Nomogram based on CT texture analysis and morphological characteristics for differentiating Borrmann Ⅳ type gastric cancer from gastric diffuse large B-cell lymphoma
10.3760/cma.j.cn112149-20220220-00134
- VernacularTitle:基于CT纹理分析和形态学特征的列线图鉴别Borrmann Ⅳ型胃癌与胃弥漫大B细胞淋巴瘤的价值
- Author:
Changfeng JI
1
;
Song LIU
;
Xiangmei QIAO
;
Ling CHEN
;
Han WANG
;
Yiwen SUN
;
Kefeng ZHOU
;
Zhengyang ZHOU
Author Information
1. 南京医科大学鼓楼临床医学院医学影像科，南京 210008
- Keywords:
Stomach neoplasms;
Lymphoma;
Tomography, X-ray computed;
Texture analysis;
Nomogram
- From:
Chinese Journal of Radiology
2023;57(4):397-403
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To explore the value in differentiating Borrmann Ⅳ type gastric cancer (BT4-GC) from gastric diffuse large B-cell lymphoma (DLBCL) using a nomogram based on CT texture analysis (CTTA) and morphological characteristics.Methods:From June 2011 to December 2020, a total of 60 patients with BT4-GC and 24 patients with DLBCL were retrospectively collected in Nanjing Drum Tower Hospital Clinical College of Nanjing Medical University. Morphological characteristics were evaluated, including major location, long axis range, circumferential range, mucosal line status, and perigastric enlarged lymph nodes. CTTA parameters were calculated using venous CT images with a manual region of interest. The morphological characteristics and CTTA parameters between BT4-GC and DLBCL were compared by χ 2 test, Fisher exact test or Mann-Whitney U test. The multivariate binary logistic regression analysis was used to filter factors into the diagnostic model and construct a nomogram. The receiver operating characteristic (ROC) curve was used to evaluate the diagnostic performance of CTTA parameters and the diagnostic model in differentiating BT4-GC from DLBCL. Results:For morphological characteristics, mucosal line status showed a significant difference between BT4-GC and DLBCL (χ 2=12.99, P<0.001). For CTTA parameters, 16 parameters showed significant differences between BT4-GC and DLBCL (all P<0.05). The area under the ROC curve (AUC) of 16 CTTA parameters in differentiating BT4-GC from DLBCL was 0.662-0.833. Percentile 90 showed the highest AUC of 0.833 (95%CI 0.736-0.906). The mucosal line status (OR 4.82, 95%CI 1.21-19.25, P=0.026) and percentile 90 (OR 1.09, 95%CI 1.04-1.15, P=0.001) were brought into the diagnostic model and constructed a nomogram. The AUC of the model in differentiating BT4-GC from DLBCL was 0.898 (95%CI 0.813-0.953), sensitivity was 0.833, and specificity was 0.817. Conclusions:The nomogram based on CTTA percentile 90 and morphological characteristics mucosal line status can effectively distinguish BT4-GC from DLBCL and shows high diagnostic efficacy.
