The inherent variability of lesions poses challenges in leveraging AI in 3D automated breast ultrasound (ABUS) for lesion detection. Traditional methods based on single scans have fallen short compared to comprehensive evaluations by experienced sonologists using multiple scans. To address this, our study introduces an innovative approach combining the multi-view co-attention mechanism (MCAM) with unsupervised contrastive learning. Rooted in the detection transformer (DETR) architecture, our model employs a one-to-many matching strategy, significantly boosting training efficiency and lesion recall metrics. The model integrates MCAM within the decoder, facilitating the interpretation of lesion data across diverse views. Simultaneously, unsupervised multi-view contrastive learning (UMCL) aligns features consistently across scans, improving detection performance. When tested on two multi-center datasets comprising 1509 patients, our approach outperforms existing state-of-the-art 3D detection models. Notably, our model achieves a 90.3% cancer detection rate with a false positive per image (FPPI) rate of 0.5 on the external validation dataset. This surpasses junior sonologists and matches the performance of seasoned experts.

Objective To assess the differential diagnostic performance of spectral characteristic parameters of radio-frequency(RF)signal time series based on ultrasonic RF flow for benign and malignant breast lesions.Methods Two dimensional B-mode ultrasound images and RF data of 137 breast lesions were collected.All ultrasonic RF data were quantitatively analyzed with a software developed by our laboratory for ultrasonic RF time series analysis.Finally,nine spectral characteristic parameters were obtained,including SMR fractal dimension,Higuchi fractal dimension,slope,intercept,mid-band fit,S1,S2,S3,and S4.All of the 116 breast lesions were pathologically diagnosed.86 lesions were confirmed to be malignant,30 lesions were benign and 21 lesions were diagnosed as benign after follow-up.The sensitivity,specificity,accuracy,positive and negative predictive values of individual parameter of RF time series spectral characteristic parameters and combined parameters of regression models were calculated,as well as Logistic regression model was established.The receiver operating characteristic(ROC)curve and the area under ROC curve(AUC)were obtained to evaluate the differential diagnostic values of these parameters for benign and malignant breast lesions.Results Multivariate regression analysis showed that the parameters finally included into the Logistic model were Higuchi fractal dimension,S2,and S4.The highest sensitivity,specificity,accuracy,positive and negative predictive values of RF time series spectral characteristic parameters in the diagnosis of breast lesions were 90.7％(S2)and 92.2％(Higuchi fractal dimension,S4),86.1％(regression model),93.9％(S4)and 79.6％(regression model),respectively.The AUCs could reach up to 0.910(S4)and 0.930(regression model),and there was no statistical significance between them(P>0.05).Conclusions The characteristic parameters of RF signal time series based on ultrasonic RF flow provide quantitative data on the sub-resolution tissue microstructure in terms of physical properties,which yields high differential diagnostic efficiency for benign and malignant breast lesions.

ObjectiveThis study aimed to evaluate a convolution neural network algorithm for breast lesion detection with multi-center ABUS image data developed based on ABUS image and Yolo v5.MethodsA total of 741 cases with 2,538 volume data of ABUS examinations were analyzed, which were recruited from 7 hospitals between October 2016 and December 2020. A total of 452 volume data of 413 cases were used as internal validation data, and 2,086 volume data from 328 cases were used as external validation data. There were 1,178 breast lesions in 413 patients (161 malignant and 1,017 benign) and 1,936 lesions in 328 patients (57 malignant and 1,879 benign). The efficiency and accuracy of the algorithm were analyzed in detecting lesions with different allowable false positive values and lesion sizes, and the differences were compared and analyzed, which included the various indicators in internal validation and external validation data.ResultsThe study found that the algorithm had high sensitivity for all categories of lesions, even when using internal or external validation data. The overall detection rate of the algorithm was as high as 78.1 and 71.2% in the internal and external validation sets, respectively. The algorithm could detect more lesions with increasing nodule size (87.4% in ≥10 mm lesions but less than 50% in <10 mm). The detection rate of BI-RADS 4/5 lesions was higher than that of BI-RADS 3 or 2 (96.5% vs 79.7% vs 74.7% internal, 95.8% vs 74.7% vs 88.4% external). Furthermore, the detection performance was better for malignant nodules than benign (98.1% vs 74.9% internal, 98.2% vs 70.4% external).ConclusionsThis algorithm showed good detection efficiency in the internal and external validation sets, especially for category 4/5 lesions and malignant lesions. However, there are still some deficiencies in detecting category 2 and 3 lesions and lesions smaller than 10 mm.

目的 探讨自动乳腺容积超声(ABUS)联合远程诊断的临床价值.方法 经过专业培训的来自基层医院或体检医疗机构的技术人员采集ABUS图像后.将患者临床资料和ABUS图像通过网络远程传输到我院读图中心工作站,由具有丰富ABUS诊断经验的超声医师进行图像质控及阅图,并给予BI-RADS分类诊断及建议,对于BI-RADS分类4A类以上的病例需经过自动乳腺容积超声专家会诊,并协助制定诊疗方案.结果 ABUS远程诊断共检测3 721例患者,其中远程专家会诊424例.BI-RADS分类:BI-RADS0类2例,BI-RADS 1类225 例,BI-RADS 2 类 823 例,BI-RADS 3 类 2 247 例,BI-RADS 4A 类 320 例,BI-RADS 4B 类 52 例,BI-RADS 4C 类 40 例,BI-RADS 5类10例,BI-RADS6类2例.所有患者均给予远程诊断BI-RADS分类恶性危险分层及处理建议,其中90.4％(3 362例)的诊断报告给予了倾向性的病理诊断结果,优化了患者的诊疗方案.结论 通过自动乳腺容积超声联合远程诊断,可以使基层单位或偏远地区人群就地共享优质专家的读图资源,解决当地人群乳腺相关疾病在筛查、分级诊疗及疾病管理的相关问题,有助于促进乳腺癌的早诊早治.

Background:The aim of this study was to develop a conventional ultrasound (US) features-based nomogram for the prediction of malignant nonmasslike (NML) breast lesions.Methods:Consecutive cases of adult females diagnosed with NML breast lesions via US screening in our center from June 1st, 2017, to April 17th, 2020, were retrospectively enrolled. Candidate variables included age, clinical symptoms, and the image features obtained from the conventional US. Nomograms were developed based on the results of the multiple logistic regression analysis via R language. One thousand bootstraps were used for internal validation. The area under the curve (AUC) and the bias-corrected concordance index (C-index) were calculated. Decision curve analysis (DCA) was also performed for further comparison between the nomogram and the Breast Imaging Reporting and Data System (BI-RADS). The study has not yet been registered.Results:A total of 229 patients were included in the study after exclusion and follow-up. The overall malignant rate of NML breast lesions was 31.0%. Age, clinical symptoms, echo pattern, calcification, orientation, and Adler's classification were selected to generate the nomogram according to the results of the multivariable logistic regression analysis. The bias-corrected C-index and the AUC of our nomogram were 0.790 and 0.828, respectively. The DCA showed that our model had larger net benefits in a range from 0.2 to 0.7 when compared with the BI-RADS.Conclusions:We developed a prediction model using a combination of age, clinical symptoms, echo pattern, calcification, orientation, and Adler's classification for malignant NML breast lesion prediction that yielded adequate discrimination and calibration.

Objective:To analyze the differences of clinical symptoms and ultrasonic image features between benign and malignant non-mass-like breast lesions and to investigate their diagnostic values.Methods:A total of 229 non-mass-like breast lesions detected by ultrasound in 210 female patients aged (44.3±11.1) years were retrospectively analyzed in The Second Affiliated Hospital, Guangzhou University of Chinese Medicine from June 2017 to July 2019. The differences of clinical symptoms and ultrasonic image features were compared between the benign and malignant groups. The ultrasonic image features included size, orientation, cord-shaped hypoecho in the lesion, echo pattern, posterior features, calcification, associated features, and blood flow signal. Independent sample t test was used to compare the measurement data, and χ2 test or Fisher's exact probability method was used to compare the count data. Results:A total of 71 malignant lesions and 158 benign lesions were confirmed by surgery or biopsy. There were statistically significant differences in clinical symptoms, cord-shaped hypoecho in the lesion, echo pattern of lesions, calcification, associated features, and blood flow between the benign group and the malignant group (all P<0.05); in the benign group, the echo pattern of lesions was mainly type Ⅱa and Ⅴ, while in the malignant group, it was mainly type Ⅰb and Ⅱb; there were no statistically significant differences in the size, orientation, and posterior features (all P>0.05). Conclusion:There are some differences in the clinical symptoms and ultrasonic image features between benign and malignant non-mass-like breast lesions, which may be helpful to the differential diagnosis for non-mass-like breast lesions.

Objective:To investigate the effect and influencing factors of microwave ablation (MVA) in the treatment of benign thyroid nodules.Methods:The clinical data of ultrasound-guided microwave ablation for thyroid benign nodules in the Second Affiliated Hospital of Guangzhou University of Traditional Chinese Medicine from April 2017 to April 2019 were retrospectively analyzed. At 1, 3 and 6 months after operation, conventional ultrasound examination was performed to calculate the volume reduction rate of the nodules. The nodules were divided into groups according to gender, age, nodule blood supply, nodule size, nodule nature and Hashimoto′s thyroiditis background, and the related factors influencing microwave ablation were analyzed.Results:68 patients (106 nodules) with benign thyroid nodules were treated with microwave ablation. The volume of benign thyroid nodules after the MWA treatment was significantly reduced after 1, 3, 6 months, and their nodule volume reduction ratio (VRR) were (39.7±6.1)% (1 months), (56.2±5.9)% (3 months), (70.3±5.4)% (6 months), respectively. There were significant differences in the volume reduction ratio of nodules at 1, 3 and 6 months after operation among different nodule size, nodule nature and Hashimoto′s thyroiditis background, with statistically significant difference ( P<0.05). However, there was no significant difference in the reduction ratio of nodules in different gender, age and nodule blood supply at 1, 3 and 6 months after operation ( P>0.05). Pearson correlation analysis showed that VRR was negatively correlated with ablation time per unit volume, with statistically significant difference ( P<0.05). Logistic regression analysis indicated that only nodule nature and ablation time per unit volume entered the regression equation. Conclusions:The size and nature of the nodules, Hashimoto′s thyroiditis background and ablation time per unit volume will affect the postoperative volume reduction rate.

女性,50岁.3月前自检发现左乳肿物,大小约1 cm,无红肿热痛,近期明显增大.无乳腺癌或其他恶性肿瘤家族史.查体:左乳巨大肿物外凸,大小约15 cm×15 cm,质硬,边界不清,表面不光滑,活动度欠佳.因肿块巨大、质硬,无法行乳腺X线检查.超声所见:左乳全乳巨大低回声为主的混合回声肿块,形态欠规则,平行生长,边缘尚光整,内回声不均匀,见少量无回声,未见明显点状高回声,后方回声稍增强,周围结构未见扭曲;CDFI示肿块内部及周边可见稍丰富血流信号(图1).双侧腋下淋巴结形态结构未见明显异常.超声提示:左乳全乳巨大混合回声肿块,BI-RADS 4C类.CT平扫+增强:左乳巨大肿块影,边界尚清,密度均匀(图2),增强扫描肿块见分隔状强化及边缘强化.诊断意见:左乳占位性病变,BI-RADS 4C类.排除手术禁忌症行左乳单纯切除术,术程顺利.病理结果:结合免疫组化符合未分化多形性肉瘤.患者出院后因自身原因未行放疗,6个月后复诊出现胸壁转移,随诊复查至今.

目的:研究乳腺影像学报告及数据系统(BI-RADS)分类对肉芽肿性乳腺炎(GLM)与乳腺癌鉴别诊断的价值.方法:选取在医院就诊的87例乳腺疾病患者超声图像资料,其中43例为GLM(45个病灶),44例为乳腺癌(46个病灶),应用第5版BI-RADS规范化超声词典和分类方法进行超声描述及分类诊断,分析其在GLM与乳腺癌鉴别诊断中的乳腺组织构成、病灶的大小、形状、方位、边缘、回声模式、后方回声特征、钙化以及结构扭曲、导管改变、皮肤改变、水肿和血管分布等相关特征.结果:所有病灶均经病理证实.GLM与乳腺癌在病灶的纵横比、边缘完整性、内部及后方回声、钙化、彩色血流分布以及组织扭曲、皮肤增厚、回缩改变和水肿等相关声像学特征比较,差异具有统计学意义(x2=13.580,x2=65.188,x2=17.160,x2=14.887,x2=6.131,x2=7.893,x2=6.284,x2=38.748,x2=35.708;P<0.05);乳腺组织构成、形状和导管变化比较,差异无统计学意义.BI-RADS分类对GLM和乳腺癌的诊断准确率、灵敏度、特异度、阳性预测值和阴性预测值分别为88.0％、95.7％、80.0％、83.0％和94.7％;ROC曲线下面积为0.878.结论:超声BI-RADS分类通过标准化的描述术语和风险评估分类方法,在GLM和乳腺癌良恶性鉴别诊断评估中操作简单方便、无创有效,具有较高的临床应用价值.

To determine the methodology of non-invasive test for evaluation of liver stiffness (LS) with tumours using two-dimensional (2D) shear wave elastography (SWE).

OBJECTIVEThis study investigated the feasibility of using strain elastography (SE) and real time shear wave elastography (RT-SWE) to evaluate early tumor response to cytotoxic chemotherapy in a murine xenograft breast cancer tumor model.METHODSMCF-7 breast cancer-bearing nude mice were treated with either cisplatin 2 mg kg-1 plus paclitaxel 10 mg kg-1 (treatment group) or sterile saline (control group) once daily for 5 days. The tumor elasticity was measured by SE or RT-SWE before and after therapy. Tumor cell density was assessed by hematoxylin and eosin staining, and the ratio of collagen fibers in the tumor was evaluated by Van Gieson staining. The correlation between tumor elasticity, as determined by SE and SWE, as well as the pathological tumor responses were analyzed.RESULTSChemotherapy significantly attenuated tumor growth compared to the control treatment (p < 0.05). Chemotherapy also significantly increased tumor stiffness (p < 0.05) and significantly decreased (p < 0.05) tumor cell density compared with the control. Moreover, chemotherapy significantly increased the ratio of collagen fibers (p < 0.05). Tumor stiffness was positively correlated with the ratio of collagen fibers but negatively correlated with tumor cell density.CONCLUSIONThe study suggests that ultrasound elastography by SE and SWE is a feasible tool for assessing early responses of breast cancer to chemotherapy in our murine xenograft model. Advances in knowledge: This study showed that the tumor elasticity determined by ultrasound elastography could be a feasible imaging biomarker for assessing very early therapeutic responses to chemotherapy.

Objective To explore the optimal methodology of shear wave elastography in assessing breast lesions.Methods 51 breast nodules in 48 women were examined by shear wave elastography.Evaluating the elastographic features and parameters of the nodules in the following situations.(1) Standard or penetration scanning modes (2) moderate or enough coupling jelly.Results (1) 39 of 51 nodules displayed filling defect color pattern at standard mode,however only 3 of 51 nodules at penetration mode when using enough coupling jelly.(2) Image artifacts were reduced in 20 nodules at penetration mode when using enough coupling jelly.The elastographic parameters gotten were higher by using moderate than enough coupling jelly(P＜ 0.05).Under the above two conditions,the parameters (Emax、Emean and Esd) were both different from malignant nodules to benign ones,but with the different Az values (area under ROC curve) and cutoff values.Conclusions The optimal methodology we advanced is essential to the shear wave elastography in assessing breast nodules qualitatively and quantitatively,the correct and strict procedure,which are essential for getting precise result.

Objective To evaluate the values of RF time-series signal based on ultrasonic radio-frequency flow in the differentiation of benign and malignant breast lesions. Methods A commercially available clinical ultrasound scanner, Sonix TOUCH (Ultrasonix Medical Corporation, Richmond, Canada) with a L14–5 linear ultrasound transducer was used to simultaneously collect B-mode images and RF data from breast lesions. The ultrasound probe displaying the maximal plane of the breast lesion was kept in the same position for 10 seconds. The ultrasound RF data from region of interest was imported into software developed by our lab for ultrasound spectral analysis and 9 spectral parameters including SMR fractal dimension, Higuchi fractal dimension, Slope, Intercept, Mid-band fit, S1, S2, S3, S4 were calculated. 137 patients with 137 breast lesions confirmed by pathological or follow-up findings were included in the study. Of the 137 breast lesions, 86 malignant and 30 benign lesions were confirmed by ultrasound guided core needle biopsy or surgical excision, and the rest 21 lesions were presumed benign as no significant change was found after at least 2 years of follow-up. Results There are significantly difference in spectral parameters including SMR fractal dimension, Higuchi fractal dimension, Slope, Intercept, Mid-band fit, S1, S2, S3 and S4 between the malignant and benign breast lesions (0.75±0.77 vs 0.82±0.10, t=-4.722, 1.31±0.07 vs 1.42±0.10, t=-7.476, -0.24±0.04 vs -0.26±0.06, t=1.986, 0.19±0.03 vs 0.21±0.048, t=-3.391, 0.067±0.011 vs 0.08±0.019, t=-5.319, 3.22±0.54 vs 3.60±0.83, t=-3.298, 0.53±0.12 vs 0.73±0.23, t=-6.467, 0.31±0.06 vs 0.45±0.13, t=-9.207, 0.24±0.05 vs 0.38±0.12, t=-9.367, all P<0.05). Conclusion RF time-series signal based on ultrasonic radio-frequency flow could provide a new imaging method with a simple, low-cost noninvasive technique for the differential diagnosis of the benign and malignant breast lesions. Key words: Breast neoplasms; Ultrasonography; Diagnosis, differential

Objective To investigate the value of ultrasound-guided supraclavicular lymph nodes biopsy in the staging of non-small cell lung cancer (NSCLC) patients. Methods The present study was a retrospective review of patients with NSCLC who underwent ultrasound-guided supraclavicular lymph node biopsy from 2005 January to 2014 February. The sensitivity, specificity, negative predictive value, positive predictive value and accuracy of ultrasound–guided supraclavicular lymph node biopsy are reported. The sizes of the lymph nodes which were biopsied and the number of instances in which treatment was changed were also reported. Results A total of 586 patients underwent ultrasound-guided supraclavicular lymph node biopsy. Of the 586 patients, 543 (92.66%) had malignant findings, 30 (5.12%) had benign findings; 11 of whom were suspected to be malignant and underwent another ultrasound-guided biopsy or incisional biopsy and 2 of them had malignant findings. Thirteen of 586 had nondiagnostic biopsy findings. As a result, ultrasound-guided supraclavicular lymph node biopsy changed the tumor stage and treatment planning in 34 (5.8%) of the 586 patients. Overall, ultrasound-guided supraclavicular lymph node biopsy had a sensitivity of 99.63%, a specificity of 100%, a negative predictive value of 93.33%, a positive predictive value of 100% and an accuracy of 97.44%. No complications were found in the patients who underwent ultrasound-guided supraclavicular lymph nodes biopsy. Conclusions Ultrasound-guided supraclavicular lymph nodes biopsy was a safe and efficient option for patients with NSCLC which had suspicious supraclavicular lymph node, and also could be used as a reliable method for clinical staging of patients with NSCLC. Key words: Ultrasound-guided biopsy; Supraclavicular lymph node; Non-small cell lung cancer; staging

目的 分析肝肿瘤患者的背景肝脏实时剪切波弹性成像(SWE)模量值的相关因素及其独立预测因子.方法 收集2015年7月至2016年3月在中山大学肿瘤防治中心就诊的肝肿瘤患者132例,对所有病例均行常规超声检查及背景肝脏SWE检测,同时记录临床及实验室相关指标,变量包括性别、年龄、乙型肝炎表面抗原(HBsAg)、甲胎蛋白(AFP)、糖类蛋白19-9 (CA19-9)、癌胚抗原(CEA)、谷草转氨酶(AST)、谷丙转氨酶(ALT)、碱性磷酸酶(ALP)、谷胺酰转肽酶(GGT)、白蛋白(ALB)、总胆红素(TBIL)、直接胆红素(DBIL)、凝血酶原时间(PT)、Child-Pugh评分、肿瘤直径及位置、肝脾测量径线、门静脉及脾静脉内径.所有病例均行肝肿瘤切除术,术后获得肿瘤病理及瘤旁肝组织脂肪变性情况及肝纤维化S分期.上述变量分别与SWE背景肝脏弹性模量值作Spearman两两相关分析,筛选出显著性相关的变量进行多重线性回归分析,得出回归方程并进行方差分析,评估方程的独立预测因子.结果 HBsAg、AFP、CA19-9、ALT、AST、ALP、GGT、ALB、肝左叶前后径、脾长径(LS)、脾厚径、肝肿瘤病理类型(pathology)、肝纤维化S分期等变量与背景肝脏SWE弹性模量值均显著性相关(r=0.406、0.317、0.288、0.317、0.391、0.265、0.340、-0.269、0.230、0.414、0.365、0.416、0.597,P值均＜0.01).拟合的回归模型:SWE=0.958+0.126 S4+0.000371 GGT+0.002 LS-0.009 ALB+0.085 pathology,调整决定系数R2=0.422(F=19.996,P＜0.01).结论 肝纤维化S分期与肝肿瘤患者背景肝脏SWE模量值有较高的相关性;肝硬化(S4)是SWE测值主要的独立预测因子.

目的 通过分析无痛性亚急性甲状腺炎(SAT)的超声误诊原因,提高对本病的认识.方法 回顾性分析术前超声误诊的20例没有临床症状的SAT临床资料与声像图特征.结果 所有病例均经病理确诊.误诊为甲状腺癌(TC) 15个病灶,结节性甲状腺肿(NG)5个病灶,甲亢及桥本氏甲状腺炎(HT)各1个病灶,超声未能明确性质2个病灶.结论 无痛性SAT易误诊,当遇到局灶性边界欠清低回声甲状腺病变时,应考虑到SAT的可能性,及时结合实验室或其他影像学检查鉴别诊断,减少不必要的误诊及手术.