目的 评价基于双参数MRI的深度学习自动分割与机器学习分类模型,探索其在临床显著性前列腺癌(CSPC)诊断中的应用.方法 纳入409例前列腺患者MRI检查资料,在DWI、ADC和T2WI中应用VB-Net模型分别进行病灶自动分割和腺体自动分割,生成感兴趣区(ROI),病灶自动分割时将分割阈值设置为不同数值分别重复进行.分别提取病灶ROI和腺体ROI中的纹理特征,进行Lasso特征选择,建立、训练随机森林、支持向量机和 Logistic回归模型并进行验证.结果 病灶分割中分割阈值分别为0.9、0.5、0.1时,假阴性率分别为0.462、0.273、0.182,假阳性率分别为0.134、0.419、0.661;当分割阈值设为0.5,病灶自动分割后进行纹理分析和机器学习分类,3种模型ROC曲线的AUC为0.76～0.792;腺体分割后进行纹理分析和机器学习分类,3种模型ROC曲线的AUC为0.827～0.855.结论 采用基于前列腺bp-MRI的VB-Net模型对CSPC病灶具有一定的自动分割、分类能力,结合进一步的机器学习能较好地诊断CSPC;VB-Net模型对腺体自动分割后再进行机器学习分类,也对CSPC有较好的诊断价值.在提高全自动病灶自动分割能力方面还需要进一步的深入研究.

目的 比较前列腺影像报告和数据系统(PI-RADS) v2和v2.1检测移行区前列腺癌的诊断效能.方法 筛选67例前列腺特异性抗原(PSA)水平升高的患者行前列腺磁共振检查,随后行超声引导下前列腺穿刺活检或前列腺根治手术,并取得病理结果.由2名放射科医师独立评估每个移行区病变,并进行PI-RADS v2和v2.1评分,比较2名医师评分一致性.采用诊断性试验方法计算PI-RADS v2和v2.1诊断移行区前列腺癌的敏感度、特异度、阳性预测值、阳性预测值、准确率,采用McNemar检验分析上述测值的差异.采用Medcalc软件绘制受试者工作特征曲线,并计算曲线下面积,采用Delong test分析二者曲线下面积差异.结果 67例患者中,前列腺癌30例,前列腺良性病变37例.两名放射科医师应用PI-RADS v2及PI-RADS v2.1评价前列腺病变的Kappa值分别为0.730及0.852.两名放射科医师使用PI-RADS v2.1评价前列腺病变的敏感度均高于PI-RADS v2,差异无统计学意义(P =0.688、P=0.063);PI-RADS v2.1的诊断特异度、阳性预测值、阳性预测值、准确率、曲线下面积均高于PI-RADS v2,且差异有统计学意义.结论 与PI-RADS v2相比,PI-RADS v2.1能更好地评价移行区前列腺病变.

To establish machine learning(ML) models for the diagnosis of clinically significant prostate cancer (csPC) using multiparameter magnetic resonance imaging (mpMRI), texture analysis (TA), dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) quantitative analysis and clinical parameters and to evaluate the stability of these models in internal and temporal validation. The dataset of 194 men was split into training (n = 135) and internal validation (n = 59) cohorts, and a temporal dataset (n = 58) was used for evaluation. The lesions with Gleason score ≥ 7 were defined as csPC. Logistic regression (LR), stepwise regression (SR), classical decision tree (cDT), conditional inference tree (CIT), random forest (RF) and support vector machine (SVM) models were established by combining mpMRI-TA, DCE-MRI and clinical parameters and validated by internal and temporal validation using the receiver operating characteristic (ROC) curve and Delong’s method. Eight variables were determined as important predictors for csPC, with the first three related to texture features derived from the apparent diffusion coefficient (ADC) mapping. RF, LR and SR models yielded larger and more stable area under the ROC curve values (AUCs) than other models. In the temporal validation, the sensitivity was lower than that of the internal validation (p < 0.05). There were no significant differences in specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV) and AUC (p > 0.05). Each machine learning model in this study has good classification ability for csPC. Compared with internal validation, the sensitivity of each machine learning model in temporal validation was reduced, but the specificity, accuracy, PPV, NPV and AUCs remained stable at a good level. The RF, LR and SR models have better classification performance in the imaging-based diagnosis of csPC, and ADC texture-related parameters are of the highest importance.

目的 对比MR双参数与多参数成像纹理及定量分析结合机器学习对高级别前列腺癌的诊断价值.资料与方法 回顾性分析疑似前列腺癌并行前列腺多参数MRI和经直肠超声引导穿刺活检取得病理结果的194例患者,采用Omni-Kinetics软件分别在T2WI、表观扩散系数(ADC)、T1加权动态磁敏感增强(T1WI_DSC)序列勾画病灶所在全部层面兴趣区,提取病变区纹理及定量特征数据后采用多因素Logistic回归分析.应用受试者工作特征(ROC)曲线评价双参数(T2WI+ADC)与多参数(T2WI+ADC+T1WI_DSC)诊断高级别前列腺癌的差异.结果 双参数诊断高级别前列腺癌的敏感度为82.61％,特异度为86.11％,准确度为84.75％;多参数诊断高级别前列腺癌的敏感度为86.97％,特异度为86.11％,准确度为86.44％.两验证组ROC曲线下面积差异无统计学意义(0.918比0.946,P=0.077).结论 MR双参数纹理分析结合机器学习诊断高级别前列腺癌有较高的准确性.

目的 探讨MR多参数成像纹理及定量分析结合机器学习在诊断高级别前列腺癌中的价值.方法 回顾性分析临床怀疑前列腺癌并接受前列腺多参数MRI的患者资料,所有患者均经直肠超声引导穿刺活检取得病理结果.采用Omni-Kinetics软件分析T2 WI、ADC、T1 WI动态增强(T1 WI_DSC)序列,提取病变区纹理特征及定量数据导入R软件进行多因素逻辑回归分析.采用ROC曲线评价各单序列(T2 WI、ADC、T1 WI_DSC)及多序列所得参数(T2 WI+ADC+T1 WI_DSC)在诊断高级别前列腺癌中的价值.结果 前列腺良性增生120例及高级别前列腺癌74例纳入研究.4种模型(T2WI、ADC、T1WI_DSC、T2WI+ADC+T1WI_DSC)预测高级别前列腺癌敏感性分别为86.96％、73.91％、65.22％、86.96％,特异性分别为75.00％、86.11％、66.67％、88.89％,准确率分别为79.66％、81.36％、66.10％、88.14％,AUC值分别为0.8925、0.8575、0.6667、0.9396.结论 MR多参数纹理及定量分析结合机器学习诊断高级别前列腺癌有较高的价值,可临床推广运用于诊断高级别前列腺癌.

目的 建立基于多参数MRI (mpMRI)和影像组学特征的机器学习模型,评价其诊断临床显著性前列腺癌(CSPC)的价值.方法 结合纹理分析、MR动态增强定量分析、前列腺影像报告与数据系统(PI-RADS)评分和部分临床资料建立Logistic回归(LR)、逐步回归(SR)、经典决策树(cDT)、条件推断树(CIT)、随机森林(RF)和支持向量机(SVM)模型,运用ROC曲线和决策曲线分析法(DCA)评价上述模型和变量的重要性.结果 验证组中RF模型诊断CSPC的AUC大于SVM、cDT、SR模型(P均＜0.05),RF模型与LR、CIT模型诊断CSPC的AUC差异无统计学意义(P均＞0.05),其余各模型间诊断CSPC的AUC差异无统计学意义(P均＞0.05).概率阈值为16％～91％时,RF模型的净获益最大,优于其他模型;概率阈值为23％～91％时,SVM模型的净获益仅次于RF模型而优于其他模型.前列腺特异性抗原密度(PSAD)和部分纹理分析参数的重要性较高.结论 RF模型诊断CSPC优于其他模型,SVM模型次之.PSAD和纹理分析相关参数诊断CSPC的重要性高于PI-RADS评分和动态增强MRI定量参数.

Objective To compare the application effects of 3 dimensional sampling perfertion with application optimized contrasts using different flip angle evolutions (3D-SPACE) sequence under the status of high-resolution and low-resolution in Magnetic Resonance Cholangiopancreatography (MRCP) examination. Methods Forty-four examinees were conducted MRCP scanning with Siemens Avanto 1.5T magnetic resonance imaging (MRI) scanner. In such examination, examinees were scanned by adopting 3D-SPACE sequence with high-resolution and low-resolution mode respectively. Then the imaging effects in the two resolution models were evaluated and scored by two senior radiologists separately. Later, the scoring results were analyzed statistically with the Wilcoxon rank sum test. Results Minimal changes were displayed more clearly in original image with high-resolution 3D-SPACE sequence, whilst imaging time of low-resolution 3D-SPACE sequence was less. There was no significant difference in image quality with Maximum Intensity Projection (MIP) reconstruction technology between the two resolution models (score difference in delineating common bile duct: P=0.899, score difference in delineating right and left hepatic duct as well as main pancreatic duct: P=0.623). Conclusion For examinees with poor cooperation, low-resolution 3D-SPACE sequence was easier to get qualified images.

Objective To summarize the optimal sequence parameters, we compared the image quality of diffusion tensor imaging (DTI) with different sequence parameters in lumbar nerve roots of Bama minipig at 1.5 T. Methods We performed lumbar DTI scan on 10 pigs for 3 times based on different combinations of slice thickness, view field and matrix. The signal to noise ratio and reconstructed diffusion tensor tractography (DTT) were measured by 3 experienced radiologists, then the DTT images were evaluated and scored. Scoring results were analyzed statistically with the statistical analysis. Results There was no significant differences in the SNR among different DTI images (P>0.05) when the voxel size were similar, whereas the slice thickness and FOV were different. The difference of DTI image score was statistically significant among the different sequences of FOV (P<0.05). The relative smaller FOV sequence had relative higher DDT image score. Conclusion Voxel size depends on slice thickness, view field and matrix, either too large or too small voxel can cause poor image quality. The DTT images show more nerve fibers and better continuity when use the small view field (72 mm×72 mm) in combination with small matrix. Therefore, this combination is a good choice for DTI evaluation of lumbar nerve root in miniature pigs.

Objective To investigate use of magnetic resonance diffusion tensor imaging (MR-DTI) for quantitatively evaluating the efficacy of Mailuoning Injection on non-compressive lumbar radiculitis.Methods Nine Bama mini pigs were selected and divided into group A,B and C,3 cases in each group.The non-compressive lumbar radiculitis model was established under CT-guiding.The corresponding therapeutic drugs (group A:10 mL Mailuoning;group B:10 mL normal saline;group C:10 mL diminishing inflammation fluid) were given by epidural injection on 14 d after constructing model.MR-DTI was performed before model construction,14 d after model construction and before treatment.One experimental pig in each group was taken on 3,7,14 d after treatment,performed MR-DTI and killed for taking the nerve root sample to conduct the immunohistochemical detection.The fractional anisotropy (FA) values of nerve root in MR-DTI imaging were measured.The FA values and immunohistochemical detection results were statistically analyzed.Results MR-DTI:the FA values after model construction in each group was decreased (P<0.05);the FA values on 14 d after treatment in the group A was increased,which showed statistically significant difference compared with before treatment (P<0.05);the FA values after 3,7 d treatment in the group A had no obvious increase,the difference between the group A and B was not statistically significant (P>0.05),but the FA values increase in the group C was earlier and more rapid than other two groups (P<0.05);the FA values after 14 d treatment in the group A was risen again,the FA values of bilateral nerve roots had no statistical difference between the group A with the group B and C (P>0.05),while the FA values had statistical difference between the group B and C (P<0.05).The immunohistochemical results:TNF-α integral absorbency value(IA value) on 7 d after treatment in the group A began to decline;the TNF-α IA value on 14 d after treatment in the group A and C was significantly decreased compared to group B,the difference was statistically significant (P<0.05).Conclusion Mailuoning Injection has a certain effect on non-compressive lumbar radiculitis,which can be evaluated by using DTI.

Objective:To explore the pathological changes of non-compressive radiculitis after CT-guided treatment,and also e-valuate the changes of DTI parameters in lumbar nerve root. Methods:9 pigs with CT-guided induced non-compressive radiculi-tis were divided into three groups with 3 swine in each group based on implemented treatment methods. Bilateral epidural nerve root injection,at segments L2~3,L3~4,and L4~5,of specific therapeutic materials were injected into the respective groups. Thereafter,each group underwent postoperative (at 3,7 and 14 days) DTI MRI scan,and immunohistochemical study was performed on nerve root specimens. Analysis of each group in the postoperative FA value and nerve root immunohistochemical changes,the correlation between FA value and immunohistochemical changes were also analyzed. Results:There were declining trend that have been observed in nerve root FA value and TNF-α average optical density values after anti-inflammatory liquid treatment. After 14 days,the FA value and TNF-α average optical density value of Mailuoning treatment group showed statistical signifi-cance then compared with saline group. Furthermore,FA value of nerve root and immunohistochemical changes were correlated negatively. Conclusions:Nerve root inflammation could be improved by anti-inflammatory liquid treatment,and DTI could be used as a non-invasive monitoring tool for non-compressive radiculitis after minimal invasive treatment.

目的 探讨变应性支气管肺曲霉病(ABPA)的胸部X线及CT表现.方法 回顾性分析2005年12月-2013年12月收治的20例ABPA的胸部X线、CT表现.结果 胸部X线检查结果显示4例为阴性,14例可见实变影,10例可见肺纹理增粗、紊乱,5例可见支气管扩张.胸部CT结果显示1例为阴性,18例可见支气管扩张,17例可见中心性支气管扩张,其中6例为中心及外周支气管均受累,仅1例表现为单纯外周支气管扩张,11例可见树芽征,6例可见黏液栓,8例可见实变影,5例可见磨玻璃影,4例可见胸膜增厚,3例可见胸腔积液,2例可见肺不张.结论 对于过敏性哮喘及肺囊性纤维化患者,出现中心性支气管扩张可以提示ABPA,但不能作为特异性表现.支气管扩张伴高密度黏液栓,虽然少见,却可能是该病的特异性表现.

Objective: To investigate the multislice CT( MSCT) manifestations of Macrocystic Serous Cystadenoma of the Pancreas( MSAP) . Methods:The MSCT manifestations of 7 cases of MSAP were analyzed retrospectively. All the 7 cases were pathologically proved. Results:Two cases were located in the pancreatic head, 3 cases in the pancreatic body and 2 cases in pancreatic tail. Lobulated contour were found in 4 cases, ovoid contour in 2 cases, irregular contour in 1 case. Multicystic were found in 4 cases, single cystic in 3 cases. Cystic size ranged from 17 mm to 61 mm with an average of 34 mm. The thickness of septa in 4 cases were less than 2mm. Calcification and fiber scar were not found in all cases. Cystic wall were uniformly thin ( thickness less than 2 mm) at all cases. Main pancreatic duct expansions were found in 2 cases. Conclusion: Characteristics manifestations of MSAP include: mainly locating in the head of pancreas 、lobulated shape、thin and homogenous septa、absence of center scar and calcification、thin capsule wall、disconnecting with the main pancreatic duct.

Objective: To investigate the multislice CT(MSCT) manifestations of renal angiomyolipoma(RAML) with minimal fat. Methods:The unenhanced and enhanced MSCT manifestations of 8 cases of RAML with minimal fat were analyzed retrospectively. All the 8 cases were pathologically proved. Results:On unenhanced CT images, the attenuation of all 8 cases was homogeneously, and none of them contained calcification. Compared with that of normal renal parenchy-ma, hyperattenuation was found in 4 cases, isoattenuation in 2 cases, hypoattenuation in 2 cases. Protrusion rate ≥1/2 and angular interface sign was positive in 6 cases. Type 1 beak appearance sign was found in 4 cases, type 2 in 3 cases, type 3 in 1 case. Homogeneously enhancement was found in all 8 cases on contrast enhanced CT images. Among them, ob-vious enhancement at cortical phase was found in 5 cases and at parenchymal phase was found in 3 cases. Enlarged vascular was not found in all 8 cases. Conclusion:MSCT manifestation of renal angiomyolipomas with minimal fat had the following characteristics:attenuation were homogeneous isoattenuation or hyperattenuation on unenhanced CT images, calcification was seldomly found. Homogeneous enhancement after injection of contrast media, protrusion rate≥1/2, angular interface was mainly positive, beak appearance sign was mainly type 1, tortuous or enlarged vessels were found in part of cases.

Objective:To study the value of MRI with multi-sequences in the diagnosis of common bile duct small stone.Methods:MRI features of 57cases with common bile duct small stone proved by clinical data and surgery were analyzed.The sensitivity of using thin-slice T2WI-FS,True FISP,MRCP HASTE and SPACE sequence separately and the above-mentioned 4sequences in combination were compared and analyzed statistically.Results:The sensitivity of thin-slice T2WI-FS,True FISP,MRCP HASTE and SPACE sequence in combination was 89.5%,which was higher than that of using one single sequence(P<0.05).Conclusion:To improve the sensitivity of the diagnosis of common bile duct small stone with MRI,MRCP HASTE,SPACE,True Fisp in combination with thin-slice T2WI-FS sequence are recommended.

Objective To study the association between the time from injury to examination and CT manifestations for the patients with gastric and bowel closed injury perforation.Methods Clinic materials and CT features of 54 patients with bowel and gastric closed injury perfora tion confirmed by surgical operation were collected.The patients were divided into several groups according to the time from injury to CT exam.The association between the time from injury to exam and CT manifestations for the patients was retrospectively analyzed.Results There was significant difference between the display rate of pneumoperitoneum on CT with the patients who were taken examination within 8 hours and those beyond 8 hours(P0.05).Generally,the display rate of ascites on CT was relatively higher(88.9%).Conclusion There was significant the association between the time from injury to exam and CT manifestations of pneumoperitoneum.CT feature of pneumoperitoneum in parts of the patients may be negative within 8 hours after injury.Aimed at the patients with only non-characteristic manifestations such as ascites on early CT exam,intimately observed and repeat CT examination should be done to confirm diagnosis and to avoid the chance of surgery treatment delayed.