Purpose

To investigate differences between pituitary adenoma and craniopharyngioma on magnetic resonance imaging (MRI) with image features and three-dimensional texture features.

Materials and methods

A total of 126 patients diagnosed with pituitary adenoma (N = 63) or craniopharyngioma (N = 63) were enrolled. Qualitative magnetic resonance (MR) image features and texture features of tumors were extracted from preoperative MRI and evaluated using chi-square test or Mann–Whitney U test. Binary logistic regression analyses were performed to assess their abilities as independent diagnostic predictors, and ROC analyses were conducted to evaluate the diagnostic value of significant features. Mann–Whitney U test and ROC analyses were performed to explore the relationship between MR image features and texture features.

Results

Five MR image features were suggested to be significantly different between pituitary adenoma and craniopharyngioma. Three texture features from contrast-enhanced T1WI (HISTO-Skewness, GLCM-Contrast and GLCM-Energy), two texture features from T2WI (HISTO-Skewness and GLCM-Contrast) showed significant differences between two types of tumors. Logistic regression analyses suggested GLCM-Energy from contrast-enhanced T1WI, HISTO-Skewness and GLCM-Contrast from T2WI could be taken as independent predictors. Moreover, HISTO-Skewness and GLCM-Contrast from T2WI were found to be significantly related to cystic change.

Conclusion

MR image features and texture features were associated with each other, and both types of features represented feasible diagnostic value in discrimination between pituitary adenoma and craniopharyngioma.

中文翻译：

---

使用基于MRI的图像特征和纹理特征区分垂体腺瘤和颅咽管瘤。

目的

要研究具有图像特征和三维纹理特征的磁共振成像（MRI），垂体腺瘤和颅咽管瘤之间的差异。

材料和方法

共有126名被诊断为垂体腺瘤（N  = 63）或颅咽管瘤（N  = 63）的患者入组。从术前MRI提取肿瘤的定性磁共振（MR）图像特征和纹理特征，并使用卡方检验或Mann-Whitney U检验进行评估。进行二进制逻辑回归分析以评估其作为独立诊断预测指标的能力，并进行ROC分析以评估重要特征的诊断价值。进行了Mann-Whitney U检验和ROC分析，以探索MR图像特征与纹理特征之间的关系。

结果

垂体腺瘤和颅咽管瘤之间的五个MR图像特征被认为是显着不同的。对比增强的T1WI的三个纹理特征（HISTO-Skewness，GLCM-Contrast和GLCM-Energy），T2WI的两个纹理特征（HISTO-Skewness和GLCM-Contrast）显示出两种类型肿瘤之间的显着差异。Logistic回归分析表明，来自对比增强的T1WI的GLCM-Energy，来自T2WI的HISTO-Skewness和GLCM-Contrast可作为独立的预测因子。此外，发现来自T2WI的HISTO偏度和GLCM对比与囊性变化显着相关。

结论

MR图像特征和纹理特征相互关联，这两种类型的特征都代表着对垂体腺瘤和颅咽管瘤的鉴别诊断价值。