一种新的正交参数选优法及其在非线性回归分析中的应用

非线性回归分析是工程中经常采用的一种用来估计数学模型参数的方法，该方法能否顺利运用与参数初始值的选择有极大关系。本研究提出一种新的正交参数选优法——阻尼正交表法，它不仅可以保证非线性回归分析算法的顺利收敛，而且能够显著提高后者的收敛速度，进而极大改善非线性回归分析算法的应用性能。本研究的数值试验及心肌造影超声心动图定量分析应用实例表明，作为对传统正交参数选优法的一种改进，阻尼正交表法在科学与工程计算或信号与信息处理领域有着很好的应用前景。     

Enhancement of hepatic hemangiomas with levovist on coded harmonic angiographic ultrasonography.

OBJECTIVE: To evaluate the pattern of contrast enhancement with Levovist on coded harmonic angiographic ultrasonography of hepatic hemangiomas. METHODS: Twenty hemangiomas were evaluated with coded harmonic angiographic ultrasonography and a microbubble contrast agent. Verification of the diagnosis of a hemangioma was made by means of dynamic computed tomography (n = 8), dynamic magnetic resonance imaging (n = 1), radionuclide scanning (n = 6), or follow-up ultrasonography (n = 5). Ultrasonographic images were obtained before contrast agent administration and with a bolus injection of 2.5 g of a microbubble contrast agent (300 mg/mL Levovist; Schering AG, Berlin, Germany) every 10 to 15 seconds for 5 minutes. The contrast enhancement patterns of the 20 hemangiomas were assessed. RESULTS: The tumor diameters as measured on ultrasonography were 7 to 97 mm (mean, 26.7 mm). Of the 20 hemangiomas, peripheral globular enhancement with progressive centripetal fill-in was shown in 15 (75%), rimlike enhancement with progressive centripetal fill-in was shown in 2 (10%), and homogeneous enhancement was shown in 1 (5%). In the remaining 2 lesions (10%), the enhancement patterns could not be seen, because they were not found on coded harmonic angiographic ultrasonography. CONCLUSIONS: Coded harmonic angiographic ultrasonography with a microbubble contrast agent can depict the typical enhancement pattern in most hepatic hemangiomas.     

Multistep phase difference phase contrast imaging

A new technique for multistep phase-contrast image processing is presented. The N-step method consists of simply forming the linear average of the N — 1 adjacent phase-difference signals. It has similar noise reduction properties as other multistep techniques, but the simplicity of the noise variance of the N-step technique allows intuitive insight into phase-difference phase-contrast processing and noise reduction, which can aid in the design of efficient and improved phase-contrast imaging sequences. As well, the computational simplicity of the N-step phase-difference technique compared with any other known multistep technique is advantageous. Like other multistep techniques, it has far more efficient noise reduction properties than simple two-step, multiple average phase-contrast imaging, even when normalized for total scan time. A three-step phase-difference velocity image has 50% less variance than an image acquired with two steps and two scans averaged but is obtained in 25% less scan time. Given its advantages, it should now be the chosen technique for increasing velocity-to-noise and contrast-to-noise ratios in all phase-difference phase-contrast clinical applications.     

Phase-derivative analysis in MR angiography: Reduced Venc dependency and improved vessel wall detection in laminar and disturbed flow

A problem of current MRA techniques is the inability to accurately depict the vascular anatomy, particularly in areas of disturbed flow. Various reasons, such as intravoxel phase dispersion, saturation, temporal variations, and maximum intensity projection (MIP) nonlinearity, cause a wrong delineation of vessel boundaries. A phase contrast (PC)-based postprocessing operation, the phase derivative (PhD), is introduced to detect phase fluctuations indicating flow. Two-dimensional and three-dimensional angiographic reconstruction algorithms are presented. Mathematical formulas are derived to predict the effect of sampling to flow profiles and the effect on the PhD of these profiles. Numerical, phantom, and preliminary in vivo experiments demonstrate that PhD images do not suffer from phase wraps and allow a velocity dynamic range extension only limited by a differential phase change. It is also shown that PhD MIPs produce higher signal-to-noise ratios than conventional PC angiograms and give a better impression of the anatomy of (stenotic) vessels and of their diameters for both laminar and disturbed flow.     

Imaging of shifted stimulated echoes and multiple spin echoes

It is shown that a repetitive pulse sequence consisting of two 90° pulses and gradients in a 1:2 ratio around the second 90° pulse generates interscan shifted stimulated echoes (SSTEs) and intrascan multiple spin echoes (MSEs). Separation of these two types of signals is accomplished using specific gradient crusher schemes. The intensity of the SSTEs is an order of magnitude larger than that of the MSEs and determines the signal contrast if both effects are selected simultaneously. The SSTE sequence generates improved contrast between gray and white matter, even at high field, which is explained in terms of increased inverse T₁-weighting for the interscan echo. The MSE image has low signal to noise and no detectable contrast. The effect of interscan diffusion weighting is also discussed.     

Effect of artifacts due to flowing blood on the reproducibility of phase-contrast measurements of myocardial motion

The reproducibility of myocardial motion trajectories calculated from cine phase-contrast (PC) velocity data is reduced by artifacts due to the inconsistent motion of intracardiac blood. Spatial presaturation reduces these artifacts but requires a longer sequence TR, with a potentially negative effect on trajectory accuracy and reproducibility. We investigated the effect of spatial presaturation on trajectory reproducibility. A midventricular transaxial slice was imaged in five normal volunteers. The same slice was imaged three times each with sequences using spatial presaturation or not. Because the most serious artifacts originate in the heart chambers and propagate in the phase-encoded direction, myocardial regions that were in line with the heart chambers (in the phase-encode direction) had the highest artifact level in the scans without spatial presaturation. The reproducibility of trajectories for regions placed in these areas (the anterior wall, septum and posterior wall in the transaxial scans with phase encoding in the anterior-posterior direction) improved by a factor of two when presaturation was used (P < .001). In areas that were not in line with the heart chambers (eg, the anterior aspect of the lateral wall in the transaxial scans), the effect of presaturation was not significant. These results correlate well with the measured reduction in artifact level. The reproducibility of myocardial motion trajectories over large areas of the heart is improved to approximately 1 mm when presaturation is used. Therefore, use of presaturation is recommended for myocardial motion studies using cine PC velocity data.     

Initial experience with dynamic MR imaging in evaluation of normal bone marrow versus malignant bone marrow infiltrations in humans

The purpose of this study was (a) evaluation of dynamic contrast-enhanced MR imaging of normal bone marrow versus malignant bone marrow infiltrations in patients with proven B-cell-type chronic lymphocytic leukemia (B-CLL) and (b) correlation with the clinical stage according to Binet (stages A, B, C) and response to therapy. Bone marrow imaging of the lumbar spine, pelvis, and proximal femurs was performed at 1.5 T in 45 patients without known malignancy and in 30 patients with B-CLL. The differences between opposed-phase and in-phase dynamic gradient-echo sequences before and up to 10 minutes after intravenous application of .1 mmol/kg body weight of gadolinium-diethylenetriamine penta-acetic acid (Gd-DTPA) were evaluated in normal bone marrow. The contrast-enhancement patterns of normal and malignant bone marrow were compared using the opposed-phase dynamic gradient-echo sequence. Ten of the patients with bone marrow infiltrations (Binet stage C) additionally underwent MR imaging follow-up during therapy. Opposed-phase gradient echo sequences demonstrated a signal decrease of normal bone marrow, and in-phase gradient echo sequences demonstrated a signal increase of normal bone marrow after administration of Gd-DTPA. The dynamic signal intensity time courses differed significantly (P < .05) between Binet stages B and C and controls as well as among the three Binet stages of B-CLL. In the 10 patients followed during therapy, MR imaging sensitively demonstrated response (n = 6), nonresponse (n = 2), or relapse after initial response (n = 2). In out-of-phase imaging, both normal bone marrow and initial bone marrow infiltration in CLL stage Binet A show signal decrease after administration of contrast agent, whereas there is increase in signal intensity in higher-grade bone marrow infiltration in Binet stage B or C disease. The signal loss of normal bone marrow in out-of-phase imaging is a phase effect rather than a T2* effect. The differentiation of initial from higher-grade bone marrow infiltration on out-of-phase images relies solely on a shift in the fat/water ratio.     

Normal morphology of sacroiliac joints in children: magnetic resonance studies related to age and sex

Objective. To determine in a prospective study the normal MRI morphology of the sacroiliac joints (SIJs) in relation to age and sex during adolescence. Design and patients. A total of 98 children (63 boys, mean age 12.7±2.8 years; 35 girls, mean age 13.7±2.3 years), ranging in age from 8 to 17 years, with juvenile chronic arthritis (JCA) but without signs of sacroiliitis fulfilled the study prerequisites (no back pain and no pathologic changes of the SIJs on physical examination before MRI in a 1.5-year follow-up). An additional eight HLA-B27-negative boys and eight HLA-B27-negative girls without arthritis served as controls. The MRI protocol comprised a T1-weighted SE sequence, an opposed-phase T2*-weighted GE sequence, and a dynamic contrast-enhanced study in single-section technique. Results. Noncontrast MRI permitted differentiation of “open” from ossified segmental and lateral apophyses of the sacral wings, with a significant difference in age (P <0.05) between children with open and ossified apophyses. Ossification of the apophyses of the sacral wings was seen significantly earlier (P <0.05) in girls than in boys. Girls also had a significantly higher incidence of transitional lumbosacral vertebrae, pelvic asymmetries, and accessory joints. In the contrast-enhanced opposed-phase MRI study, normal cartilage of the SIJs showed no contrast enhancement whereas the joint capsule showed a moderate enhancement. Conclusion. There are significant age- and sex-related differences in the normal MRI morphology of juvenile SIJs. Our findings might serve as a standard of comparison for the evaluation of pathologic changes – in particular for the early identification of juvenile sacroiliitis.     

Effects of ami-25 on liver vessels and tumors on T1-weighted turbo-field-echo images: Implications for tumor characterization

This study was devoted to tumor differentiation in liver MR T1-weighted imaging with superparamagnetic iron oxide (SPIO). Twenty-one patients with 40 liver lesions were studied at 1.5 T. Before and at least 45 minutes after SPIO administration, turbo-field-echo (TFE) T1-weighted, TFE T1 × T2*-weighted (MXT), and fat-suppressed turbo-spin-echo T2-weighted images were acquired. A quantitative analysis was performed blindly. On TFE T1-weighted images, the signal enhancement was ?33% ± 12 for the liver, ?24% ± 2 for adenomas and focal nodular hyperplasia, +60% ± 33 for the hemangiomas; metastases and cyst enhancement were not significant. After SPIO on TFE T1-weighted images, the hemangioma-to-liver signal ratio (149% ± 18) was definitely higher than the mean metastasis-to-liver signal ratio (90% ± 16). This T1-related differentiation ability lacked dramatically on TFE MXT images and, in one case, was reduced on post-SPIO TFE T1-weighted images by a long imaging delay after SPIO administration (2 hours).