多层螺旋CT对乳腺形态特征与血流灌注检测结果的Fisher判别

Fisher discriminant of morphological and blood perfusion findings of breast lesions by multislice spiral CT

目的:评价Fisher判别模式应用于乳腺多层螺旋CT(MSCT)检测结果分析的临床价值,并在此模式下比较乳腺MSCT形态特征、血流灌注及两者综合的临床应用价值。方法:对81例体检或钼靶发现乳腺占位性病变者行肿块(84个)MSCT平扫+组织灌注扫描,按病理随访结果分为良性组及恶性组两大类。对CT主要形态学特征(钙化、毛刺、坏死、分叶)进行评价,按病理良、恶性分组,以两组间有统计学意义的形态学征象为判别指标并行Fisher判别。由去卷积法得出血流灌注参数:血流量(bloodflow,BF)、血容量(bloodvolume,BV)、平均通过时间(meantransittime,MTT),按病理良、恶性分组,以两组间有统计学意义的参数为判别指标行Fisher判别。综合形态学特征及血流灌注参数行Fisher判别。采用刀切法(jack-knife)估计误判概率。结果:以CT形态学征象为判别指标者误判率为16.67%,以血流灌注参数为判别指标者误判率为23.81%,两方面综合判别的误判率为8.33%。结论:Fisher判别模式可有效应用于乳腺MSCT检测结果的分析;乳腺疾病的CT形态学特征是其良恶性定性诊断的较可靠依据,血流灌注参数是有效的参考指标,两者综合应用可提高临床诊断价值。

Objective: To evaluate the diagnostic value of Fisher discriminant in analyzing the breast findings by multislice spiral CT(MSCT) and to compare the clinical value of the morphological appearance, perfusion parameters and integration of both under the mode of Fisher discriminant. Methods: Eighty-four masses were found in 81 people during health examination or mammography and they were scanned by MSCT with routine and perfusion sequence. Benign and malignant groups were classified according to pathological results after surgery or biopsy. The CT appearances(calcified, spiculated, necrotic, or lobulated) were evaluated and Fisher discriminant was applied to the parameters with significant difference. Hemodynamic parameters were obtained by deconvolution arithmetic (blood flow, blood volume, mean transit time), and Fisher discriminant was applied to the parameters with significant difference. Finally, Fisher discriminant was also applied to the integration of morphological and hemodynamic findings. Jackknife method was used for estimating the rate of miscarriage of justice. Results: According to Fisher discriminant, the rates of miscarriage of justice based on morphological appearance, hemodynamic parameters and integration of both were 16.67%, 23.81% and 8.33%, respectively. Conclusion: Fisher discriminant is effective in analyzing the findings of breast lesions on MSCT. CT appearance of breast is reliable to determine the nature of breast tumors and the hemodynamic parameters are effective references. Clinical diagnosis can be more accurate by the combination of both.