本研究应用图像分析技术测量了４８例（良、恶性各２４例）乳腺肿块细针穿刺涂片的细胞学参数，发现恶性组平均细胞核面积、周长、等效直径、圆度系数、各灰度参数、质量以及各参数标准差，均明显大于良性组；并应用ＳＡＳ软件包建立了Ｂａｙｅ＇ｓ判别模型，通过回代检验，灵敏度为９３．３３％，特异度为１００％。 This study applied image analysis techniques to measure the cytological parameters of fine needle aspiration smears of 48 cases (24 cases of benign and malignant) of breast masses, and found that the mean nuclei area, perimeter, equivalent diameter, roundness factor, and gray levels of the malignant group. Degree parameter, quality and standard deviation of each parameter were significantly larger than benign group; and Bayesian’s discriminant model was established by SAS software package. The sensitivity was 93.33% and specificity was 100%.