Daoyu Hu

Classification of Hepatic Tissues from CT Images Based on Texture Features and Multiclass Support Vector Machines

A computer-aided diagnosis (CAD) of X-ray Computed Tomography (CT) liver images with contrast agent injection is presented. Regions of interests (ROIs) on CT liver images are defined by experienced radiologists. For each ROI, texture features based on first order statistics (FOS), spatial gray level dependence matrix (SGLDM), gray level run length matrix (GLRLM) and gray level difference matrix (GLDM) are extracted. Support vector machine (SVM) is originally for binary classification. In order to classify hepatic tissues from CT images into primary hepatic carcinoma, hemangioma and normal liver, we utilize two methods to construct multiclass SVMs: one-against-all (OAA), one-against-one (OAO) and compare their performance. The result shows that a total accuracy rate of 97.78% is obtained with the multiclass SVM using the OAO method. Our study has some practical significance for clinical diagnosis.

Depiction of Transplant Renal Vascular Anatomy and Complications: Unenhanced MR Angiography by Using Spatial Labeling with Multiple Inversion Pulses

PURPOSE To evaluate the ability to depict anatomy and complications of renal vascular transplant with unenhanced magnetic resonance (MR) angiography with spatial labeling with multiple inversion pulses (SLEEK) and to compare the results with color Doppler (CD) ultrasonography (US), digital subtraction angiography (DSA), and intraoperative findings. MATERIALS AND METHODS This study was approved by the institutional review board, and written informed consent was received before examination. Seventy-five patients who underwent renal transplantation were examined with unenhanced MR angiography with SLEEK and CD US. DSA was performed in 15 patients. Surgery was performed in eight patients. The ability of SLEEK to show transplant renal vascular anatomy and complications was evaluated by two experienced radiologists who compared the results with CD US, DSA, and intraoperative findings. RESULTS Patients successfully underwent SLEEK MR angiography. Transplant renal vascular anatomy was assessed in 87 arteries and 78 veins. Renal vascular complications from transplantation were diagnosed in 23 patients, which included 14 with arterial stenosis, three with arterial kinking, two with arteriovenous fistulas, two with venous stenosis, one with pseudoaneurysms, and one with fibromuscular dysplasia. Three patients had two renal transplants and nine patients had nine accessory renal arteries. More accessory renal arteries were detected with SLEEK than with CD US. Correlation was excellent between the stenosis degree with SLEEK and DSA (r = 0.96; P < .05). For those with significant artery stenosis (>50% narrowing) proved with DSA (n = 7) or surgery (n = 3), positive predictive value was 91% (10 of 11). CONCLUSION Unenhanced MR angiography with SLEEK preliminarily proved to be a reliable diagnostic method for depiction of anatomy and complications of renal vascular transplant. It may be used for evaluation of patients with renal transplant, and in particular for those with renal insufficiency.

Assessing the Early Response of Advanced Cervical Cancer to Neoadjuvant Chemotherapy Using Intravoxel Incoherent Motion Diffusion-weighted Magnetic Resonance Imaging: A Pilot Study

Background: Diffusion-weighted imaging (DWI) with the intravoxel incoherent motion (IVIM) model has shown promising results for providing both diffusion and perfusion information in cervical cancer; however, its use to predict and monitor the efficacy of neoadjuvant chemotherapy (NACT) in cervical cancer is relatively rare. The study aimed to evaluate the use of DWI with IVIM and monoexponential models to predict and monitor the efficacy of NACT in cervical cancer. Methods: Forty-two patients with primary cervical cancer underwent magnetic resonance exams at 3 time points (pre-NACT, 3 weeks after the first NACT cycle, and 3 weeks after the second NACT cycle). The response to treatment was determined according to the response evaluation criteria in solid tumors 3 weeks after the second NACT treatment, and the subjects were classified as two groups: responders and nonresponders groups. The apparent diffusion coefficient (ADC), true diffusion coefficient (D), perfusion-related pseudo-diffusion coefficient (D*), and perfusion fraction (f) values were determined. The differences in IVIM-derived variables and ADC between the different groups at the different time points were calculated using an independent samples t-test. Results: The D and ADC values were all significantly higher for the responders than for the nonresponders at all 3 time points, but no significant differences were observed in the D* and f values. An analysis of the receiver operating characteristic (ROC) curves indicated that a D value threshold <0.93 × 10−3 mm2/s and an ADC threshold <1.11 × 10−3 mm2/s could differentiate responders from nonresponders at pre-NACT time point, yielding area under the curve (AUC) of which were 0.771 and 0.806, respectively. The ROC indicated that the AUCs of D and ADC at the 3 weeks after the first NACT cycle and 3 weeks after the second NACT cycle were 0.823, 0.763, and 0.787, 0.794, respectively. The AUC values of D and ADC at these 3 time points were not significantly different (P = 0.641, 0.512, and 0.547, respectively). Conclusions: D and ADC values may be useful for predicting and monitoring the efficacy of NACT in cervical cancer. An IVIM model may be equal to monoexponential model in predicting and monitoring the efficacy of NACT in cervical cancer.

Adrenal and nephrogenic hypertension: an image quality study of low tube voltage, low-concentration contrast media combined with adaptive statistical iterative reconstruction.

The aim of this study was to investigate the effect of using low tube voltage, low‐concentration contrast media and adaptive statistical iterative reconstruction (ASIR) for reducing the radiation and iodine contrast doses in adrenal and nephrogenic hypertension patients.

Did low tube voltage CT combined with low contrast media burden protocols accomplish the goal of "double low" for patients? An overview of applications in vessels and abdominal parenchymal organs over the past 5 years.

Imaging communities have already reached a consensus that the radiation dose of computed tomography (CT) should be reduced as much as reasonably achievable to lower population risks. Increasing attention is being paid to iodinated contrast media (CM) induced nephrotoxicity (CIN); a decrease in the intake of iodinated CM is required by increasingly more radiologists. Theoretically, the radiation dose varies with the tube current time and square of the tube voltage, with higher iodine contrast at low photon energies (Huda et al. [2000] Radiology, 21 7, 430–435).The use of low tube voltage is a promising strategy to reduce both the radiation dose and CM burden. The term ‘double low’ has been coined to describe scanning protocols that reduce radiation dose and iodine intake synchronously. These protocols are becoming increasingly popular in the clinical setting.

Assessment of different mathematical models for diffusion‐weighted imaging as quantitative biomarkers for differentiating benign from malignant solid hepatic lesions

To quantitatively compare the monoexponential, biexponential, and stretched‐exponential diffusion‐weighted imaging (DWI) models in differentiating benign from malignant solid hepatic lesions. The institutional review board approved this retrospective study and waived the informed consent requirement. A total of 188 patients with 288 hepatic lesions included 202 malignant lesions and 86 benign lesions were assessed (confirmed by pathology or clinical follow‐up for 6 months). All patients underwent hepatic 3.0‐T MRI, including multi‐b DWI that used 12 b values. The ADC, Dp, Dt, perfusion fraction (fp), α, and DDC values for normal liver, benign liver lesions, and malignant liver lesions were calculated. Independent sample t tests were used for comparisons. The diagnostic performance of the parameters was evaluated using ROC analysis. The AUC value for each model was also calculated. The value of Dp was significantly lower in benign lesions than in normal hepatic parenchyma while others were significantly higher (P < .001). Whereas Values of Dt and α in malignant hepatic lesions were significantly higher than in normal hepatic parenchyma (P < .001), and the Dp value was significantly lower (P < .001). Values of ADC, fp, DDC, and α for malignant hepatic lesions were significantly lower than those for benign hepatic lesions (P < .001). ROC analysis showed that the diagnostic value of the biexponential model of normal hepatic parenchyma vs benign hepatic lesions and normal hepatic parenchyma vs malignant hepatic lesions was high (0.946 and 0.876, respectively). In the differential diagnosis of benign and malignant hepatic lesions, DDC had the highest AUC value (0.819). The biexponential and stretched‐exponential DWI may provide additional information and improve the differential diagnosis of benign and malignant hepatic lesions compared with the monoexponential DWI.

Subtype Differentiation of Small (≤ 4 cm) Solid Renal Mass Using Volumetric Histogram Analysis of DWI at 3-T MRI

OBJECTIVE The purpose of this article is to evaluate the utility of volumetric histogram analysis of apparent diffusion coefficient (ADC) derived from reduced-FOV DWI for small (≤ 4 cm) solid renal mass subtypes at 3-T MRI. MATERIALS AND METHODS This retrospective study included 38 clear cell renal cell carcinomas (RCCs), 16 papillary RCCs, 18 chromophobe RCCs, 13 minimal fat angiomyolipomas (AMLs), and seven oncocytomas evaluated with preoperative MRI. Volumetric ADC maps were generated using all slices of the reduced-FOV DW images to obtain histogram parameters, including mean, median, 10th percentile, 25th percentile, 75th percentile, 90th percentile, and SD ADC values, as well as skewness, kurtosis, and entropy. Comparisons of these parameters were made by one-way ANOVA, t test, and ROC curves analysis. RESULTS ADC histogram parameters differentiated eight of 10 pairs of renal tumors. Three subtype pairs (clear cell RCC vs papillary RCC, clear cell RCC vs chromophobe RCC, and clear cell RCC vs minimal fat AML) were differentiated by mean ADC. However, five other subtype pairs (clear cell RCC vs oncocytoma, papillary RCC vs minimal fat AML, papillary RCC vs oncocytoma, chromophobe RCC vs minimal fat AML, and chromophobe RCC vs oncocytoma) were differentiated by histogram distribution parameters exclusively (all p < 0.05). Mean ADC, median ADC, 75th and 90th percentile ADC, SD ADC, and entropy of malignant tumors were significantly higher than those of benign tumors (all p < 0.05). Combination of mean ADC with histogram parameters yielded the highest AUC (0.851; sensitivity, 80.0%; specificity, 86.1%). CONCLUSION Quantitative volumetric ADC histogram analysis may help differentiate various subtypes of small solid renal tumors, including benign and malignant lesions.

Automatic Skeletonization for 3D Hepatic Portal Vein in CT angiography

The analysis of Hepatic Portal Vein (HPV) is important in liver surgical planning and diseases diagnosis, such as living-related liver transplant and oncologic resections. Skeletonization is always the necessary way for vasculature quantification analysis. In this paper, we presented an automatic skeletonization algorithm for 3D HPV based on 3D topological thinning. Since the topological thinning method cannot avoid rings, we analyzed the three types of ring and utilized the graph analysis to remove the loop. To effectively prune artificial twigs resulting from the coarse boundary, the developed approach not only takes into account the length of branches but also the radius of the start and the end points. Also this approach adopts the direction change between the branches to distinguish the other vascular systems. This method has been implemented in physical phantoms and human abdominal Multi-slice Spriral CT Angiography (MSCTA). For the variety of image data about HPV, some other vasculars’ branches cannot be removed, we will extend our work to add user intervention for easily modification.

Comparison of reduced field-of-view diffusion-weighted imaging (DWI) and conventional DWI techniques in the assessment of rectal carcinoma at 3.0T: Image quality and histological T staging

To compare image quality (IQ) of reduced field‐of‐view (rFOV) and full FOV (fFOV) diffusion‐weighted imaging (DWI) sequences at 3T, with histological T staging of rectal cancer as a reference standard.

Comparison of the Diagnostic Value of Monoexponential, Biexponential, and Stretched Exponential Diffusion-weighted MRI in Differentiating Tumor Stage and Histological Grade of Bladder Cancer

RATIONALE AND OBJECTIVES We aimed to determine the utility of various diffusion parameters obtained from monoexponential, biexponential, and stretched exponential diffusion-weighted imaging (DWI) models in differentiating tumor stage and grade of bladder cancer. MATERIALS AND METHODS Forty-five patients with pathologically confirmed bladder cancer underwent multi-b-value DWI. An apparent diffusion coefficient (ADC) was calculated from DWI by using a monoexponential model. A true diffusion coefficient (D), perfusion-related pseudo-diffusion coefficient (D*), and perfusion fraction (f) were calculated from DWI by using a biexponential model. A water molecular diffusion heterogeneity index (α) and distributed diffusion coefficient (DDC) were calculated from DWI by using a stretched exponential model. All parameters were compared between different stages and grades by using the Mann-Whitney U test. Receiver operating characteristic and intrareader correlation coefficient analysis were used for statistical evaluations. RESULTS ADC, D, f, and DDC values were significantly higher in the non-muscle-invasive vs muscle-invasive bladder cancers (P = .000, .000, .002, and .000, respectively) and in low-grade vs high-grade ones (P = .000, .000, .018, and .000, respectively). D* value was significantly lower in the low-grade bladder cancers compared to high-grade ones (P = .012). The areas under the receiver operating characteristic curve of ADC, D, and DDC values were 0.945, 0.912, and 0.946 in staging bladder cancers; 0.866, 0.862, and 0.856 in grading bladder cancers, respectively. CONCLUSION Biexponential and stretched exponential DWI models may provide more parameters in staging and grading bladder cancers and show a slight difference between DDC and ADC values in staging bladder cancers. These two DWI models, as well as the monoexponential models, were very helpful in staging and grading bladder cancers.