Department of Oral and Maxillofacial Surgery, First Affiliated Hospital of Shihezi University, Shihezi, Xinjiang, China; School of Medicine, Shihezi University, Shihezi, Xinjiang, China.
Department of Oral and Maxillofacial Surgery, First Affiliated Hospital of Shihezi University, Shihezi, Xinjiang, China.
Oral Surg Oral Med Oral Pathol Oral Radiol. 2023 Aug;136(2):240-246. doi: 10.1016/j.oooo.2023.04.011. Epub 2023 May 1.
We evaluated the diagnostic value of histogram analysis (HA) using ultrasonographic (US) images for differentiation among pleomorphic adenoma (PA), adenolymphoma (AL), and malignant tumors (MT) of the parotid gland.
Preoperative US images of 48 patients with PA, 39 patients with AL, and 17 patients with MT were retrospectively analyzed for gray-scale histograms. Nine first-order texture features derived from histograms of the tumors were compared. Area under the receiver operating characteristic curve (AUC) was used to evaluate the diagnostic performance of texture features. The Youden index maximum exponent was used to calculate sensitivity and specificity.
Statistically significant differences were discovered in Mean and Skewness HA values between PA and AL (P<0.001), and in Mean values between AL and MT (P<0.001). However, comparison of PA and MT showed no statistically significant differences (P>0.01). Excellent discrimination was detected between PA and AL (AUC=0.802), and between AL and MT (AUC=0.822). The combination of Mean plus Skewness improved discrimination between PA and AL (AUC=0.823) with sensitivity values reaching 1.00. However, Mean plus Skewness applied to differentiate PA from AL and Mean values applied to distinguish AL and MT resulted in low specificity, indicating many false positive interpretations.
Histogram analysis is useful for differentiating PA from AL and AL from MT but not PA from MT.
我们评估了基于超声图像的直方图分析（HA）对腮腺多形性腺瘤（PA）、腺淋巴瘤（AL）和恶性肿瘤（MT）的鉴别诊断价值。
回顾性分析了 48 例 PA、39 例 AL 和 17 例 MT 患者的术前超声图像，对肿瘤的灰度直方图进行了分析。比较了从直方图中提取的 9 个一阶纹理特征。利用受试者工作特征曲线（ROC）下的面积（AUC）评估纹理特征的诊断性能。采用 Youden 指数最大值指数计算灵敏度和特异性。
PA 和 AL 之间的 HA 值均值和偏度差异有统计学意义（P<0.001），AL 和 MT 之间的 HA 值均值差异有统计学意义（P<0.001）。然而，PA 和 MT 之间的比较无统计学意义（P>0.01）。PA 和 AL 之间（AUC=0.802），AL 和 MT 之间（AUC=0.822）的鉴别能力均较好。Mean 值和 Skewness 值联合可提高 PA 和 AL 之间的鉴别能力（AUC=0.823），其灵敏度值可达 1.00。然而，Mean 值和 Skewness 值联合用于区分 PA 和 AL，Mean 值用于区分 AL 和 MT 特异性较低，表明存在较多假阳性解释。
直方图分析有助于鉴别 PA 与 AL、AL 与 MT，但不能鉴别 PA 与 MT。
