18F-FDG PET/CT Radiomics for Preoperative Prediction of Lymph Node Metastases and Nodal Staging in Gastric Cancer

Abstract

ObjectivesThe accurate assessment of lymph node metastases (LNMs) and the preoperative nodal (N) stage are critical for the precise treatment of patients with gastric cancer (GC). The diagnostic performance, however, of current imaging procedures used for this assessment is sub-optimal. Our aim was to investigate the value of preoperative 18F-FDG PET/CT radiomic features to predict LNMs and the N stage.MethodsWe retrospectively collected clinical and 18F-FDG PET/CT imaging data of 185 patients with GC who underwent total or partial radical gastrectomy. Patients were allocated to training and validation sets using the stratified method at a fixed ratio (8:2). There were 2,100 radiomic features extracted from the 18F-FDG PET/CT scans. After selecting radiomic features by the random forest, relevancy-based, and sequential forward selection methods, the BalancedBagging ensemble classifier was established for the preoperative prediction of LNMs, and the OneVsRest classifier for the N stage. The performance of the models was primarily evaluated by the AUC and accuracy, and validated by the independent validation methods. Analysis of the feature importance and the correlation were also conducted. We also compared the predictive performance of our radiomic models to that with the contrast-enhanced CT (CECT) and 18F-FDG PET/CT.ResultsThere were 185 patients—127 men, 58 women, with the median age of 62, and an age range of 22–86 years. One CT feature and one PET feature were selected to predict LNMs and achieved the best performance (AUC: 82.2%, accuracy: 85.2%). This radiomic model also detected some LNMs that were missed in CECT (19.6%) and 18F-FDG PET/CT (35.7%). For predicting the N stage, four CT features and one PET feature were selected (AUC: 73.7%, accuracy: 62.3%). Of note, a proportion of patients in the validation set whose LNMs were incorrectly staged by CECT (57.4%) and 18F-FDG PET/CT (55%) were diagnosed correctly by our radiomic model.ConclusionWe developed and validated two machine learning models based on the preoperative 18F-FDG PET/CT images that have a predictive value for LNMs and the N stage in GC. These predictive models show a promise to offer a potentially useful adjunct to current staging approaches for patients with GC.

Radiomics in Oncological PET Imaging: A Systematic Review—Part 2, Infradiaphragmatic Cancers, Blood Malignancies, Melanoma and Musculoskeletal Cancers

“…A total of 7 studies on gastric cancer were found [ 29 , 30 , 31 , 32 , 33 , 34 , 35 ], all using 18F-FDG as radiopharmaceutical. The average number of patients included was 163.7 (range 79–214), with 5/7 (71.4%) studies including more than 100 patients, 5/7 (71.4%) using a separate validation dataset and 1/7 (14.3%) using prospective data.…”

“…Furthermore, 4 studies were conducted for diagnostic purposes: 2 for nodal involvement prediction (AUC between 0.74 and 0.81) [ 29 , 35 ], 1 for peritoneal involvement prediction (AUC 0.88 in the validation cohorts) [ 30 ] and 1 to differentiate between gastric cancer and primary gastric lymphoma (AUC 0.77) [ 31 ]. The remaining three were prognosis-oriented [ 32 , 33 , 34 ].…”

18F-FDG PET/CT-based radiomics nomogram for the preoperative prediction of lymph node metastasis in gastric cancer

“…Other studies ( 29 – 31 ) also have shown that the combination of CT anatomical images and PET metabolic images for radiomics analysis is more effective in differential diagnosis and prognosis evaluation of diseases than PET alone, and the accuracy of predicting LNM is significantly better than that of conventional PET/CT diagnosis. Recently, a study using 18 F-FDG PET/CT radiomics features ( 31 ) randomly assigned 185 patients with GC to the training cohort and verification cohort in a ratio of 8:2 and established BalancedBagging ensemble classifier for predicting LNM in GC. Although it has different sample sizes and methodology, the PET/CT-score for predicting LNM preoperatively in the present study is similar to that of the aforementioned study ( 31 ).…”

“…Recently, a study using 18 F-FDG PET/CT radiomics features ( 31 ) randomly assigned 185 patients with GC to the training cohort and verification cohort in a ratio of 8:2 and established BalancedBagging ensemble classifier for predicting LNM in GC. Although it has different sample sizes and methodology, the PET/CT-score for predicting LNM preoperatively in the present study is similar to that of the aforementioned study ( 31 ). In clinical practice, adequate analysis of clinical and imaging data can contribute to the correct diagnosis and management of GC, so we further combined PET/CT radiomics features with clinical risk factors to construct a radiomics nomogram to predict LNM.…”

Predictors of Metastatic Lymph Nodes at Preoperative Staging CT in Gastric Adenocarcinoma

“…In future, a large multicenter cohort study or prospective study with a standardized protocol would be needed to overcome these limitations and to establish reliable volumetric ranges and sums of SA ranges for loco-regional nodal stations in GC. Secondly, measuring every single LN detected at preoperative CT is too time-consuming to be feasible at every center in everyday clinical practice and artificial intelligence could be helpful in this field, since different radiomics models, both in CT and PET/CT images analysis, proved to be effective in prediction of nodal metastases in GC [ 38 , 39 , 40 , 41 ]. Nonetheless, in this study we considered easily-measurable parameters that do not demand specific training for radiologists.…”