基于经验小波变换的超声射频信号特征在乳腺病变良恶性鉴别中的临床价值

打开文本图片集

ABSTRACTObjectiveTo extract ultrasonicradirequencysignalfeaturesbasedonempirical wavelettransform （EWT），andtoinvestigateclinicalvaluediferentfeaturecombinationsindiferentiatingbenignfrommalignantbreast lesions.MethodsUtrasonicradirequencysignalsfromtransverseand longitudinal sections1OO breast lesionareas werecolleced.QuantitativeanalysiswasperformedbyEWTonregion interest withinlesion，andstatisticalfeatures empiricalmodefunctions（EMFs）in2-9MHzfrequencybandwereextracted，including mean，variance，medianrtosis， root mean square（RMS），and skewnessResultsWhen frequency band was EMFs2-8， feature combination was variance，kurtosis，andRMS，andsize heregioninterestinbreastlesionwas4Ox40pixels，averageclaiication accuracy rates based on 10-fold cross-validation using support vector machine and random forest algorithm were 92.30%± （20 2.67% and 92.50%±1.78% ，respectively.ConclusionA quantitative analysis method based on extraction ultrasonic radirequencysignalfeatures viaEWTprovidesanewandefective approachfordifferential diagnosis betweenbenignand malignant breastlesions，and demonstrates good clinical application value.

KEYWORDS Breast lesions，benign and malignant；Ultrasonic radirequency signal;Empirical mode functions; ）ifferential diagnosis

乳腺癌作为女性发病率最高的恶性肿瘤，严重威胁着其生命健康。（剩余8434字）