(-3) mm(2)/s, 0.756 x 10(-3) mm(2)/s, 10.3 mm, 13.2 mm, 1.32, respectively) and PET/CT-negative LNs (0.8893 x 10(-3) mm(2)/s, 1.019 x 10(-3) mm(2)/s, 7.4 mm, 11.0 mm, 1.49, respectively) (P < 0.05). The Az value of the minimum ADC (0.864) was greater than those of mean ADC (0.836), short-axis diameter (0.764), long-axis diameter (0.640) and L/S ratio (0.652) (P < 0.05). The sensitivity and accuracy of the minimum ADC (86%, 82%) were greater than those of the short-axis diameter (55%, 74%), long-axis diameter (73%, 58%) and L/S ratio (52%, 66%) (P < 0.05). ADC showed superior correlation with PET/CT compared with conventional size-based criteria on T2WI.
    European Radiology 03/2009; 19(8):2024-32. · 4.34 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To determine the reference site for relative apparent diffusion coefficient (rADC) and to evaluate the benefit of rADC for detecting metastatic lymph nodes in uterine cervical cancer. Two observers independently measured ADCs in the spleen, liver, renal cortex, lumbar spine, lumbar spinal cord, and gluteus maximus on diffusion-weighted images (b value, 0 and 1000 mm/sec(2)) in 50 patients. The reference site for rADC was determined using the intra- and interobserver coefficient of variation (CV) of ADC in these organs. rADC was calculated by ADC(lesion)/ADC(reference site). The benefit of rADC over ADC was validated by comparing the area under the receiver operating curve for identifying metastatic lymph nodes in uterine cervical cancer in 130 patients. The renal cortex was determined to be the reference site for rADC, as its CVs (intraobserver, 5%-7%; interobserver, 5%) were less than those of the other organs (P < 0.05). The ADC and rADC of metastatic lymph nodes (n = 29, ADC, 0.7483 x 10(-3) mm(2)/sec; rADC, 0.3832) were less than those of nonmetastatic lymph nodes (n = 229, ADC, 0.9960 x 10(-3) mm(2)/sec; rADC, 0.5383) (P < 0.05). The area under the receiver operating characteristics curve for differentiating metastatic from nonmetastatic lymph nodes was greater for rADC (0.914; 95% confidence interval [CI], 0.872-0.945) than for ADC (0.872; 95% CI, 0.825-0.910) (P = 0.007). The renal cortex is an appropriate reference site for rADC and rADC may improve the accuracy for diagnosing metastatic lymph nodes in uterine cervical cancer.
    Journal of Magnetic Resonance Imaging 01/2009; 29(2):383-90. · 2.57 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: This study was designed to develop an automated system for quantification of various regional disease patterns of diffuse lung diseases as depicted on high-resolution computed tomography (HRCT) and to compare the performance of the automated system with human readers. A total of 600 circular regions-of-interest (ROIs), 10 pixels in diameter, were utilized. The 600 ROIs comprised 100 ROIs that represented six typical regional patterns (normal, ground-glass opacity, reticular opacity, honeycombing, emphysema, and consolidation). The ROIs were used to train the automated classification system based on the use of a Support Vector Machine classifier and 37 features of texture and shape. The performance of the classification system was tested with a 5-fold cross-validation method. An automated quantification system was developed with a moving ROI in the lung area, which helped classify each pixel into six categories. A total of 92 HRCT images obtained from patients with different diseases were used to validate the quantification system. Two radiologists independently classified lung areas of the same CT images into six patterns using the manual drawing function of dedicated software. Agreement between the automated system and the readers and between the two individual readers was assessed. The overall accuracy of the system to classify each disease pattern based on the typical ROIs was 89%. When the quantification results were examined, the average agreement between the system and each radiologist was 52% and 49%, respectively. The agreement between the two radiologists was 67%. An automated quantification system for various regional patterns of diffuse interstitial lung diseases can be used for objective and reproducible assessment of disease severity.
    Korean journal of radiology: official journal of the Korean Radiological Society 01/2009; 10(5):455-63. · 1.32 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE To perform node-by-node comparison of diffusion-weighted imaging (DWI) and size-based criteria on T2-weighted imaging (T2WI) for detection of pelvic lymph node (LN) metastasis in cervical cancer patients using PET/CT as the reference standard. METHOD AND MATERIALS We included 163 patients with 340 pelvic LN with the short-axis diameter equal to or greater than 5 mm. A threshold ADC was obtained by receiver-operating-characteristic analysis of the testing set (n = 50) and subsequently applied to the training set (n =113). In the training set, two reviewers measured the short- and long-axis diameters on T2WI and apparent-diffusion coefficient (ADC) in pelvic LNs. The ADC, short-axis diameter, and long-to-short diameter (L/S)-ratio were compared between metastatic and non-metastatic LNs. Using the results of PET/CT as the reference standard, the detection of LN metastasis was compared on per-node and per-patient bases between DWI and and the followng size-based criteria on T2WI : short-axis diameter (8 mm, 9 mm, and 10 mm) and a long-to-short diameter (L/S)-ratio of 1.25. RESULTS The mean ADC, short-axis diameter, and L/S-ratio were significantly different between metastatic (0.6500, 10.5 cm, and 1.23, respectively) and non-metastatic LNs (0.8689, 7.6 cm, and 1.43, respectively) (P<.001). From the receiver-operating-characteristic analysis (Az = .911) of the testing set, an ADC of 0.710 was demonstrated as the optimal threshold. Per-node sensitivity and specificity for detecting metastatic LNs by ADC were 82% (73/89) and 76% (116/152), respectively. The sensitivity of ADC was significantly greater than those of the short-axis diameter criteria (37%-64%) and L/S-ratio criterion (58%) (P <.0001). Per-patient accuracy of ADC (92%) was greater than all short-axis diameter criteria (66%-81%) and the L/S-ratio criterion (73%) (P<.05). CONCLUSION DWI can improve the diagnostic accuracy of MR in the detection of metastatic pelvic LNs in cervical cancer patients, as it is more sensitive than T2WI. CLINICAL RELEVANCE/APPLICATION With the addition of diffusion-weighted imaging, MR examination can significantly improve the assessment of tumor lymphatic spread and thereby improve the care of patients with cervical cancer.
    Radiological Society of North America 2008 Scientific Assembly and Annual Meeting; 11/2008
  • [Show abstract] [Hide abstract]
    ABSTRACT: To evaluate diagnostic performance of apparent diffusion coefficient (ADC) in differentiating prostate cancer from noncancerous tissue according to anatomical region. In 47 patients with diffusion-weighted-MR (b-value, 0 and 1000 sec/mm2) on a 1.5 T unit, ADCs were measured in prostate cancer and in three noncancerous tissues (transitional zone, peripheral zone, and prostatic base). Diagnostic performance of ADC for differentiating cancer from noncancerous tissue was evaluated using receiver-operating-characteristics (ROC) analysis. Mean ADC of prostate cancer (0.963x10(-3) mm2/s) was lower than those of all noncancerous tissues (P<0.001). In noncancerous tissue, ADC differed according to anatomical region (peripheral zone, 1.572x10(-3) mm2/sec; transitional zone, 1.441x10(-3) mm2/sec; prostatic base, 1.146x10(-3) mm2/sec) (P<0.01). ADC was lower in prostate cancer than in all noncancerous tissues in 34 (72%) patients. Area under the ROC curve for differentiating cancer from noncancerous tissue in prostatic base (0.725) was less than those for differentiating cancer from noncancerous tissue in peripheral (0.952) and transitional zones (0.906) (P<0.05). Sensitivity differed according to anatomical region (peripheral zone, 98%; transitional zone, 82%; prostatic base, 66%) (P<0.05). Variable ADC in noncancerous tissue according to anatomical region may limit diagnostic performance of ADC for cancer detection.
    Journal of Magnetic Resonance Imaging 10/2008; 28(5):1173-9. · 2.57 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: To investigate the feasibility of diffusion-weighted imaging (DWI) in the differentiation of metastatic from nonmetastatic lymph nodes. In 125 patients who underwent lymph node dissection for uterine cervical cancer, DWI was performed at b value of 0 and 1000 s/mm2. By referring to the surgical maps of the pelvic lymph nodes, the apparent diffusion coefficient (ADC) was compared in the metastatic and nonmetastatic lymph nodes, and receiver-operating-characteristics analysis was performed to evaluate the diagnostic performance of the ADC in differentiating metastatic from nonmetastatic lymph nodes. The ADC were significantly lower in the metastatic lymph nodes (0.7651x10(-3) mm2/s+/-0.1137) than in the nonmetastatic lymph nodes (1.0021x10(-3) mm2/s+/-0.1859; P<0.001). The area-under-the-curve of ADC for differentiating metastatic from nonmetastatic lymph nodes, was 0.902. The sensitivity and specificity of ADC for differentiating metastatic from nonmetastatic lymph nodes, were 87% for the ADC and 80%, respectively. DWI is feasible for differentiating metastatic from nonmetastatic lymph nodes in patients with uterine cervical cancer.
    Journal of Magnetic Resonance Imaging 09/2008; 28(3):714-9. · 2.57 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE To evaluate the usefulness of automated classification system using perfusion parameters of the Brix two compartment model in the differentiation of prostate cancer from non-cancerous tissue. METHOD AND MATERIALS In 18 patients with radical prostatectomy and dynamic contrast enhanced MR imaging (1.5 T; Gradient echo; temporal resolution, 3 sec; 200 dynamics; slice thickness, 5 mm), eight parameters of Brix two compartment model (Kep, Kel, AH, time of arrival, time to peak, plateau signal, base signal, and root mean squared error) were measured at 153 locations, including 51 prostate cancer, 51 non-cancerous transitional zone and 51 non-cancerous peripheral zone tissues. For automated classification, a support vector machine (SVM), of which the parameters were optimized, was employed. To assess the cross-validation of the system, a leave-one-out method was employed, which split the data into training set (n = N-1) and test (n = 1) and repeated alternate learning until all data were tested. RESULTS The accuracy for differentiating cancer from non-cancerous tissue was similar (range, 77%-88%) in eight parameters. The result from a sequential forward selection method showed that all parameters are necessary for the classification. The sensitivity and specificity of the automated classification system were 91% and 93% for differentiating cancer from all non-cancerous tissue, 88% and 88% for differentiating cancer from non-cancerous transitional zone tissue, and 96% and 97% for differentiating cancer from non-cancerous peripheral zone tissue. The specificity for differentiating cancer from non-cancerous peripheral zone tissue was greater than that for differentiating cancer from non-cancerous transitional zone tissue (P = .03). CONCLUSION The automatic classification system using perfusion parameters of Brix two compartment model may be useful for differentiating prostate cancer from non-cancerous tissue. CLINICAL RELEVANCE/APPLICATION Automatic classification system using perfusion parameters of Brix two compartment model can be used for prostate cancer detection and localization.
    Radiological Society of North America 2007 Scientific Assembly and Annual Meeting; 11/2007
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE To verify the usefulness of relative apparent-diffusion-coefficient (rADC) and to determine the reference organ for calculating rADC. METHOD AND MATERIALS Fifty-five patients with uterine cervix cancer and diffusion-weighted MR (b value, 0 and 1000) on 1.5 T unit, were enrolled. Two radiologists measured ADC values at three points in each of various organs, including the spleen, liver, right kidney, lumbar spine, spinal cord, and right gluteus muscle. Reliability for measuring ADC in organs was evaluated by coefficient-of-variation (CV). rADC was calculated by the follows: rADC = ADC of lesion/ADC of reference organ. ADC and rADC were measured in uterine cervix cancer and adjacent endometrium. ROC analysis was used for comparing the performance of ADC and rADC in differentiating cancer tissue from endometrium. RESULTS The mean CV in two radiologists was lower in right kidney (6.2% and 4.7%) than in the other organs (spleen, 11.6% and 7.5%; liver, 15.7% and 15.4%; lumbar spine, 16.8% and 14.0%; spinal cord, 11.3% and 11.2%; and right gluteus muscle, 16.2% and 18.4%). ADC and rADC were 1.012 ± 0.171 and 0.525 ± 0.070 in cancer tissue and 1.328 ± 0.275 and 0.687 ± 0.112 in endometrium, respectively, The area under ROC for differentiating cervix cancer from endometrium was greater in rADC (0.919) than in ADC (0.847) (P = .009). CONCLUSION rADC is superior to ADC in differentiating cancer tissue from adjacent non-cancerous tissue and the right kidney is adequate for the reference organ for calculating rADC. CLINICAL RELEVANCE/APPLICATION The rADC, a normalized form of ADC, can reduce individual variation of ADC, and the right kidney is the optimal internal organ for the reference ADC of rADC.
    Radiological Society of North America 2007 Scientific Assembly and Annual Meeting; 11/2007
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE To investigate the added value of diffusion-weighted imaging (DWI) for discrimination of metastatic lymph nodes in preoperative MR staging of rectal cancer. METHOD AND MATERIALS Institutional review board approval and informed consent were obtained. Eighty-four consequent patients with rectal cancer underwent preoperative MRI using a 1.5T system and surgical resection. In addition to conventional T1 and T2-weighted imaging, DWI was obtained in the axial plane using EPI-STIR sequence with high b-value of 1000s/mm2. Two experienced radiologists performed blind review of conventional MR imaging (c-MRI) and DWI focusing on identification of malignant lymph nodes. In each c-MRI and DWI, the reader’s confidence for depiction of malignant node was graded using a five point scale (1, 90%). All lymph nodes greater than 4 mm in size were recorded of size, location, and apparent diffusion coefficient (ADC) values. With surgical and pathologic reports as reference standards, metastatic lymph nodes were matched by size and location. RESULTS Among 524 matched lymph nodes, metastasis was confirmed in 35 nodes (6.7%) (12 of 19 nodes larger than 10 mm in size, 19 of 153 nodes of 5-10 mm, 4 out of 152 nodes less than 5 mm). The ADC was significantly greater in metastatic lymph nodes (1.31 ± 0.29 X 10-3 mm2/s) than benign nodes (1.09  0.51 X 10-3 mm2/s) (p<0.05). On receiver operating characteristic (ROC) analysis, diagnostic performance was significantly improved with DWI (0.899 vs. 0.783, p<0.05). The differences in the area under the ROC curve between both imaging was maximum in nodes of 5-10 mm (p<0.05). CONCLUSION In MR evaluation of rectal cancer, the application of DWI can provide improved diagnostic accuracy in discriminate metastatic lymph nodes with higher diagnostic confidence. CLINICAL RELEVANCE/APPLICATION In MR evaluation of rectal cancer, the application of DWI can provide improved diagnostic accuracy in discriminate metastatic lymph nodes with higher diagnostic confidence.
    Radiological Society of North America 2007 Scientific Assembly and Annual Meeting; 11/2007
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE To improve classification accuracy and reduce feature dimension by a feature selection method using hybrid genetic algorithm (hGA) for the differentiation among diffuse infiltrative lung diseases (DILD). METHOD AND MATERIALS Six-hundred circular regions of interest (ROI) with 10, 20 and 30 pixel diameter, comprising of each 100 ROIs representing six regional disease patterns (normal, ground-glass opacity, reticular opacity, honeycombing, emphysema, and consolidation) were marked by an experienced radiologist from selected HRCT images. For each ROI, 37 features from texture (histogram, gradient, run length encoding and co-occurrence matrix) and shape (lower attenuation area and tophat transform) analyses were extracted. Extracted features were applied to a classifier for the training of system. For automated classification, a support vector machine (SVM), of which the parameters were optimized, and 20-folding cross-validation method were employed. For feature selection, hGA and typical GA (tGA) were used. hGA consisted of GA and a local optimizer. After generations of chromosomes, local optimizer (Sequential Backward Selection) was employed to adjust best-fitted local optimum of feature combination. To evaluate the usefulness of hGA, the performance and number of selected features of hGA and tGA was compared. RESULTS Overall accuracy and number of selected features of hGA were 89.013%, 8 in ROI with 10 pixel diameter, 95.670%, 9 in ROI with 20 pixel diameter, and 97.031%, 15 in ROI with 30 pixel diameter, respectively. Otherwise, overall accuracy and number of selected features of tGA were 88.876%, 15 in ROI with 10 pixel diameter, 94.521%, 15 in ROI with 20 pixel diameter, and 96.670%, 17 in ROI with 30 pixel diameter, respectively. In 20 ROI, SD of hGA and tGA were 0.34 and 0.31, respectively. CONCLUSION hGA reduced features dimension and equal accuracy than tGA. In addition, hGA performed good results than tGA in small ROI. CLINICAL RELEVANCE/APPLICATION This result can be used to set the effective feature selection method based on the ROI size for developing automatic classification system of differentiation between diffuse infiltrative lung diseases
    Radiological Society of North America 2007 Scientific Assembly and Annual Meeting; 11/2007
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE To investigate the usefulness of diffusion-weighted-imaging (DWI) in the differentiation of metastatic from non-metastatic lymph nodes in patients with uterine cervical cancer. METHOD AND MATERIALS In 108 patients who underwent radical hysterectomy and lymph node dissection for uterine cervical cancer, we performed diffusion-weighted imaging with background body signal suppression (b value, 0 and 1,000 s/mm2). By referring to the surgical maps of the pelvic lymph nodes, the apparent diffusion coefficient (ADC) and relative ADC (rADC, ADClymph node/ADCkidney) were measured in pelvic lymph nodes on both sides. The ADC and rADC were compared in the metastatic and non-metastatic lymph nodes, and receiver-operating-characteristics analysis was performed to evaluate the diagnostic performance of the ADC and rADC in differentiating metastatic from non-metastatic lymph nodes. RESULTS The ADC and rADC were significantly lower in the metastatic lymph nodes (0.763 ± 0.154, 0.388 ± 0.076) than in the non-metastatic lymph nodes (0.996 ± 0.199, 0.535 ± 0.102) (P < .001). The area-under-the-curve for differentiating metastatic from non-metastatic lymph nodes was greater in the rADC (0.917) than in the ADC (0.868) (P = .003). The sensitivity and specificity for differentiating metastatic from non-metastatic lymph nodes were 81% and 85% for the ADC and 87% and 93% for the rADC, respectively. CONCLUSION : DWI is useful for differentiating metastatic from non-metastatic lymph nodes in patients with uterine cervical cancer, while the rADC provides better diagnostic performance than the ADC. CLINICAL RELEVANCE/APPLICATION Diffusion weighted MR can be used to differentiate metastatic from non metastatic lymph nodes.
    Radiological Society of North America 2007 Scientific Assembly and Annual Meeting; 11/2007
  • [Show abstract] [Hide abstract]
    ABSTRACT: To find optimal binning, variable binning size linear binning (LB) and non-linear binning (NLB) methods were tested. In case of small binning size (Q

Publication Stats

184 Citations
25.61 Total Impact Points

Institutions

  • 2009–2011
    • Ulsan University Hospital
      Urusan, Ulsan, South Korea
  • 2010
    • Korea Basic Science Institute KBSI
      Sŏul, Seoul, South Korea
  • 2008–2009
    • Asan Medical Center
      • Department of Radiology
      Seoul, Seoul, South Korea
    • University of Ulsan
      • Department of Radiology
      Ulsan, Ulsan, South Korea