Please click here to download the model.
Nomogram_OPM_AGC: CT radiomic nomogram for preoperatively identify occult peritoneal metastasis (PM) in patients with advanced gastric cancer (AGC). The nomogram combines the signatures of primary tumor and nearby peritoneal region, as well as the Lauren type. We suggest the clinicians and physicians to use it to avoid suboptimal clinical practice, such as the improper use of surgical treatment for occult peritoneal metastasis patient which negatively affects patient prognosis. To use this model, one should outline the 2D primary tumor (ROI-1) in one slice with the maximum-tumor-area, and also outline the 2D peritoneum region (ROI-2, area>2cm2) nearest to the center of the primary tumor. All these processes are done in venous phase CT. Besides, Lauren type is also required for better prediction performance. The signatures of primary tumor (RS1) and the nearby peritoneum (RS2) will extracted from ROI-1 and ROI-2 respectively. The two signatures and Lauren type are input into the radiomic nomogram. Then the possible PM status will be given by the nomogram.
Please click here to download the dataset.
We share the venous phase CT imaging data and clinical characteristics of 554 advanced gastric cancer (AGC).
CT imaging data: For each patient, we share the DICOM files of one slice with maximum-tumor-area, and one slice with its nearby peritoneum. Moreover, we share the segmentation results of the primary tumor and peritoneum by the radiologists. The segmentation results can be opened with ITK SNAP software (www.itksnap.org).
Clinical characteristics: For eaeach patient, we share the clinical characteristics including: Sex (male or female), Age, mild CT-defined ascites (+ or -), Locations (gastric antrum, gastric body, esophagogastric junction, whole stomach), Pathology (adenocarcinoma, signet ring and mucinous cell carcinoma, Differentiation, (poorly differentiated moderately or well differentiated), Lauren type (intestinal type and mixed type, or diffuse type), Borrmann type ( type 2 and type 3, or type 4), CEA (normal or elevated), and CA19-9 (normal or elevated).
Radiomic features: For each patient, we share the 546 features from the primary tumor and peritoneum region.
Occult peritoneal metastasis (PM) in advanced gastric cancer (AGC) patients is highly possible to be missed on CT images, and the patients have high risk to be lately detected during the subsequent laparotomy and may receive improper surgical procedures. We aimed to develop a noninvasive radiomic nomogram to individually identify occult PMs in AGC.
A total of 554 AGC patients from four centers were divided into a training cohort, an internal validation cohort, and two external validation cohorts. All patients were diagnosed as free from PM by CT, but confirmed the PM status by laparoscopy (occult PM positive n=122, real PM negative n=432). 546 quantitative image features were extracted from both the primary tumor and nearby peritoneum region on CT. Then, radiomic signatures reflecting phenotypes of primary tumor (RS1) and peritoneal (RS2) were built as predictors of PM via machine learning methods. An individualized nomogram incorporating RS1, RS2, and clinical factors was constructed using multivariable logistic regression and assessed by stratification analysis, receiver operating characteristic (ROC) curve, and calibration curve.
Multivariable analysis revealed RS1, RS2, and the clinical factor of Lauren type were significant predictors for occult PM (all p <0·05). The nomogram combining RS1, RS2 and Lauren type had powerful predictive ability with AUCs of 0·945 (95% confidence interval [CI]: 0·906-0·985) in the training cohort, 0·880 (95% CI: 0·820-0·941), 0·883 (95% CI: 0·816-0·951) and 0·892 (95% CI: 0·821-0·964) in the three validation cohorts. Moreover, the nomogram demonstrated better diagnostic accuracy than the model with only RS1, RS2, or clinical factors (net reclassification improvement p<0·05). Furthermore, the stratification analysis showed that our nomogram had potential generalization ability and it was not affected by patient sex, age, BMI, version of CT, type of CT contrast agent, contrast agent concentration, contrast agent infused rate or image thickness (Delong test p>0·05).
CT phenotypes of both primary tumor and nearby peritoneum are significantly associated with occult PM status. Through internal and external validation, a nomogram built based on CT phenotypes of primary tumor and nearby peritoneum as well as Lauren type has an excellent predictive ability of occult PM in AGC.