Abstract: The purpose of this study was to explore the application of energy spectrum CT omics nomogram in the detection and misdiagnosis analysis of gastric neurosecretory tumors. A total of 34 patients with gastric neurosecretory tumors were selected and divided into training set (25 cases) and test set (9 cases) according to the completely random method. Energy spectrum CT imaging omics analysis of lesion characteristics; LASSO regression to establish imaging omics risk score; Bootstrap method to analyze the AUC, specificity, sensitivity and misdiagnosis rate under the ROC curve. The training set and the test set showed significant differences in tumor size, lymph node enlargement, and venous tumor thrombus (P<0.05). Compared with low-grade patients, high-grade patients had higher risk scores (P<0.05). Validated by Bootstrap method, the misdiagnosis rate in the training set is 8.00%, AUC 0.940, specificity 92.0%, and sensitivity 95.4%; the misdiagnosis rate in the test set is 11.11%, AUC 0.859, specificity 88.89%, and sensitivity 77.6%. In conclusion, the power spectrum CT omics nomogram has good predictive performance in the detection and misdiagnosis analysis of gastric neurosecretory tumors.

Key words: gastric neurosecretory tumor, clinical effect, energy spectrum CT omics nomogram, misdiagnosis rate