CT图像纹理分析在鉴别良恶性肺结节中的价值

目的 评价CT纹理分析在鉴别肺良恶性结节中的价值.方法 回顾性分析110例孤立性肺结节(SPNs)(恶性55例,良性55例)的CT平扫图像,所有结节均经组织病理学或临床随访证实.提取肺结节CT平扫图像的对比度、能量、熵、相关以及一致性等5个灰度共生矩阵纹理参数进行分析.使用t检验比较良恶性结节纹理特征的差异.使用受试者工作特征曲线(ROC)分析确定各参数的最佳临界值,并评价各参数诊断肺良恶性结节的敏感性、特异性、准确性、阳性及阴性预测值.结果 5个纹理特征中,能量、一致性和熵的ROC曲线下面积(AUC)分别为0.997、0.918、0.976.恶性结节的能量、一致性小于良性结节,熵大于良性结节(P＜0.001).使用最佳临界值标准,能量、一致性和熵的敏感性分别为0.982,0.818,1.000,特异性分别为0.964,0.909,0.910,准确性为0.973,0.864,0.945,阳性预测值为0.964,0.9,0.915,阴性预测值为0.981,0.833,0.980.结论 CT纹理特征定量分析是一种有前景的能够准确鉴别良恶性肺结节的手段.     

CT直方图分析鉴别肺良恶性纯磨玻璃结节的价值

【摘要】目的：探讨CT直方图分析对肺良恶性纯磨玻璃结节（pGGO）的鉴别诊断价值。方法：回顾性分析经组织病理学证实的52例pGGO的CT平扫图像,其中恶性28例,良性24例。选取横轴面图像中结节最大层面勾画感兴趣区（避开血管、支气管）生成结节的CT像素直方图。比较良恶性组间直方图参数及直方图分布特点的差异,采用Logistic回归法分析pGGO恶变的最佳预测指标及阈值。结果：良、恶性组的直方图参数中跨度分别为（590.3±141.5）和（907.2±298.0）HU,最大CT值分别为（-333.5±140.4）和（-105.1±244.4）HU,峰度对应CT值分别为（-675.1±145.3）和（-468.4±187.0）HU,组间差异均有统计学意义（P<0.05）;良性pGGO的直方图主要表现为“速升缓降”型,恶性pGGO主要表现为“缓升速降”型,差异具有统计学意义（P<0.05）。鉴别良恶性pGGO的峰度值临界值为-553HU,相应的诊断敏感度和特异度分别为64.3%和87.5%。结论：基于像素的CT直方图分析技术可为肺良恶性pGGO的鉴别诊断提供重要参考依据。     

多层螺旋CT对孤立性肺结节的诊断价值

目的 探讨多层螺旋CT鉴别孤立性肺良恶性结节的诊断价值.方法 回顾性分析2005-10-2007-12在本院接受治疗的临床与病理资料完整的肺孤立性结节(SPN)患者76例,以手术或穿刺活组织检查的病理结果作为诊断金标准.以动态扫描中增强值＞20 HU为界值计算出CT诊断SPN的敏感性、特异性、准确性、阳性预测值和阴性预测值.结果 多层螺旋CT对良性孤立性肺结节诊断的敏感度、特异度、准确性、阳性预测值、阴性预测值分别为92.6%,89.8%,90.8%,83.3%,95.7%.对恶性孤立性肺结节诊断的敏感度、特异度、准确性、阳性预测值、阴性预测值分别为83.7%,80.5%,82.9%,89.1%,73.3%.结论 多层螺旋CT诊断良恶性孤立性肺结节具有较高的敏感性、特异性和准确性,可作为诊断孤立性肺结节首选的非侵入性检查方法.     

声脉冲辐射力成像对乳腺良恶性结节的鉴别诊断价值

目的 :探讨声脉冲辐射力成像(acoustic radiation force impulse,ARFI)在诊断乳腺良恶性结节中的价值。方法:分析128例共145个经病理证实的乳腺结节的声脉冲辐射力成像特征。采用ROC曲线评价剪切波速度值对乳腺病灶良恶性的鉴别诊断价值,并确定临界值。结果:145个结节中,良性结节107个,恶性结节38个。乳腺病灶良性组与恶性组的剪切波速度(shearwave velocity,SWV)值分别为(2.89±1.13)m/s和(6.57±1.36)m/s,2组差异有统计学意义(P0.05)。以SWV值3.37 m/s为临界值诊断乳腺病灶的良恶性的敏感度、特异度、准确度分别为89.4%、89.7%、90.3%。结论:ARFI为乳腺结节的良恶性鉴别诊断提供了新的分析方法,具有重要的临床价值。     

首过时间动态增强CT评价孤立性肺结节良恶性——与延迟增强模式对照

目的 探讨16层CT首过时间动态增强在确定孤立性肺结节(solitary pulmonary nodules, SPNs)良恶性的价值并与延迟动态增强CT对照. 资料与方法 61例SPNs患者同时行首过及延迟时间动态增强CT扫描.记录SPNs首过时间强化峰值(peak height,PH),受试者操作曲线(receiver operating characteristic,ROC)用于确定首过及延迟时间PH值诊断结节良恶性的最佳界值点.计算并比较两种增强方式最佳界值点对病变良恶性诊断的敏感度、特异度、准确率,阳性、阴性预测值. 结果 良性结节首过及延迟时间PH值[(10.912±5.452) HU,(18.172±11.567) HU]均小于恶性结节首过及延迟时间PH值[(22.864±13.514) HU,(26.350±12.744 )HU].首过及延迟时间PH值诊断结节良恶性的最佳界值点为分别为15 HU、17 HU,对SPNs良恶性的诊断,首过时间PH值诊断的敏感度、特异度、准确率,阳性、阴性预测值和曲线下面积(Az值)分别为77.78%、84%、 80.33%、87.5%、72.41%和0.84;延迟时间PH值分别88.89%、48.00%、72.13%、71.11%、75.00%和0.76.两种检查方法比较差异有统计学意义(P＝0.042). 结论 胸部首过时间动态增强CT较延迟动态增强模式能更准确评价SPNs良恶性.     

CT纹理分析技术鉴别甲状腺良恶性结节可行性研究

目的探讨CT纹理分析技术在鉴别甲状腺良恶性结节中的价值。方法回顾性分析经我院手术病理证实的甲状腺病变病人35例,共42个病灶,其中恶性结节26个,良性结节16个。所有病人治疗前均行颈部增强CT扫描。将DICOM格式的CT增强图像(层厚和层间距均为5 mm)导入CT Kinetics软件进行纹理及直方图分析得到未经滤过的原始细纹理图像。CT纹理分析主要参数包括熵值、偏度、峰态、平均像素值和像素分布的标准差。甲状腺良恶性结节间纹理参数比较采用独立样本t检验或Mann-Whitney U检验,并对有统计学意义的纹理参数进行受试者操作特征(ROC)曲线分析,确定诊断阈值。结果甲状腺恶性结节的熵值、偏度、峰态、像素值和标准差分别为6.65±0.92、0.63±1.37、0.69±1.23、84.08±23.36和18.14±3.31;良性结节分别为5.96±0.54、0.59±1.42、0.51±1.17、72.00±24.52和20.05±6.10。熵值在甲状腺良恶性结节间差异有统计学意义(P0.05),偏度、峰态、像素值和标准差在甲状腺良恶性结节间差异均无统计学意义(均P0.05)。ROC曲线分析显示,以熵值6.09为鉴别甲状腺结节良恶性的阈值,其ROC曲线下面积、敏感度和特异度分别为0.733、71.3%和70.0%。结论 CT纹理参数对鉴别甲状腺结节的良恶性有一定帮助。     

实时剪切波弹性成像鉴别甲状腺结节良恶性的定量分析

目的应用剪切波弹性成像测定甲状腺结节的杨氏模量值,探讨其在结节良恶性鉴别诊断中的价值.资料与方法对79例甲状腺结节患者进行实时剪切波弹性成像,测得结节的杨氏模量值,以病理结果为“金标准”,绘制 ROC 曲线,分析鉴别良恶性病灶的参考界值.结果79例共分析甲状腺结节88个,良性63个,恶性25个,良性结节的平均、最大杨氏模量值分别为(27.56±25.22)kPa 和(41.79±30.66)kPa,恶性结节的平均、最大杨氏模量值分别为(88.38±35.27)kPa 和(137.36±56.81)kPa ;良恶性结节杨氏模量最大值和平均值差异有统计学意义(P <0.05);甲状腺结节杨氏模量最大值与平均值诊断恶性病变的 ROC 曲线下面积为0.971和0.802(P <0.05),分别以结节杨氏模量最大值90.34kPa 和平均值60.25kPa 作为诊断界值,其诊断敏感度、特异度为90.34%、86.89%和63.67%、91.66%;2名观察者间符合率为93.2%(Kappa =0.83).结论剪切波弹性成像测得的甲状腺结节杨氏模量最大值与平均值可以定量评价病灶硬度,有助于甲状腺结节良恶性的鉴别诊断.     

能谱CT钙基图像对甲状腺良恶性结节的鉴别诊断价值

目的 探讨能谱CT钙基图像定量分析对甲状腺良恶性结节的鉴别诊断价值.方法 收集经手术病理证实的82例甲状腺结节患者(共98个结节),所有患者术前均行甲状腺能谱CT平扫及双期增强扫描.根据病理结果,将甲状腺结节分为恶性组(61个)和良性组(37个),另选50例患者的对侧正常甲状腺组织作为正常组,比较3组平扫期钙浓度的差异.绘制受试者工作特征(ROC)曲线,得出最佳诊断阈值及诊断的敏感度和特异度.结果 恶性组、良性组和正常组的平扫期钙浓度(mg/cm3)分别为5.52±2.72、10.72±4.68和24.66±7.58,两两比较差异均有统计学意义(P值均<0.001).平扫期诊断甲状腺恶性结节的最佳钙浓度值为6.065 mg/cm3,敏感度为65.6%,特异度为91.9%.结论 能谱CT能够定量评估甲状腺结节的钙浓度,对鉴别甲状腺良恶性结节有一定价值.     

Q-analysis实时弹性成像技术对乳腺良恶性病变鉴别诊断的应用价值

目的探讨Q-analysis实时弹性成像技术在诊断乳腺良恶性病变中的应用价值。方法选取2016年8月~2017年8月在连云港市第一人民医院行乳腺超声检查的乳腺结节患者79例,共94个结节,所有结节均获得明确病理结果。应用Q-analysis系统实时组织弹性定量分析,获得乳腺肿块的整体和局部弹性比值,并绘制受试者操作特性(ROC)曲线,分析结节整体及局部弹性比值的曲线下面积,确定诊断的最佳临界值及两种方法评估乳腺肿块良恶性的敏感度及特异度,并应用Z检验分析ROC曲线,比较两种方法在乳腺疾病诊断中的差异是否具有统计学意义。结果 94个病灶中良性肿块72个,恶性肿块22个。弹性图像显示良性结节的Q-analysis曲线较平稳,峰值较低;恶性结节曲线波动较大,峰值较高;乳腺恶性肿块与良性肿块的整体弹性比值分别为3.49±1.02、2.40±0.80,二者比较具有统计学意义(P0.05),恶性肿块与良性肿块的局部弹性比值分别为3.33±0.93、2.13±0.85,两者比较差异有统计学意义(P0.05);实时弹性分析乳腺病灶整体及局部弹性比率值法的ROC曲线下面积分别为0.795和0.893,两种方法分析比较,对良恶性的乳腺病变的鉴别诊断具有较高的准确性;乳腺病灶整体及局部临界值分别为3.2和2.6,其诊断的敏感度和特异度分别为86.36%、83.33%和95.45%、90.28%;两种方法对乳腺良恶性诊断效能的比较差异有统计学意义(Z=1.96,P=0.027)。结论 Q-analysis实时弹性分析技术可评估病灶的动态的实时的弹性图像征象,为乳腺病变良恶性的鉴别诊断提供依据,而且局部弹性比率值的敏感性及特异度优于整体弹性比率值。     

彩色多普勒超声评分法在甲状腺结节良恶性鉴别诊断中的应用价值

目的 :探讨彩色多普勒超声评分法在甲状腺良恶性结节鉴别诊断中的应用价值。方法 :选择与甲状腺癌鉴别相关性较高的8项评价指标,设定评分标准,拟定甲状腺良恶性结节鉴别的超声评分临界值,对72例甲状腺结节进行超声评分。结果:本研究通过超声评分法结果与病理对照,进行相关性分析,界定8分为良恶性结节超声评分临界值,其诊断甲状腺癌的敏感度为92.8%,特异度为90.9%,准确率为91.6%,阳性预测值为86.6%,阴性预测值为95.2%。良恶性结节评分差异有统计学意义(P0.05)。结论:甲状腺超声评分法提高了良恶性甲状腺结节定性评估的敏感度、特异度和准确率,对甲状腺结节良恶性的鉴别具有较高的应用价值。     

Stress MR imaging for evaluation of popliteal artery entrapment

The popliteal artery entrapment (PAE) syndrome has been recognized as a cause of arterial occlusion in young people. It is the result of an anomaly of the relationship between the popliteal artery and the gastrocnemius muscle. Eight young healthy volunteers (16 legs) and six patients (10 legs) with suspected PAE underwent magnetic resonance (MR) imaging. Gradient-echo images were obtained in axial planes with the leg at rest and during active plantar flexion against resistance. Imaging at rest allowed identification of PAE signs in only one leg, which had an anomalous medial course of the popliteal artery. In the other cases, only the stress technique was able to show signal loss in the popliteal artery due to muscular compression (two legs) or the presence of accessory muscle slip around the vessel (two legs), as confirmed at surgery. MR imaging is therefore a useful technique for the diagnosis of PAE because of its capability of combining information obtainable with other modalities.     

MR imaging of muscle and tender points in fibromyalgia

Fibromyalgia is a syndrome manifested by chronic, diffuse muscu-loskeletal aching and soreness, palpable muscle tender points, and other symptoms. Standardized clinical diagnostic criteria have recently been developed. Skeletal muscle has been postulated as the end organ in this disease. Biochemical, histologic, electromyographic, and conventional radiographic studies have demonstrated no definitive abnormality. This study sought to establish whether magnetic resonance (MR) imaging could demonstrate any abnormality in these patients. Eighteen patients were entered in the study, 14 of whom were able to complete their examinations. T1 -weighted, T2-weighted, gradient-echo, and STIR (short-tau inversion-recovery) sequences were performed in all patients, with selected patients examined with T1weighted, gadopentetate dimeglu-mine-enhanced sequences. The trapezius and suboccipital regions were imaged in patients who, clinically, had active fibro-myalgia. No abnormalities could be detected. The authors conclude that the conventional MR imaging used in this study was unable to depict any primary skeletal muscle abnormality in fibromyalgia.     

MR angiography of congenital heart disease: Value of segmented two-dimensional inflow technique and maximum-intensity-projection display

Magnetic resonance (MR) angiography of the cardiovascular system was evaluated in 41 patients with congenital heart disease by using a two-dimensional (2D) inflow technique based on a magnetization-prepared gradient-echo pulse sequence with segmented k-space data acquisition and electrocardiographic gating at 0.5 T. Inversion and saturation prepulses were used to suppress stationary tissue and enhance intravascular signal. Presaturation slabs were applied where certain vascular structures had to be suppressed. Sequence parameters were optimized by evaluating signal intensity and contrast characteristics for various flip angles and inversion and saturation delay times. The heart and intrathoracic vasculature were encompassed with 40–50 overlapping sections. Both 2D angiograms and maximum-intensity-projection images were evaluated. Combining data sets acquired in the sagittal and transverse orientations provided the most satisfactory information about the pulmonary arteries. The highest signal-to-noise ratios were obtained with a flip angle of 65° and short prepulse delay times. Two-dimensional MR angiography can provide useful diagnostic information but requires a thorough understanding of in-plane and hemodynamically induced signal intensity changes.     

MR imaging of disorders of the brachial plexus

The authors evaluated 64 consecutive patients with suspected brachial plexus (BP) abnormalities of diverse cause with magnetic resonance (MR) imaging, using the body coil and a standardized protocol. Of the 43 patients for whom follow-up was available, 25 were suspected of having neoplastic involvement of the BP, nine had sustained injuries, and nine presented with BP symptoms of uncertain cause. MR imaging was 63% sensitive, 100% specific, and 77% accurate in demonstrating the abnormality in this diverse patient population. When patients with neoplastic and traumatic disorders were considered separately, sensitivity increased to 81%, accuracy to 88%, and specificity remained unchanged. In the patients with a clinical diagnosis of idiopathic or viral plexitis, the MR imaging findings were normal, serving to exclude other structural abnormalities. It is concluded that MR imaging is valuable in the assessment of a wide range of BP disorders.     

MR imaging characteristics of noncancerous lesions of the prostate.

Radical prostatectomy specimens from 53 men with clinical stage A or B prostate cancer were retrospectively reviewed and compared with correlative axial T2-weighted magnetic resonance (MR) images obtained just before surgery. Non-cancerous lesions were evaluated for signal intensity and location. Focal high-signal-intensity areas (n = 72) were present in 81% of patients. The 26% of lesions seen in the central gland all correlated with cystic atrophy. Of the 53 lesions seen in the peripheral prostate, 47 (89%) were cystic atrophy without associated cancer, four (7.5%) cystic atrophy with cancer, and two (3.8%) focal inflammation. Focal low-signal-intensity areas (n = 42) were present in 60% of patients. Of the 31% of lesions in the central prostate, one-fifth correlated with benign prostatic hyperplasia (BPH) and four-fifths with fibrous tissue. Of the 69% of peripheral lesions, 83% corresponded to fibrous tissue, 10% to BPH, and 7% to normal tissue. Mixed lesions (n = 42) were present in 64% of patients; 86% of these were located centrally and 14% peripherally. All mixed central lesions were BPH; the peripheral lesions were areas of combined cystic atrophy and fibrosis. BPH of low or mixed signal intensity can extend into the peripheral prostate and mimic cancer. High-intensity cystic atrophy associated with cancer can mimic normal tissue.     

肾细胞癌的MRI研究进展

肾细胞癌是最常见的成人肾脏恶性肿瘤。近年来，多种功能MRI成像技术（如扩散加权成像、灌注加权成像等）、多参数MRI联合分析以及影像组学等新兴影像处理技术被证实在肾细胞癌的诊断中具有较大的价值。目前，研究热点多集中于良恶性肿瘤的鉴别、组织学亚型的区分、肿瘤分期、预测核分级及判断预后。就MRI新技术及图像处理技术在肾细胞癌中的研究进展予以综述。     

Dynamic contrast-enhanced MR imaging assessment of vascularized free fibular grafts

Magnetic resonance (MR) imaging may be a noninvasive method for assessing perfusion of vascularized bone grafts placed for treatment of avascular necrosis. One proximal femur of seven beagles was devascularized, with insertion of a vascularized fibular graft. MR imaging at 1 week (seven dogs) and 6 weeks (five dogs) after surgery included pre- and postcontrast spin-echo sequences, unenhanced twodimensional time-of-flight (TOF) vascular imaging, and dynamic gradient-echo imaging during infusion of gadolinium. Relative signal intensity values of selected regions obtained from the dynamic gradientecho images were plotted as percent enhancement versus time. In the operated hip, MR imaging did not show enhancement in six of seven femoral heads and greater trochanters at 1 week after surgery, with similar results after 6 weeks. MR imaging of fibular grafts 6 weeks after surgery showed an initial rapid increase in enhancement and a subsequent slower increase in five of five dogs, although no enhancement was seen in six of seven dogs at 1 week. These findings contrasted with a rapid initial increase in enhancement followed by slow decline in non-operated hips. Two-dimensional TOP imaging did not show the vascular pedicle of the graft in any dog. Findings of radionuclide bone scanning performed 1 week after surgery were consistent with devascularization of the operated femur and fibular graft. However, tetracycline distribution and histologic findings confirmed the viability of five of five grafts within the devascularized femurs 6 weeks after surgery. Thus, dynamic contrast-enhanced MR imaging at 6 weeks after surgery is valuable for assessing vascular bone graft perfusion, while similar imaging at 1 week may suggest otherwise.     

Influence of in vivo copper on MR images of the liver in rats

To investigate the effects of in vivo copper on magnetic resonance (MR) images, the authors studied Long-Evans cinnamon rats, which develop hepatitis and hepatocellular carcinoma as a result of abnormal copper metabolism. The livers of the rats were imaged before hepatitis developed; the absence of hepatic disease was confirmed histopathologically. The copper that accumulated in the liver of the rats was thought to exist in the form of divalent ions, which were suspected of reducing the T1 and T2 of neighboring protons. However, the signal intensities of the liver on T1- and T2*-weighted images did not change, suggesting that in vivo copper, even when accumulated abnormally, does not influence the signal intensity of MR images.     

Techniques for high-resolution MR imaging of atherosclerotic plaque

Atherosclerotic cardiovascular disease is the most common cause of death in the United States. Investigation of atherosclerotic plaque morphology and composition is important because the findings may be useful in predicting prognosis or response to therapy. This study presents high-resolution magnetic resonance (MR) imaging techniques developed on a 1.5-T whole-body imager with a custom-built surface coil, for characterizing the composition and morphology of plaque removed at carotid endarterectomy. The initial comparison of MR imaging and histologic results showed good correlation. In conjunction with MR angiography, these techniques could be used in in vivo imaging to define the size, location, and contents of atherosclerotic plaque at the carotid bifurcation.     

Dynamic MRI of the gallbladder lesions: Differentiation of benign from malignant

Forty-nine pathologically proven gallbladder lesions were evaluated in 45 patients using dynamic MRI with a spoiled gradient pulse sequence (SPGR), to access the ability of this technique to differentiate benign from malignant gallbladder lesions. The studies were reviewed retrospectively. Signal intensity of the lesions were measured. Twenty-one malignant and 28 benign lesions were classified into three categories: polypoid, diffuse wall thickening, and exophytic. Early and delayed enhancement patterns were evaluated. For the polypoid masses, malignant lesions (n = 9) demonstrated early and prolonged enhancements, whereas benign lesions (n = 14) had early enhancement with subsequent washout (P < .05). For diffuse gallbladder wall thickening, malignant lesions (n = 6) demonstrated early and prolonged enhancement and benign lesions (n = 14) showed relatively slow, prolonged enhancement (P < .05). The exophytic masses (n = 6) all were malignant and demonstrated early and prolonged enhancement. Dynamic MRI can help differentiate benign from malignant gallbladder lesions.