Imaging local scatterer concentrations by the Nakagami statistical model

The ultrasonic B-mode image is an important clinical tool used to examine the internal structures of the biological tissue. Due to the fact that the conventional B-scans cannot fully reflect the nature of the tissue, some useful quantitative parameters have been applied to quantify the properties of the tissue. Among various possibilities, the Nakagami parameter was demonstrated to have an outstanding ability to detect the variation of the scatterer concentration. This study is aimed to develop a scatterer concentration image based on the Nakagami parameter map to assist in the B-mode image for tissue characterization. In particular, computer simulations are carried out to generate phantoms of different scatterer concentrations and echogenicity coefficients and their B-mode and Nakagami parametric images are compared to evaluate the performance of the Nakagami image in differentiating the properties of the scatterers. The simulated results show that the B-mode image would be affected by the system settings and user operations, whereas the Nakagami parametric image provides a comparatively consistent image result when different diagnosticians use different dynamic ranges and system gains. This is largely because the Nakagami image formation is only based on the backscattered statistics of the ultrasonic signals in local tissues. Such an imaging principle allows the Nakagami image to quantify the local scatterer concentrations in the tissue and to extract the backscattering information from the regions of the weaker echoes that may be lost in the B-mode image. These findings suggest that the Nakagami image can be combined with the use of the B-mode image simultaneously to visualize the tissue structures and the scatterer properties for a better medical diagnosis.     

计算机辅助多模态融合超声诊断乳腺良恶性肿瘤

目的 观察利用深度学习（DL）融合常规超声和超声弹性成像诊断乳腺良、恶性肿瘤的效能。方法 利用DL卷积神经网络（CNN）提取乳腺肿瘤超声灰阶与超声弹性特征,并进行多模态融合,评价融合弹性图像或弹性比值等不同信息方式对乳腺良、恶性肿瘤的诊断效能;绘制受试者工作特征（ROC）曲线,计算曲线下面积（AUC）,评估多模态融合模型的诊断效能。结果 多模态融合模型鉴别乳腺良、恶性肿物的效能优于单模态常规超声或弹性模型,其中融合灰阶与弹性图像模型鉴别诊断效能优于融合灰阶与弹性比值模型,分类准确率达93.51%,敏感度为94.88%,特异度为92.25%,AUC达0.975。结论 计算机辅助多模态融合有助于提高超声对乳腺良、恶性肿瘤的诊断效能。     

Effect of adaptive threshold filtering on ultrasonic nakagami parameter to detect variation in scatterer concentration

The Nakagami parameter is associated with the Nakagami distribution estimated from ultrasonic backscattered signals and closely reflects the scatterer concentrations in tissues. There is an interest in exploring the possibility of enhancing the ability of the Nakagami parameter to characterize tissues. In this paper, we explore the effect of adaptive thresholdfiltering based on the noise-assisted empirical mode decomposition of the ultrasonic backscattered signals on the Nakagami parameter as a function of scatterer concentration for improving the Nakagami parameter performance. We carried out phantom experiments using 5 MHz focused and nonfocused transducers. Before filtering, the dynamic ranges of the Nakagami parameter, estimated using focused and nonfocused transducers between the scatterer concentrations of 2 and 32 scatterers/mm3, were 0.44 and 0.1, respectively. After filtering, the dynamic ranges of the Nakagami parameter, using the focused and nonfocused transducers, were 0.71 and 0.79, respectively. The experimental results showed that the adaptive threshold filter makes the Nakagami parameter measured by a focused transducer more sensitive to the variation in the scatterer concentration. The proposed method also endows the Nakagami parameter measured by a nonfocused transducer with the ability to differentiate various scatterer concentrations. However, the Nakagami parameters estimated by focused and nonfocused transducers after adaptive threshold filtering have different physical meanings: the former represents the statistics of signals backscattered from unresolvable scatterers while the latter is associated with stronger resolvable scatterers or local inhomogeneity due to scatterer aggregation.     

Use of Nakagami Statistics and Empirical Mode Decomposition for Ultrasound Tissue Characterization by a Nonfocused Transducer

The Nakagami parameter associated with the Nakagami distribution estimated from ultrasonic backscattered signals reflects the scatterer concentration in a tissue. A nonfocused transducer does not allow tissue characterization based on the Nakagami parameter. This paper proposes a new method called the noise-assisted Nakagami parameter based on empirical mode decomposition of noisy backscattered echoes to allow quantification of the scatterer concentration based on data obtained using a nonfocused transducer. To explore the practical feasibility of the proposed method, the current study performed experiments on phantoms and measurements on rat livers in vitro with and without fibrosis induction. The results show that using a nonfocused transducer makes it possible to use the noise-assisted Nakagami parameter to classify phantoms with different scatterer concentrations and different stages of liver fibrosis in rats more accurately than when using techniques based on the echo intensity and the conventional Nakagami parameter. However, the conventional Nakagami parameter and the noise-assisted Nakagami parameter have different meanings: the former represents the statistics of signals backscattered from unresolvable scatterers, whereas the latter is associated with stronger resolvable scatterers or local inhomogeneity caused by scatterer aggregation. (E-mail: mechang@gate.sinica.edu.tw; mcho1215@ntu.edu.tw)     

超声定位光散射成像技术在乳腺肿块定性诊断中的价值

目的探讨运用超声定位光散射乳腺成像(US-guided optical imaging system,OPTIMUS)系统对乳腺肿块的定性诊断价值。方法回顾分析经病理检查证实且同时应用OPTIMUS系统和彩色多普勒超声诊断的36例乳腺肿块,评价两种方法的诊断结果与病理诊断的一致性。并绘制受试者工作特征(receiver operating characteristic,ROC)曲线,计算曲线下面积(area under the curve,AUC),判断OPTIMUS系统及彩色多普勒超声两种检查手段的准确性。结果 36例乳腺肿块术后病理诊断良性16个,恶性20个。彩色多普勒超声诊断乳腺肿块中良性17个,恶性19个;OPTIMUS系统诊断乳腺肿块中良性14个,恶性22个。OPTIMUS系统、彩色多普勒超声诊断与病理诊断结果的一致性检验Kappa值分别为0.772、0.497,其两者之间一致性检验Kappa值为0.606。OPTIMUS系统、彩色多普勒超声诊断的AUC分别为0.923(95%CI:0.834～1.013)、0.802(95%CI:0.654～0.949)。结论 OPTIMUS对乳腺肿块的定性判断较彩色多普勒超声诊断准确性及敏感性高,两者结合可进一步提高诊断准确性。     

Image retrieval with principal component analysis for breast cancer diagnosis on various ultrasonic systems.

OBJECTIVES: We present a computer-aided diagnostic (CAD) system with textural features and image retrieval strategies for classifying benign and malignant breast tumors on various ultrasonic systems. Effective applications of CAD have used different types of texture analysis. Nevertheless, most approaches performed in a specific ultrasonic machine do not indicate whether the technique functions satisfactorily for other ultrasonic systems. This study evaluated a series of pathologically proven breast tumors using various ultrasonic systems. METHODS: Altogether, 600 ultrasound images of solid breast nodules comprising 230 malignant and 370 benign tumors were investigated. All ultrasound images were acquired from four diverse ultrasonic systems. The suspicious tumor area in the ultrasound image was manually chosen as the region-of-interest (ROI) subimage. Textural features extracted from the ROI subimage are supported in classifying the breast tumor as benign or malignant. However, the textural feature always behaves as a high-dimensional vector. In practice, high-dimensional vectors are unsatisfactory at differentiating breast tumors. This study applied the principal component analysis (PCA) to project the original textural features into a lower dimensional principal vector that summarized the original textural information. The image retrieval techniques were employed to differentiate breast tumors, according to the similarities of the principal vectors. The query ROI subimages were identified as malignant or benign tumors according to characteristics of retrieved images from the ultrasound image database. RESULTS: Using the proposed CAD system, historical cases could be directly added into the database without a retraining program. The area under the receiver-operating characteristics curve for the system was 0.970+/-0.006. CONCLUSION: The CAD system identified solid breast nodules with comparatively high accuracy in the different ultrasound systems investigated.     

Computer-aided diagnosis based on speckle patterns in ultrasound images

For breast ultrasound, the scatterer number density from backscattered echo was demonstrated in previous research to be a useful feature for tumor characterization. To take advantage of the scatterer number density in B-mode images, spatial compound imaging was obtained, and the statistical properties of speckle patterns were analyzed in this study for use in distinguishing between benign and malignant lesions. A total of 137 breast masses (95 benign cases and 42 malignant cases) were used in the proposed computer-aided diagnosis (CAD) system. For each mass, the average number of speckle pixels in a region of interest (ROI) was calculated to use the concept of scatterer number density. In addition, the first-order and second-order statistics of the speckle pixels were quantified to obtain the distributions of the pixel values and the spatial relations among the pixels. The performance of the speckle features extracted from each ROI was compared with the performance of the segmentation features extracted from each segmented tumor. As a result, the proposed CAD system using the speckle features achieved an accuracy of 89.1% (122/137); a sensitivity of 81.0% (34/42); and a specificity of 92.6% (88/95). All of the differences between the speckle features and the segmentation features are not statistically significant (p > 0.05). In a receiver operating characteristic (ROC) curve analysis, the Az value, area under ROC curve, of the speckle features was significantly better than the Az value of the segmentation features (0.93 vs. 0.86, p = 0.0359). The performance of this approach supports the notion that the speckle patterns induced by the scatterers in tissues can provide information for classifying tumors. The proposed speckle features, which were extracted readily from drawing an ROI without any preprocessing, also provide a more efficient classification approach than tumor segmentation.     

声辐射力脉冲弹性成像技术结合三维超声检查鉴别诊断乳腺肿块良恶性的价值

目的评价声辐射力脉冲弹性成像（ARFI）技术结合三维超声检查对乳腺影像报告和数据系统（BI-RADS）3级以上乳腺肿块良恶性鉴别诊断的价值。方法选择2012年5-12月上海市第十人民医院经病理证实的BI—RADS3级以上乳腺肿块女性患者66例,共69个肿块。其中乳腺良性肿块24个,乳腺癌45个。首先对69个肿块进行三维超声检查,观察有无汇聚征;然后应用ARFI技术测量肿块内部及同一深度肿块周围正常组织的剪切波速度值（SWV）。以手术病理结果作为金标准,以SWV值绘制受试者操作特性（ROC）曲线,选取ARFI技术鉴别诊断乳腺肿块良恶性的最佳诊断闽值,并分析三维超声检查、ARFI技术、ARFI技术与三维超声检查结合鉴别诊断乳腺肿块良恶性的准确性、敏感度、特异度。结果三维超声检查结果显示,31个肿块出现汇聚征,其中乳腺癌28个,乳腺良性肿块3个;三维超声检查诊断乳腺肿块良恶性的准确性为71．0％,敏感度62．2％,特异度为87．5％。肿块内部SWV值诊断BI-RADS3级以上乳腺肿块良恶性的最佳诊断阈值为4．34m／s,其ROC曲线下面积为0．76,敏感度为64．4％,特异度为87．5％;11个肿块SWV值〉4．34m／s但未出现汇聚征,其中乳腺癌8个,乳腺良性肿块3个;10个肿块出现汇聚征但SWV值≤4．34m／s,其中乳腺癌5个,乳腺良性肿块5个;21个肿块SWV值〉4．34m／s且出现汇聚征,病理证实均为乳腺癌;ARFI技术结合三维超声检查鉴别诊断乳腺肿块良恶性的准确性为75．4％,敏感度为80．0％,特异度为66．7％。结论对于BI—RADS3级以上乳腺肿块,ARFI技术和三维超声检查均有一定的鉴别诊断价值,但2种方法结合并不能明显提高单一超声影像技术诊断准确性。     

超声弹性成像硬度评分及面积比法在术前乳腺良恶性肿块诊断价值

目的：探讨超声弹性成像硬度评分与面积比对乳腺良恶性肿块鉴别的诊断价值。方法2014年2月至2015年6月本科超声检查发现乳腺单发肿块患者68例,术前均采用超声弹性成像检查,获取病灶弹性成像图和灰阶声像图的面积比值资料,并获取术后病理结果,探讨两者在乳腺肿瘤定性方面的价值。结果恶性组43例,良性组25例;恶性组其中83．7％病灶弹性硬度4～5分,明显高于良性组12．8％（ P＜0．05）;恶性组肿块的弹性成像图与灰阶声像图面积比平均为（1．48±0．26）,明显高于非恶性组肿块（1．26±0．21）（ P＜0．05）。硬度评分诊断乳腺恶性病变的灵敏度82．6％特异度84．3％,以面积比为1．3界点诊断乳腺恶性病变的灵敏度67．4％,特,异度为88．6％。结论超声弹性成像硬度评分及面积比法在鉴别乳腺肿块的良恶性具有较高的参考价值。     

加速时间指数及弹性应变率比值鉴别诊断乳腺良恶性肿瘤的价值

目的探讨加速时间指数(ATI)及弹性应变率比值(SR)鉴别诊断乳腺良恶性肿瘤的应用价值。方法回顾性分析120例乳腺肿瘤患者的临床资料,根据病理结果分为良性组76例(98个病灶)和恶性组44例(57个病灶),均行常规超声和超声弹性成像检查,计算并比较两组二维超声图像特征、ATI及SR值。绘制受试者工作特征(ROC)曲线分析ATI、SR值及其联合鉴别诊断乳腺恶性肿瘤的价值,计算曲线下面积(AUC)。结果良、恶性组二维超声图像特征(病灶形态、长轴方向、回声强度、内部回声是否均匀、边界情况、钙化程度、后方回声强度)比较,差异均有统计学意义(均P<0.05);恶性组ATI、SR值均高于良性组(0.18±0.05 vs.0.11±0.03,3.05±0.65 vs.2.33±0.41),差异均有统计学意义(均P<0.05)。ROC曲线分析显示,ATI、SR值鉴别诊断乳腺恶性肿瘤的AUC分别为0.842、0.813,且联合应用的AUC(0.906)高于其单独应用(均P<0.05)。结论ATI、SR值均能准确鉴别乳腺良、恶性肿瘤,两者联合应用价值更高。     

超声弹性成像与常规超声检测肿块面积比在乳腺肿块诊断中的应用

目的探讨超声弹性成像与常规超声面积比对乳腺肿块的诊断价值。方法在常规超声检查的基础上应用超声弹性成像技术对94例乳腺肿块患者进行检查(共计120个肿块),首先观察乳腺肿块的灰阶声像图特征,然后在弹性模式和灰阶超声下测量肿块的面积,计算两者的面积比值,并与病理结果进行对照。以病理结果为金标准分别计算超声弹性成像诊断乳腺恶性肿块的敏感性、特异性及准确性,并绘制ROC曲线计算其曲线下面积。结果 94例患者共120个肿块,其中良性肿块79个,恶性肿块41个。以面积比值≥1.5诊断为恶性肿块,﹤1.5诊断为良性肿块,超声弹性成像诊断乳腺恶性肿块的敏感性、特异性、准确性分别为85.4%、87.3%、86.7%,其诊断乳腺恶性肿块的ROC曲线下面积为0.941。结论超声弹性成像与常规超声面积比对乳腺良恶性肿块的鉴别诊断具有较高的临床应用价值。     

乳腺癌超声造影与微血管密度的相关性研究

目的探讨乳腺癌微血管密度(MVD)与造影时间-强度曲线各参数之间的相关性,以及在鉴别乳腺肿瘤良恶性中的临床应用价值。方法对74例78个乳腺肿瘤行超声造影检查,利用时间-强度曲线进行评估,并结合CD34标记的MVD特征,研究二者相关关系及超声造影对乳腺癌的诊断价值。结果乳腺癌组MVD与峰值强度差值呈正相关(r=0.497,P<0.01),与峰值强度呈正相关(r=0.857,P<0.01),与上升及下降斜率、达峰时间无相关关系;乳腺良性肿瘤组MVD与上述时间-强度曲线定量指标参数间均未见明显相关。乳腺癌组MVD、峰值强度、峰值强度差值、上升斜率、下降斜率均高于良性组(P<0.05)。乳腺癌组与乳腺良性肿瘤组间起始强化时间、达峰时间、峰值持续时间比较差异无统计学意义(P>0.05)。结论超声造影微血管成像技术可以评估乳腺肿瘤MVD情况,可提高对乳腺肿瘤良恶性的鉴别诊断能力。     

超声妇科影像报告和数据系统分类与16层螺旋CT诊断良恶性卵巢肿瘤的价值

目的探讨超声妇科影像报告和数据系统（GI-RADS）分类与16层螺旋CT对良恶性卵巢肿瘤鉴别诊断价值。方法选取2015年1月~2019年8月在我院就诊的卵巢肿瘤患者85例为研究对象,分别进行超声及16层螺旋CT检查,采用GI-RADS系统评价超声声像图表现,并检测其癌胚抗原（CEA）水平。比较超声GI-RADS系统、16层螺旋CT联合CEA联合检查结果与病理学检查结果的一致性;以病理学检查结果为金标准,比较超声GI-RADS系统、16层螺旋CT联合CEA检查诊断鉴别良恶性卵巢肿瘤的灵敏度、特异度、阳性预测值、阴性预测值及诊断准确率,并采用ROC曲线分析超声GI-RADS系统、16层螺旋CT联合CEA检查对良恶性卵巢肿瘤的诊断鉴别价值。结果超声GI-RADS系统联合CEA检查结果与病理学检查结果的一致性（Kappa=0.791）大于16层螺旋CT联合CEA（Kappa=0.487）;超声GI-RADS系统、16层螺旋CT联合CEA联合检查诊断良恶性卵巢肿瘤的灵敏度、特异度、恶性预测值、良性预测值对比,差异无统计学意义（P>0.05）;超声GI-RADS系统联合CEA诊断良恶性卵巢肿瘤的准确率高于16层螺旋CT联合CEA（P < 0.05）;经ROC曲线分析得,超声GI-RADS系统联合CEA诊断良恶性卵巢肿瘤的AUC大于16层螺旋CT联合CEA（P < 0.05）。结论超声GI-RADS系统联合CEA检测对良恶性卵巢肿瘤具有较高的诊断价值,且诊断准确率较高。      

三维超声鉴别诊断乳腺肿块良恶性的优势

目的 探讨三维超声鉴别诊断乳腺肿块良恶性的优势征象.方法 由5名超声医师独立评估109例女性患者中120个乳腺实性肿块的二维及三维超声图像(手术病理诊断69个恶性,51个良性).评估内容包括肿块的7项二维及10项三维声像特征,通过受试者工作特征曲线(ROC) 分析比较三维超声相对于二维超声鉴别诊断肿块良恶性的优势.结果 二维及三维声像图上良恶性肿块的声像特征间各自均显著不同(P<0.001);三维声像上恶性特征显示率高于二维声像,其中三维冠状面上"汇聚征"诊断恶性的敏感性为56.8%、特异性为93.7%,边缘不规则或毛刺征诊断恶性的敏感性为92.2%,特异性为51.8%;完整界面回声诊断良性的敏感性为77.6%,特异性为90.4%.ROC分析显示三维与二维的曲线下面积分别为0.948 3和0.908 6(P<0.05).结论 三维超声提供更多的影像评估特征,使乳腺肿块的良恶性鉴别比二维超声更可靠.三维超声的诊断价值高于二维超声.     

Performance evaluation of ultrasonic Nakagami image in tissue characterization

Conventional ultrasonic B-mode images qualitatively describe tissue structures but are unsuitable for quantitative analyses of scatterer properties. We have recently developed an ultrasonic parametric imaging technique based on the Nakagami statistical distribution that is able to quantify scatterer concentrations. The aim of the present study is to further explore both the behavior of a Nakagami image in characterizing different scatterer structures at different signal-to-noise ratios (SNRs) and the feasibility of Nakagami imaging using a general commercial ultrasound scanner for tissue examinations. Simulations, experiments on a tissue-mimicking phantom and in vitro measurements on a muscle tissue before and after microwave treatment were carried out. The SNR and contrast-to-noise ratio (CNR) were estimated to quantify image performance. The results demonstrate that a Nakagami image can differentiate different scatterer concentrations for single, hypoechoic and hyperechoic targets. Also, a Nakagami image, when combined with an ultrasound scanner, can complement the B-scan to characterize tissue and to identify the region of interest with a larger CNR. However, the noise effect can degrade the performance of a Nakagami image. When the signal SNR decreased to 15 dB in simulations and to 8 dB in experiments, the CNR of the hyperechoic Nakagami image decreased by 4% and 27%, respectively, and that of the hypoechoic one decreased by 42% and 80%, respectively. These results indicate that a Nakagami image behaves well in identifying regions with high scatterer concentrations but does not perform well when both the scatterer concentration and SNR are low.     

彩色多普勒超声显像对乳腺肿块的诊断价值

目的探讨灰阶超声、彩色多普勒血流显像(CDFI)及能量多普勒(CDE)超声在乳腺实质性病灶中的血流信号及超声特点,评价其在乳腺良、恶性肿瘤鉴别诊断中的价值。方法采用二维超声、CDFI、CDE对133例(196个肿块)乳腺实质性肿瘤进行了超声检查,分析病变的形态、边缘、内部回声、包膜、纵/横比、有无钙化灶、血流信号分布等情况,并与术后病理结果对比。结果病理诊断良性肿瘤110例,恶性肿瘤23例。其中恶性肿瘤内部及周边见丰富彩色血流信号显示(21/23),穿通血管和分支血管居多(21/23),且内部可见大量散在钙化或簇样钙化(18/23);而良性肿瘤内彩色血流信号及穿通血管较少,仅2例肿块内部可见钙化。结论二维超声结合CDFI、CDE检查在乳腺良、恶性肿瘤鉴别诊断中有很好的临床应用价值。     

彩色超声判断实性浅表软组织肿物良恶性的ROC曲线分析

目的：应用受试者工作特征曲线（ROC曲线）评价彩色超声判断实性浅表软组织肿物良恶性的价值。方法：应用二维超声、彩色多普勒成像结合新近发展的三维成像、拓宽视野成像技术等观察92例实性浅表软组织肿物的超声表现并与手术及病理对照；对实性肿物的大小、边界、形态、内部均匀度、血流等超声指标进行半定量分级评分并把各分数相加，绘制综合以上各种超声表现判定实性浅表肿物良恶性的ROC曲线，同时计算Youden指数（YI，也称为正确诊断指数）确定最佳诊断点。结果：本组92例浅表软组织肿物超声检出率为100％，其中良性肿物72例，恶性肿物20例。ROC曲线下面积（Az）为0.994（P<0.001）， Youden指数为10。结论：超声对浅表软组织肿物的诊断尤其是对良恶性判断有很高的准确度，从而对治疗方法起到非常重要的指导作用。     

Classification of breast masses in ultrasonic B-mode images using a compounding technique in the Nakagami distribution domain

Classification of masses in ultrasonic B-mode images of the breast tissue using "normalized" parameters of the Nakagami distribution was recently investigated. The technique, however, did not yield performances that were comparable to those of an experienced radiologist, and utilized only a single image for tissue characterization. Because radiologists commonly use two to four images of a mass for characterization, a similar procedure is developed here. A simple summation of the normalized Nakagami parameters from two different images of a mass is utilized for classification as benign or malignant. The performance of the normalized Nakagami parameters before and after the summation has been carried out through a receiver operating characteristic (ROC) study. The bootstrap procedure has been utilized to compute the mean and SD of the ROC area, A(z), obtained for each parameter. It has been observed that combining normalized Nakagami parameters from two images of the mass may help to improve classification performance over that from utilizing the parameters of just a single image. The performance of this automated parameter-based approach appears to match that of a trained radiologist.     

三维超声冠状面成像鉴别乳腺实性病灶良恶性的应用价值

目的：探讨三维超声冠状面成像鉴别乳腺肿块良恶性的应用价值。方法观察分析97例患者106个乳腺实性病灶的二维和三维超声冠状面成像,对二维超声图像进行乳腺超声影像报告与数据系统（BI-RADS-US）分类,并与病理结果对照,计算二维超声对乳腺病灶良恶性病灶的鉴别诊断价值;根据良恶性病灶在三维超声冠状面上的声像图特征,建立Logistic回归模型,绘制受试者操作特性（ROC）曲线及计算曲线下面积来分析其对乳腺癌的诊断价值。结果106个乳腺病灶中,恶性病灶71个,良性病灶35个。二维超声诊断准确性85.8%,敏感度84.5%,特异度88.6%。多因素回归分析显示最后进入Logistic模型的特征分别为病灶边缘的成角或毛刺和“太阳征”。ROC曲线下面积为0.899,标准误为0.033,95%可信区间（0.834,0.965）。以成角或毛刺、“太阳征”为自变量的Logistic回归模型诊断乳腺肿块的准确性为88.7%（94/106）,敏感度为90.1%（64/71）,特异度为85.7%（30/35）,阳性预测值为92.8%（64/69）,阴性预测值为81.1%（30/37）。结论乳腺三维超声冠状面,特别是成角或毛刺征及“太阳征”在乳腺肿块的良恶性鉴别中具有重要价值。对于疑难病灶,三维超声冠状面上的信息有助于提高医生的诊断自信心。     

基于迁移学习的乳腺肿瘤超声图像智能分类诊断

目的 探讨迁移学习方法对乳腺良恶性肿瘤超声图像分类的价值。方法 回顾性分析经病理证实的447例乳腺肿瘤的超声声像图,采用主成分分析法对原始图像进行分析提取;在Matlab 7.0软件中编程实现迁移学习,将量化的图像特征作为输入数据,利用迁移学习对乳腺良恶性肿瘤进行智能分类。结果 乳腺恶性肿瘤的边缘粗糙度、坚固度、邻域灰度差矩阵粗糙度、肿瘤后方与周围区域回声差异及水平方向高频分量和垂直方向低频分量的直方图能量均明显高于良性肿瘤（P均<0.05）。超声和迁移学习方法诊断乳腺恶性肿瘤的敏感度分别为96.21%（127/132）和96.04%（97/101）,特异度为66.35%（209/315）和98.49%（196/199）,准确率为75.17%（336/447）和97.67%（293/300）。结论 超声图像特征定量化可为识别良恶性乳腺肿瘤提供客观的量化参数;迁移学习可有效对乳腺良恶性肿瘤的声像图进行分类。