Predictive features of CT for risk stratifications in patients with primary gastrointestinal stromal tumour. Eur Radiol. 2016;26:3086-3093. [PMID: 26699371 DOI: 10.1007/s00330-015-4172-7]

Preoperative CT-based radiomics and deep learning model for predicting risk stratification of gastric gastrointestinal stromal tumors

BACKGROUND

Gastrointestinal stromal tumors (GISTs) are clinically heterogeneous with various malignant potential in different individuals. It is crucial to explore a reliable method for preoperative risk stratification of gastric GISTs noninvasively.

PURPOSE

To establish and evaluate a machine learning model using the combination of computed tomography (CT) morphology, radiomics, and deep learning features to predict the risk stratification of primary gastric GISTs preoperatively.

METHODS

The 193 gastric GISTs lesions were randomly divided into training set, validation set, and test set in a ratio of 6:2:2. The qualitative and quantitative CT morphological features were assessed by two radiologists. The tumors were segmented manually, and then radiomic features were extracted using PyRadiomics and the deep learning features were extracted using pre-trained Resnet50 from arterial phase and venous phase CT images, respectively. Pearson correlation analysis and recursive feature elimination were used for feature selection. Support vector machines were employed to build a classifier for predicting the risk stratification of GISTs. This study compared the performance of models using different pre-trained convolutional neural networks (CNNs) to extract deep features for classification, as well as the performance of modeling features from single-phase and dual-phase images. The arterial phase, venous phase and dual-phase machine learning models were built, respectively, and the morphological features were added to the dual-phase machine learning model to construct a combined model. Receiver operating characteristic (ROC) curves were used to evaluate the efficacy of each model. The clinical application value of the combined model was determined through the decision curve analysis (DCA) and the net reclassification index (NRI) was analyzed.

RESULTS

The area under the curve (AUC) of the dual-phase machine learning model was 0.876, which was higher than that of the arterial phase model or venous phase model (0.813, 0.838, respectively). The combined model had best predictive performance than the above models with an AUC of 0.941 (95% CI: 0.887-0.974) (p = 0.012, Delong test). DCA demonstrated that the combined model had good clinical application value with an NRI of 0.575 (95% CI: 0.357-0.891).

CONCLUSION

In this study, we established a combined model that incorporated dual-phase morphology, radiomics, and deep learning characteristics, which can be used to predict the preoperative risk stratification of gastric GISTs.

Deep Learning-Based Segmentation and Risk Stratification for Gastrointestinal Stromal Tumors in Transabdominal Ultrasound Imaging

PURPOSE

To develop a deep neural network system for the automatic segmentation and risk stratification prediction of gastrointestinal stromal tumors (GISTs).

METHODS

A total of 980 ultrasound (US) images from 245 GIST patients were retrospectively collected. These images were randomly divided (6:2:2) into a training set, a validation set, and an internal test set. Additionally, 188 US images from 47 prospective GIST patients were collected to evaluate the segmentation and diagnostic performance of the model. Five deep learning-based segmentation networks, namely, UNet, FCN, DeepLabV3+, Swin Transformer, and SegNeXt, were employed, along with the ResNet 18 classification network, to select the most suitable network combination. The performance of the segmentation models was evaluated using metrics such as the intersection over union (IoU), Dice similarity coefficient (DSC), recall, and precision. The classification performance was assessed based on accuracy and the area under the receiver operating characteristic curve (AUROC).

RESULTS

Among the compared models, SegNeXt-ResNet18 exhibited the best segmentation and classification performance. On the internal test set, the proposed model achieved IoU, DSC, precision, and recall values of 82.1, 90.2, 91.7, and 88.8%, respectively. The accuracy and AUC for GIST risk prediction were 87.4 and 92.0%, respectively. On the external test set, the segmentation models exhibited IoU, DSC, precision, and recall values of 81.0, 89.5, 92.8, and 86.4%, respectively. The accuracy and AUC for GIST risk prediction were 86.7 and 92.5%, respectively.

CONCLUSION

This two-stage SegNeXt-ResNet18 model achieves automatic segmentation and risk stratification prediction for GISTs and demonstrates excellent segmentation and classification performance.

Clinical outcomes of endoscopic resection for the treatment of intermediate- or high-risk gastric small gastrointestinal stromal tumors: a multicenter retrospective study

BACKGROUND AND AIMS

Many studies of gastric gastrointestinal stromal tumors (g-GISTs) following endoscopic resection (ER) have typically focused on tumor size, with most tumors at low risk of aggressiveness after risk stratification. There have been few systematic studies on the oncologic outcomes of intermediate- or high-risk g-GISTs after ER.

METHODS

From January 2014 to January 2020, we retrospectively collected patients considered at intermediate- or high-risk of g-GISTs according to the modified NIH consensus classification system. The primary outcome was overall survival (OS).

RESULTS

Six hundred and seventy nine (679) consecutive patients were diagnosed with g-GISTs and treated by ER between January 2014 and January 2020 in three hospitals in Shanghai, China. 43 patients (20 males and 23 females) were confirmed at intermediate-or high-risk. The mean size of tumors was 2.23 ± 1.01 cm. The median follow-up period was 62.02 ± 15.34 months, with a range of 28 to 105 months. There were no recurrences or metastases, even among patients having R1 resections. The 5-year OS rate was 97.4% (42/43).

CONCLUSION

ER for intermediate- or high-risk gastric small GISTs is a feasible and safe method, which allows for a wait-and-see approach before determining the necessity for imatinib adjuvant or surgical treatment. This approach to g-GISTs does require that patients undergo close follow-up.

Computed tomography radiogenomics: A potential tool for prediction of molecular subtypes in gastric stromal tumor

BACKGROUND

Preoperative knowledge of mutational status of gastrointestinal stromal tumors (GISTs) is essential to guide the individualized precision therapy.

AIM

To develop a combined model that integrates clinical and contrast-enhanced computed tomography (CE-CT) features to predict gastric GISTs with specific genetic mutations, namely KIT exon 11 mutations or KIT exon 11 codons 557-558 deletions.

METHODS

A total of 231 GIST patients with definitive genetic phenotypes were divided into a training dataset and a validation dataset in a 7:3 ratio. The models were constructed using selected clinical features, conventional CT features, and radiomics features extracted from abdominal CE-CT images. Three models were developed: ModelCT sign, modelCT sign + rad, and model CTsign + rad + clinic. The diagnostic performance of these models was evaluated using receiver operating characteristic (ROC) curve analysis and the Delong test.

RESULTS

The ROC analyses revealed that in the training cohort, the area under the curve (AUC) values for modelCT sign, modelCT sign + rad, and modelCT sign + rad + clinic for predicting KIT exon 11 mutation were 0.743, 0.818, and 0.915, respectively. In the validation cohort, the AUC values for the same models were 0.670, 0.781, and 0.811, respectively. For predicting KIT exon 11 codons 557-558 deletions, the AUC values in the training cohort were 0.667, 0.842, and 0.720 for modelCT sign, modelCT sign + rad, and modelCT sign + rad + clinic, respectively. In the validation cohort, the AUC values for the same models were 0.610, 0.782, and 0.795, respectively. Based on the decision curve analysis, it was determined that the modelCT sign + rad + clinic had clinical significance and utility.

CONCLUSION

Our findings demonstrate that the combined modelCT sign + rad + clinic effectively distinguishes GISTs with KIT exon 11 mutation and KIT exon 11 codons 557-558 deletions. This combined model has the potential to be valuable in assessing the genotype of GISTs.

Preoperative differentiation of gastric schwannomas and gastrointestinal stromal tumors based on computed tomography: a retrospective multicenter observational study

Introduction

Gastric schwannoma is a rare benign tumor accounting for only 1-2% of alimentary tract mesenchymal tumors. Owing to their low incidence rate, most cases are misdiagnosed as gastrointestinal stromal tumors (GISTs), especially tumors with a diameter of less than 5 cm. Therefore, this study aimed to develop and validate a diagnostic nomogram based on computed tomography (CT) imaging features for the preoperative prediction of gastric schwannomas and GISTs (diameters = 2-5 cm).

Methods

Gastric schwannomas in 47 patients and GISTs in 230 patients were confirmed by surgical pathology. Thirty-four patients with gastric schwannomas and 167 with GISTs admitted between June 2009 and August 2022 at Hospital 1 were retrospectively analyzed as the test and training sets, respectively. Seventy-six patients (13 with gastric schwannomas and 63 with GISTs) were included in the external validation set (June 2017 to September 2022 at Hospital 2). The independent factors for differentiating gastric schwannomas from GISTs were obtained by multivariate logistic regression analysis, and a corresponding nomogram model was established. The accuracy of the nomogram was evaluated using receiver operating characteristic and calibration curves.

Results

Logistic regression analysis showed that the growth pattern (odds ratio [OR] 3.626; 95% confidence interval [CI] 1.105-11.900), absence of necrosis (OR 4.752; 95% CI 1.464-15.424), presence of tumor-associated lymph nodes (OR 23.978; 95% CI 6.499-88.466), the difference between CT values during the portal and arterial phases (OR 1.117; 95% CI 1.042-1.198), and the difference between CT values during the delayed and portal phases (OR 1.159; 95% CI 1.080-1.245) were independent factors in differentiating gastric schwannoma from GIST. The resulting individualized prediction nomogram showed good discrimination in the training (area under the curve [AUC], 0.937; 95% CI, 0.900-0.973) and validation (AUC, 0.921; 95% CI, 0.830-1.000) datasets. The calibration curve showed that the probability of gastric schwannomas predicted using the nomogram agreed well with the actual value.

Conclusion

The proposed nomogram model based on CT imaging features can be used to differentiate gastric schwannoma from GIST before surgery.

Predicting the risk stratification of gastrointestinal stromal tumors using machine learning-based ultrasound radiomics

PURPOSE

This study aimed to use conventional ultrasound features, ultrasound radiomics, and machine learning algorithms to establish a predictive model to assess the risk of post-surgical recurrence of gastrointestinal stromal tumors (GISTs).

METHODS

This retrospective analysis included 230 patients with pathologically diagnosed GISTs. Radiomic features were extracted from manually annotated images. Radiomic features plus conventional ultrasound features were selected using the SelectKbest analysis of variance and stratified tenfold cross-validation recursive elimination methods. Finally, five different machine learning algorithms (logistic regression [LR], support vector machine [SVM], random forest [RF], extreme gradient boosting [XGBoost], and multilayer perceptron [MLP]) were established to predict risk stratification of GISTs. The predictive performance of the established model was mainly evaluated based on the area under the receiver operating characteristic (ROC) curve (AUC) and accuracy, whereas the predictive performance of the optimal machine learning algorithm and a radiologist's subjective assessment were compared using McNemar's test.

RESULTS

Seven radiomics features and one conventional ultrasound feature were selected to construct the machine learning models for GIST risk classification. The mentioned five machine learning models were able to predict the malignant potential of GISTs. LR and SVM outperformed other classifiers on the test set, with LR achieving an accuracy of 0.852 (AUC, 0.881; sensitivity, 0.871; specificity, 0.826) and SVM achieving an accuracy of 0.852 (AUC, 0.879; sensitivity, 0.839; specificity, 0.870), and proved significantly better than the radiologist (accuracy, 0.691; sensitivity, 0.645; specificity, 0.813).

CONCLUSION

Machine learning-based ultrasound radiomics features are able to noninvasively predict the biological risk of GISTs.

Possibility of Using Conventional Computed Tomography Features and Histogram Texture Analysis Parameters as Imaging Biomarkers for Preoperative Prediction of High-Risk Gastrointestinal Stromal Tumors of the Stomach

BACKGROUND

The objective of this study is to determine the morphological computed tomography features of the tumor and texture analysis parameters, which may be a useful diagnostic tool for the preoperative prediction of high-risk gastrointestinal stromal tumors (HR GISTs).

METHODS

This is a prospective cohort study that was carried out in the period from 2019 to 2022. The study included 79 patients who underwent CT examination, texture analysis, surgical resection of a lesion that was suspicious for GIST as well as pathohistological and immunohistochemical analysis.

RESULTS

Textural analysis pointed out min norm (p = 0.032) as a histogram parameter that significantly differed between HR and LR GISTs, while min norm (p = 0.007), skewness (p = 0.035) and kurtosis (p = 0.003) showed significant differences between high-grade and low-grade tumors. Univariate regression analysis identified tumor diameter, margin appearance, growth pattern, lesion shape, structure, mucosal continuity, enlarged peri- and intra-tumoral feeding or draining vessel (EFDV) and max norm as significant predictive factors for HR GISTs. Interrupted mucosa (p < 0.001) and presence of EFDV (p < 0.001) were obtained by multivariate regression analysis as independent predictive factors of high-risk GISTs with an AUC of 0.878 (CI: 0.797-0.959), sensitivity of 94%, specificity of 77% and accuracy of 88%.

CONCLUSION

This result shows that morphological CT features of GIST are of great importance in the prediction of non-invasive preoperative metastatic risk. The incorporation of texture analysis into basic imaging protocols may further improve the preoperative assessment of risk stratification.

Multi-Slice CT Features Predict Pathological Risk Classification in Gastric Stromal Tumors Larger Than 2 cm: A Retrospective Study

BACKGROUND

The Armed Forces Institute of Pathology (AFIP) had higher accuracy and reliability in prognostic assessment and treatment strategies for patients with gastric stromal tumors (GSTs). The AFIP classification is frequently used in clinical applications. But the risk classification is only available for patients who are previously untreated and received complete resection. We aimed to investigate the feasibility of multi-slice MSCT features of GSTs in predicting AFIP risk classification preoperatively.

METHODS

The clinical data and MSCT features of 424 patients with solitary GSTs were retrospectively reviewed. According to pathological AFIP risk criteria, 424 GSTs were divided into a low-risk group (n = 282), a moderate-risk group (n = 72), and a high-risk group (n = 70). The clinical data and MSCT features of GSTs were compared among the three groups. Those variables (p < 0.05) in the univariate analysis were included in the multivariate analysis. The nomogram was created using the rms package.

RESULTS

We found significant differences in the tumor location, morphology, necrosis, ulceration, growth pattern, feeding artery, vascular-like enhancement, fat-positive signs around GSTs, CT value in the venous phase, CT value increment in the venous phase, longest diameter, and maximum short diameter (all p < 0.05). Two nomogram models were successfully constructed to predict the risk of GSTs. Low- vs. high-risk group: the independent risk factors of high-risk GSTs included the location, ulceration, and longest diameter. The area under the receiver operating characteristic curve (AUC) of the prediction model was 0.911 (95% CI: 0.872-0.951), and the sensitivity and specificity were 80.0% and 89.0%, respectively. Moderate- vs. high-risk group: the morphology, necrosis, and feeding artery were independent risk factors of a high risk of GSTs, with an AUC value of 0.826 (95% CI: 0.759-0.893), and the sensitivity and specificity were 85.7% and 70.8%, respectively.

CONCLUSIONS

The MSCT features of GSTs and the nomogram model have great practical value in predicting pathological AFIP risk classification between high-risk and non-high-risk groups before surgery.

Clinical-radiomics-based treatment decision support for KIT Exon 11 deletion in gastrointestinal stromal tumors: a multi-institutional retrospective study

Objective

gastrointestinal stromal tumors (GISTs) with KIT exon 11 deletions have more malignant clinical outcomes. A radiomics model was constructed for the preoperative prediction of KIT exon 11 deletion in GISTs.

Methods

Overall, 126 patients with GISTs who underwent preoperative enhanced CT were included. GISTs were manually segmented using ITK-SNAP in the arterial phase (AP) and portal venous phase (PVP) images of enhanced CT. Features were extracted using Anaconda (version 4.2.0) with PyRadiomics. Radiomics models were constructed by LASSO. The clinical-radiomics model (combined model) was constructed by combining the clinical model with the best diagnostic effective radiomics model. ROC curves were used to compare the diagnostic effectiveness of radiomics model, clinical model, and combined model. Diagnostic effectiveness among radiomics model, clinical model and combine model were analyzed in external cohort (n=57). Statistics were carried out using R 3.6.1.

Results

The Radscore showed favorable diagnostic efficacy. Among all radiomics models, the AP-PVP radiomics model exhibited excellent performance in the training cohort, with an AUC of 0.787 (95% CI: 0.687-0.866), which was verified in the test cohort (AUC=0.775, 95% CI: 0.608-0.895). Clinical features were also analyzed. Among the radiomics, clinical and combined models, the combined model showed favorable diagnostic efficacy in the training (AUC=0.863) and test cohorts (AUC=0.851). The combined model yielded the largest AUC of 0.829 (95% CI, 0.621-0.950) for the external validation of the combined model. GIST patients could be divided into high or low risk subgroups of recurrence and mortality by the Radscore.

Conclusion

The radiomics models based on enhanced CT for predicting KIT exon 11 deletion mutations have good diagnostic performance.

Development of a CT image analysis-based scoring system to differentiate gastric schwannomas from gastrointestinal stromal tumors

Purpose

To develop a point-based scoring system (PSS) based on contrast-enhanced computed tomography (CT) qualitative and quantitative features to differentiate gastric schwannomas (GSs) from gastrointestinal stromal tumors (GISTs).

Methods

This retrospective study included 51 consecutive GS patients and 147 GIST patients. Clinical and CT features of the tumors were collected and compared. Univariate and multivariate logistic regression analyses using the stepwise forward method were used to determine the risk factors for GSs and create a PSS. Area under the receiver operating characteristic curve (AUC) analysis was performed to evaluate the diagnostic efficiency of PSS.

Results

The CT attenuation value of tumors in venous phase images, tumor-to-spleen ratio in venous phase images, tumor location, growth pattern, and tumor surface ulceration were identified as predictors for GSs and were assigned scores based on the PSS. Within the PSS, GS prediction probability ranged from 0.60% to 100% and increased as the total risk scores increased. The AUC of PSS in differentiating GSs from GISTs was 0.915 (95% CI: 0.874-0.957) with a total cutoff score of 3.0, accuracy of 0.848, sensitivity of 0.843, and specificity of 0.850.

Conclusions

The PSS of both qualitative and quantitative CT features can provide an easy tool for radiologists to successfully differentiate GS from GIST prior to surgery.

Relationship between multi-slice computed tomography features and pathological risk stratification assessment in gastric gastrointestinal stromal tumors

BACKGROUND

Computed tomography (CT) imaging features are associated with risk stratification of gastric gastrointestinal stromal tumors (GISTs).

AIM

To determine the multi-slice CT imaging features for predicting risk stratification in patients with primary gastric GISTs.

METHODS

The clinicopathological and CT imaging data for 147 patients with histologically confirmed primary gastric GISTs were retrospectively analyzed. All patients had received dynamic contrast-enhanced CT (CECT) followed by surgical resection. According to the modified National Institutes of Health criteria, 147 lesions were classified into the low malignant potential group (very low and low risk; 101 lesions) and high malignant potential group (medium and high-risk; 46 lesions). The association between malignant potential and CT characteristic features (including tumor location, size, growth pattern, contour, ulceration, cystic degeneration or necrosis, calcification within the tumor, lymphadenopathy, enhancement patterns, unenhanced CT and CECT attenuation value, and enhancement degree) was analyzed using univariate analysis. Multivariate logistic regression analysis was performed to identify significant predictors of high malignant potential. The receiver operating curve (ROC) was used to evaluate the predictive value of tumor size and the multinomial logistic regression model for risk classification.

RESULTS

There were 46 patients with high malignant potential and 101 with low-malignant potential gastric GISTs. Univariate analysis showed no significant differences in age, gender, tumor location, calcification, unenhanced CT and CECT attenuation values, and enhancement degree between the two groups (P > 0.05). However, a significant difference was observed in tumor size (3.14 ± 0.94 vs 6.63 ± 3.26 cm, P < 0.001) between the low-grade and high-grade groups. The univariate analysis further revealed that CT imaging features, including tumor contours, lesion growth patterns, ulceration, cystic degeneration or necrosis, lymphadenopathy, and contrast enhancement patterns, were associated with risk stratification (P < 0.05). According to binary logistic regression analysis, tumor size [P < 0.001; odds ratio (OR) = 26.448; 95% confidence interval (CI): 4.854-144.099)], contours (P = 0.028; OR = 7.750; 95%CI: 1.253-47.955), and mixed growth pattern (P = 0.046; OR = 4.740; 95%CI: 1.029-21.828) were independent predictors for risk stratification of gastric GISTs. ROC curve analysis for the multinomial logistic regression model and tumor size to differentiate high-malignant potential from low-malignant potential GISTs achieved a maximum area under the curve of 0.919 (95%CI: 0.863-0.975) and 0.940 (95%CI: 0.893-0.986), respectively. The tumor size cutoff value between the low and high malignant potential groups was 4.05 cm, and the sensitivity and specificity were 93.5% and 84.2%, respectively.

CONCLUSION

CT features, including tumor size, growth patterns, and lesion contours, were predictors of malignant potential for primary gastric GISTs.

Prediction of the mitotic index and preoperative risk stratification of gastrointestinal stromal tumors with CT radiomic features

OBJECTIVE

The objective is to develop a mitotic prediction model and preoperative risk stratification nomogram for gastrointestinal stromal tumor (GIST) based on computed tomography (CT) radiomic features.

METHODS

A total of 267 GIST patients from 2009.07 to 2015.09 were retrospectively collected and randomly divided into (6:4) training cohort and validation cohort. The 2D-tumor region of interest was delineated from the portal-phase images on contrast-enhanced (CE)-CT, and radiomic features were extracted. Lasso regression method was used to select valuable features to establish a radiomic model for predicting mitotic index in GIST. Finally, the nomogram of preoperative risk stratification was constructed by combining the radiomic features and clinical risk factors.

RESULTS

Four radiomic features closely related to the level of mitosis were obtained, and a mitotic radiomic model was constructed. The area under the curve (AUC) of the radiomics signature model used to predict mitotic levels in training and validation cohorts (training cohort AUC = 0.752; 95% confidence interval [95%CI] 0.674-0.829; validation cohort AUC = 0.764; 95% CI 0.667-0.862). Finally, the preoperative risk stratification nomogram combining radiomic features was equivalent to the clinically recognized gold standard AUC (0.965 vs. 0.983) (p = 0.117). The Cox regression analysis found that the nomogram score was one of the independent risk factors for the long-term prognosis of the patients.

CONCLUSION

Preoperative CT radiomic features can effectively predict the level of mitosis in GIST, and combined with preoperative tumor size, accurate preoperative risk stratification can be performed to guide clinical decision-making and individualized treatment.

The relationship between Ki-67 expression and imaging signs and pathological features in GISTs

Introduction

To investigate the correlations between the Ki-67 index and plain-scan computerized tomography (CT) signs and pathological features of gastrointestinal stromal tumor (GIST) tissue.

Materials and methods

Data from 186 patients with GIST diagnosed by pathology and immunohistochemistry (IHC) in Peking University First Hospital from May 2016 to May 2022 were analyzed. The patients were divided into two groups: Ki-67 ≤5% and &gt;5%. Correlation analysis, univariate and multivariate Logistic regression analysis were used to explore the correlations between CT signs, pathological features, and Ki-67 expression.

Results

Univariate indicators correlated with the Ki-67 index were mitotic count, pathological grade, tumor hemorrhage, tumor necrosis, tumor size, and tumor density. Multivariate Logistic regression indicated that the mitotic count [odds ratio (OR) 10.222, 95% confidence interval (CI) 4.312–31.039], pathological grade (OR 2.139, 95% CI 1.397–3.350), and tumor size (OR 1.096, 95% CI 1.020–1.190) were independently associated with the Ki-67 expression level. The concordance indexes (C-index) for the pathological features and CT signs models were 0.876 (95% CI 0.822–0.929) and 0.697 (95% CI 0.620–0.774), respectively, with positive predictive values of 93.62% and 58.11% and negative predictive values of 81.29% and 75.89%, respectively. After internal verification by the Bootstrap method, the fitting degree of the pathological features model was found to be better than that of the CT signs model.

Conclusion

Mitotic count, pathological risk grading, and tumor size are independent risk factors correlating with high Ki-67 index. These results indicate that the Ki-67 index reflects tumor malignancy and can predict recurrence and metastasis of GIST.

CT features of gastrointestinal spindle cell, epithelial, and round cell tumors in 41 dogs

There is sparse published information on computed tomographic (CT) characteristics of canine gastrointestinal tumors. The purposes of this multi-center, retrospective, descriptive study were to describe the CT features of histologically-confirmed canine gastrointestinal spindle cell, epithelial, and round cell tumors and, when available, describe the corresponding ultrasound findings. The inclusion criteria were as follows: availability of pre-and post-contrast CT study, and a histopathological diagnosis of the lesions. Recorded parameters were tumor size, location, gastrointestinal wall layers involvement, lesion's growth and enhancement patterns, tumor margination, presence of stenosis, mineralization, ulcerations, lymphadenopathy, or other lesions in the abdomen/thorax. When available, ultrasound images were evaluated. Forty-one dogs met the inclusion criteria and had the following histological diagnoses: 21/41 (51%) spindle cells (7 leiomyomas, 14 leiomyosarcomas/gastrointestinal stromal tumors (GISTs)), 13/41 (32%) epithelial (adenocarcinoma), and 7/41 (17%) round cell (lymphoma) tumors. The growth pattern was concentric, eccentric, and mixed in epithelial, spindle cell, and round cell tumors, respectively. Spindle cell tumors had the largest main volume and involved the outer gastrointestinal layer with an unaffected inner layer. Leiomyosarcomas/GISTs showed irregular margins compared to leiomyomas. Only lymphomas showed multifocal gastrointestinal involvement. Nine carcinomas and six spindle cell tumors caused partial stenosis with secondary sub-obstruction. Mineralizations were more frequent in spindle cell tumors (10/21) and absent in lymphomas. Lymphadenomegaly was widespread in lymphomas, regional in leiomyosarcomas-GISTs and adenocarcinomas, and absent in leiomyomas. The reported CT features may be useful in prioritizing the differential diagnosis between spindle cell, epithelial, and round cell tumors, similar to those reported on ultrasound.

Prognostic CT features in patients with untreated thymic epithelial tumors

To determine the prognostic CT features in patients with untreated thymic epithelial tumors (TETs). Clinical data and CT imaging features of 194 patients with pathologically confirmed TETs were retrospectively reviewed. The subjects included 113 male and 81 female patients between 15 and 78 years of age, with a mean age of 53.8 years. Clinical outcomes were categorized according to whether relapse, metastasis or death occurred within 3 years after the first diagnosis. Associations between clinical outcomes and CT imaging features were determined using univariate and multivariate logistic regression analyses, while the survival status was analyzed by Cox regression. In this study, we analyzed 110 thymic carcinomas, 52 high-risk thymomas and 32 low-risk thymomas. Percentages of poor outcome and patient death in thymic carcinomas were much higher than those in patients with high-risk and low-risk thymomas. In the thymic carcinomas groups, 46 patients (41.8%) experienced tumor progression, local relapse or metastasis and were categorized as having poor outcomes; vessel invasion and pericardial mass were confirmed to be independent predictors by logistic regression analysis (p < 0.01). In the high-risk thymoma group, 11 patients (21.2%) were categorized as having poor outcomes, and the CT feature pericardial mass was confirmed to be an independent predictor (p < 0.01). In survival analysis, Cox regression showed that CT features of lung invasion, great vessel invasion, lung metastasis and distant organ metastasis were independent predictors for worse survival in the thymic carcinoma group (p < 0.01), while lung invasion and pericardial mass were independent predictors for worse survival in high-risk thymoma group. No CT features were related to poor outcome and worse survival in the low-risk thymoma group. Patients with thymic carcinoma had poorer prognosis and worse survival than those with high-risk or low-risk thymoma. CT can serve as an important tool for predicting the prognosis and survival of patients with TETs. In this cohort, CT features of vessel invasion and pericardial mass were related to poorer outcomes in those with thymic carcinoma and pericardial mass in those with high-risk thymoma. Features including lung invasion, great vessel invasion, lung metastasis and distant organ metastasis indicate worse survival in thymic carcinoma, whereas lung invasion and pericardial mass indicate worse survival in high-risk thymoma.

Prognostic value of preoperative CT features for disease-free survival in patients with primary gastric gastrointestinal stromal tumors after resection

PURPOSE

Tumor size is an important prognostic factor without consideration of the necrotic and cystic components within tumor for patients with gastrointestinal stromal tumors (GISTs). We aimed to extract the enhancing viable component from the tumor using computed tomography (CT) post-processing software and evaluate the value of preoperative CT features for predicting the disease-free survival (DFS) after curative resection for patients with primary gastric GISTs.

METHODS

132 Patients with primary gastric GISTs who underwent preoperative contrast-enhanced CT and curative resection were retrospectively analyzed. We used a certain CT attenuation of 30 HU to extract the enhancing tissue component from the tumor. Enhancing tissue volume and other CT features were assessed on venous-phase images. We evaluated the value of preoperative CT features for predicting the DFS after surgery. Univariate and multivariate Cox regression analyses were performed to find the independent risk factor for predicting the DFS.

RESULTS

Of the 132 patients, 68 were males and 64 were females, with a mean age of 61 years. The median follow-up duration was 60 months, and 28 patients experienced disease recurrence and distant metastasis during the follow-up period. Serosal invasion (p < 0.001; HR = 5.277) and enhancing tissue volume (p = 0.005; HR = 1.447) were the independent risk factors for predicting the DFS after curative resection for patients with primary gastric GISTs.

CONCLUSION

Preoperative contrast-enhanced CT could be useful for predicting the DFS after the surgery of gastric GISTs, and serosal invasion and enhancing tissue volume were the independent risk factors.

Radiomics signatures based on contrast-enhanced CT for preoperative prediction of the Ki-67 proliferation state in gastrointestinal stromal tumors

PURPOSE

This study aimed to evaluate the Ki-67 proliferation state in patients with gastrointestinal stromal tumors (GISTs) using radiomics prediction signatures based on contrast-enhanced computed tomography (CE-CT).

MATERIALS AND METHODS

This single-center, retrospective study involved 103 patients (48 men and 55 women, mean age 61.1 ± 10.6 years) who had pathologically confirmed GISTs after curative resection, including 63 with low Ki-67 proliferation level (Ki-67 labeling index ≤ 6%) and 40 with high Ki-67 proliferation level (Ki-67 labeling index > 6%). Radiomics features of the delineated lesions were preoperatively extracted from three-phase CE-CT images, including the arterial, venous, and delayed phases. The most relevant features were selected to construct the radiomics signatures using a logistic regression algorithm. Significant demographic characteristics and semantic features on CT were selected to develop a nomogram along with the optimal radiomics feature. We calculated the sensitivity, specificity, accuracy, F1 score, and area under the receiver operating characteristic (ROC) curve to evaluate the predictive performance of radiomics signatures.

RESULTS

Ten quantitative radiomics features (two first-order and eight texture features) were selected to construct radiomics signatures. The radiomics signature based on the three-phase CE-CT images showed better predictive performance than that based on the single-phase CE-CT images, with an area under the curve (AUC) of 0.83 (95% CI 0.73-0.92) and F1 score of 82% in the training dataset and an AUC of 0.80 (95% CI 0.63-0.95) and F1 score of 75% in the testing dataset. The nomogram showed good calibration.

CONCLUSION

Radiomics signatures using CE-CT images are generalizable and could be used in clinical practice to determine the proliferation state of Ki-67 in GISTs.

Simple Scoring Model Based on Enhanced CT in Preoperative Prediction of Biological Risk of Gastrointestinal Stromal Tumor

Objective: To construct a simple scoring model for predicting the biological risk of gastrointestinal stromal tumors based on enhanced computed tomography (CT) features. Methods: The clinicopathological and imaging data of 149 patients with primary gastrointestinal stromal tumor were retrospectively analyzed in our hospital. According to the risk classification, the patients were divided into low-risk group and high-risk group. The features of enhanced CT were observed and recorded. Univariate and multivariate logistic regression models were used to determine the predictors of high-risk biological behaviors of gastrointestinal stromal tumor, and then a simple scoring model was constructed according to the regression coefficients of each predictor. The receiver operating characteristic curve was used to evaluate the predictive ability of the model. Results: There was no significant difference between the risk classification of gastrointestinal stromal tumor with gender and age (P = .168, .320), while significant difference was found between the tumor size and location (P < .001). Univariate and multivariate logistic regression analyses showed that tumor size, enlarged vessels feeding or draining the mass, peritumoral lymph node enlargement, and venous phase contrast enhancement rate were independent predictors of the biological risk of gastrointestinal stromal tumor (P < .05). The area under the curve value of tumor size, enlarged vessels feeding or draining the mass, peritumoral lymph node enlargement, and venous phase contrast enhancement rate as the high-risk predictor of gastrointestinal stromal tumor were 0.955, 0.729, 0.680, and 0.807, respectively. Receiver operating characteristic curve results showed that the area under the curve of the scoring model constructed based on enhanced CT features was 0.941 (95% confidence interval: 0.891-0.973). When the total score was >1, the sensitivity of the scoring model in diagnosing gastrointestinal stromal tumor was 85.58%, the specificity was 88.89%, the positive predictive value was 88.51%, the negative predictive value was 86.04%, and the accuracy was 86.18%. The results of DeLong test showed that the area under the curve of the scoring model was better than that of the receiver operating characteristic curve of tumor size, enlarged vessels feeding or draining the mass, peritumoral lymph node enlargement, venous phase contrast enhancement rate, and other indicators alone in predicting the high risk of gastrointestinal stromal tumor, and the differences were statistically significant (Z = 26.510, P < .001; Z = 3.992, P < .001; Z = 6.353, P < .001; Z = 4.052, P = .013). Conclusion: The simple scoring model based on enhanced CT features is a simple and practical clinical prediction model, which is helpful to make preoperative individualized treatment plan and improve the prognosis of gastrointestinal stromal tumor patients.

Deep learning and radiomics to predict the mitotic index of gastrointestinal stromal tumors based on multiparametric MRI

Introduction

Preoperative evaluation of the mitotic index (MI) of gastrointestinal stromal tumors (GISTs) represents the basis of individualized treatment of patients. However, the accuracy of conventional preoperative imaging methods is limited. The aim of this study was to develop a predictive model based on multiparametric MRI for preoperative MI prediction.

Methods

A total of 112 patients who were pathologically diagnosed with GIST were enrolled in this study. The dataset was subdivided into the development (n = 81) and test (n = 31) sets based on the time of diagnosis. With the use of T2-weighted imaging (T2WI) and apparent diffusion coefficient (ADC) map, a convolutional neural network (CNN)-based classifier was developed for MI prediction, which used a hybrid approach based on 2D tumor images and radiomics features from 3D tumor shape. The trained model was tested on an internal test set. Then, the hybrid model was comprehensively tested and compared with the conventional ResNet, shape radiomics classifier, and age plus diameter classifier.

Results

The hybrid model showed good MI prediction ability at the image level; the area under the receiver operating characteristic curve (AUROC), area under the precision-recall curve (AUPRC), and accuracy in the test set were 0.947 (95% confidence interval [CI]: 0.927-0.968), 0.964 (95% CI: 0.930-0.978), and 90.8 (95% CI: 88.0-93.0), respectively. With the average probabilities from multiple samples per patient, good performance was also achieved at the patient level, with AUROC, AUPRC, and accuracy of 0.930 (95% CI: 0.828-1.000), 0.941 (95% CI: 0.792-1.000), and 93.6% (95% CI: 79.3-98.2) in the test set, respectively.

Discussion

The deep learning-based hybrid model demonstrated the potential to be a good tool for the operative and non-invasive prediction of MI in GIST patients.

The Utility of Conventional CT, CT Perfusion and Quantitative Diffusion-Weighted Imaging in Predicting the Risk Level of Gastrointestinal Stromal Tumors of the Stomach: A Prospective Comparison of Classical CT Features, CT Perfusion Values, Apparent Diffusion Coefficient and Intravoxel Incoherent Motion-Derived Parameters

Background: The role of advanced functional imaging techniques in prediction of pathological risk categories of gastrointestinal stromal tumors (GIST) is still unknown. The purpose of this study was to evaluate classical CT features, CT-perfusion and magnetic-resonance-diffusion-weighted-imaging (MR-DWI)-related parameters in predicting the metastatic risk of gastric GIST. Patients and methods: Sixty-two patients with histologically proven GIST who underwent CT perfusion and MR-DWI using multiple b-values were prospectively included. Morphological CT characteristics and CT-perfusion parameters of tumor were comparatively analyzed in the high-risk (HR) and low-risk (LR) GIST groups. Apparent diffusion coefficient (ADC) and intravoxel-incoherent-motion (IVIM)-related parameters were also analyzed in 45 and 34 patients, respectively. Results: Binary logistic regression analysis revealed that greater tumor diameter (p < 0.001), cystic structure (p < 0.001), irregular margins (p = 0.007), irregular shape (p < 0.001), disrupted mucosa (p < 0.001) and visible EFDV (p < 0.001), as well as less ADC value (p = 0.001) and shorter time-to-peak (p = 0.006), were significant predictors of HR GIST. Multivariate analysis extracted irregular shape (p = 0.006) and enlarged feeding or draining vessels (EFDV) (p = 0.017) as independent predictors of HR GIST (area under curve (AUC) of predicting model 0.869). Conclusion: Although certain classical CT imaging features remain most valuable, some functional imaging parameters may add the diagnostic value in preoperative prediction of HR gastric GIST.

Gastrointestinal stromal tumors: a comprehensive radiological review

Gastrointestinal stromal tumors (GISTs) originating from the interstitial cells of Cajal in the muscularis propria are the most common mesenchymal tumor of the gastrointestinal tract. Multiple modalities, including computed tomography (CT), magnetic resonance imaging (MRI), fluorodeoxyglucose positron emission tomography, ultrasonography, digital subtraction angiography, and endoscopy, have been performed to evaluate GISTs. CT is most frequently used for diagnosis, staging, surveillance, and response monitoring during molecularly targeted therapy in clinical practice. The diagnosis of GISTs is sometimes challenging because of the diverse imaging findings, such as anatomical location (esophagus, stomach, duodenum, small bowel, colorectum, appendix, and peritoneum), growth pattern, and enhancement pattern as well as the presence of necrosis, calcification, ulceration, early venous return, and metastasis. Imaging findings of GISTs treated with antineoplastic agents are quite different from those of other neoplasms (e.g. adenocarcinomas) because only subtle changes in size are seen even in responsive lesions. Furthermore, the recurrence pattern of GISTs is different from that of other neoplasms. This review discusses the advantages and disadvantages of each imaging modality, describes imaging findings obtained before and after treatment, presents a few cases of complicated GISTs, and discusses recent investigations performed using CT and MRI to predict histological risk grade, gene mutations, and patient outcomes.

Radiomics Nomogram Based on Contrast-enhanced CT to Predict the Malignant Potential of Gastrointestinal Stromal Tumor: A Two-center Study

RATIONALE AND OBJECTIVES

Contrast-enhanced computed tomography (CE-CT) was used to establish radiomics nomogram to evaluate the malignant potential of gastrointestinal stromal tumors (GISTs).

MATERIALS AND METHODS

A total of 500 GIST patients were enrolled in this study and divided into training cohort (n = 346, our center) and validation cohort (n = 154, another center). Minimum redundancy maximum relevance (mRMR) and least absolute shrinkage and selection operator (LASSO) algorithms were used to select the feature subset with the best discriminant features from the three phases image, and five classifiers were used to establish four radiomics signatures. Preoperative radiomics nomogram was constructed by adding the clinical features determined by multivariate logistic regression analysis. The performance of radiomics signatures and nomogram were evaluated by area under the curve (AUC) of the receiver operating characteristic (ROC). The calibration of nomogram was appraised by calibration curve.

RESULTS

A total of 13 radiomic features were extracted from tri-phase combined CE-CT images. Tri-phase combined CE-CT features + Support Vector Machine (SVM) was the best combination at predicting the malignant potential of GIST, with an AUC of 0.895 (95% CI 0.858-0.931) in the training cohort and 0.847 (95% CI 0.778-0.917) in the validation cohort. The nomogram also had good calibration. In the training cohort and the validation cohort, preoperative radiomics nomogram reached AUCs of 0.927 and 0.905, respectively, which were higher than clinical.

CONCLUSION

The radiomics nomogram had a good predictive effect and generalization on the malignant potential of GIST, which could effectively help guide preoperative clinical decision.

Can lymphovascular invasion be predicted by contrast-enhanced CT imaging features in patients with esophageal squamous cell carcinoma? A preliminary retrospective study

Background

To investigate the value of contrast-enhanced CT (CECT)-derived imaging features in predicting lymphovascular invasion (LVI) status in esophageal squamous cell carcinoma (ESCC) patients.

Methods

One hundred and ninety-seven patients with postoperative pathologically confirmed esophageal squamous cell carcinoma treated in our hospital between January 2017 and January 2019 were enrolled in our study, including fifty-nine patients with LVI and one hundred and thirty-eight patients without LVI. The CECT-derived imaging features of all patients were analyzed. The CECT-derived imaging features were divided into quantitative features and qualitative features. The quantitative features consisted of the CT attenuation value of the tumor (CTV_Tumor), the CT attenuation value of the normal esophageal wall (CTV_Normal), the CT attenuation value ratio of the tumor-to-normal esophageal wall (TNR), the CT attenuation value difference between the tumor and normal esophageal wall (ΔTN), the maximum thickness of the tumor measured by CECT (Thickness), the maximum length of the tumor measured by CECT (Length), and the gross tumor volume measured by CECT (GTV). The qualitative features consisted of an enhancement pattern, tumor margin, enlarged blood supply or drainage vessels to the tumor (EVFDT), and tumor necrosis. For the clinicopathological characteristics and CECT-derived imaging feature analysis, the chi-squared test was used for categorical variables, the Mann–Whitney U test was used for continuous variables with a nonnormal distribution, and the independent sample t-test was used for the continuous variables with a normal distribution. The trend test was used for ordinal variables. The association between LVI status and CECT-derived imaging features was analyzed by univariable logistic analysis, followed by multivariable logistic regression and receiver operating characteristic (ROC) curve analysis.

Results

The CTV_Tumor, TNR, ΔTN, Thickness, Length, and GTV in the group with LVI were higher than those in the group without LVI (P < 0.05). A higher proportion of patients with heterogeneous enhancement pattern, irregular tumor margin, EVFDT, and tumor necrosis were present in the group with LVI (P < 0.05). As revealed by the univariable logistic analysis, the CECT-derived imaging features, including CTV_Tumor, TNR, ΔTN and enhancement pattern, Thickness, Length, GTV, tumor margin, EVFDT, and tumor necrosis were associated with LVI status (P < 0.05). Only the TNR (OR 8.655; 95% CI 2.125–37.776), Thickness (OR 6.531; 95% CI 2.410–20.608), and tumor margin (OR 4.384; 95% CI 2.004–9.717) were independent risk factors for LVI in the multivariable logistic regression analysis. The ROC curve analysis incorporating the above three CECT-derived imaging features showed that the area under the curve obtained by the multivariable logistic regression model was 0.820 (95% CI 0.754–0.885).

Conclusion

The CECT-derived imaging features, including TNR, Thickness, tumor margin, and their combination, can be used as predictors of LVI status for patients with ESCC.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12880-022-00804-7.

Preoperative prediction of malignant potential of 2-5 cm gastric gastrointestinal stromal tumors by computerized tomography-based radiomics

BACKGROUND

The use of endoscopic surgery for treating gastrointestinal stromal tumors (GISTs) between 2 and 5 cm remains controversial considering the potential risk of metastasis and recurrence. Also, surgeons are facing great difficulties and challenges in assessing the malignant potential of 2-5 cm gastric GISTs.

AIM

To develop and evaluate computerized tomography (CT)-based radiomics for predicting the malignant potential of primary 2-5 cm gastric GISTs.

METHODS

A total of 103 patients with pathologically confirmed gastric GISTs between 2 and 5 cm were enrolled. The malignant potential was categorized into low grade and high grade according to postoperative pathology results. Preoperative CT images were reviewed by two radiologists. A radiological model was constructed by CT findings and clinical characteristics using logistic regression. Radiomic features were extracted from preoperative contrast-enhanced CT images in the arterial phase. The XGboost method was used to construct a radiomics model for the prediction of malignant potential. Nomogram was established by combing the radiomics score with CT findings. All of the models were developed in a training group (n = 69) and evaluated in a test group (n = 34).

RESULTS

The area under the curve (AUC) value of the radiological, radiomics, and nomogram models was 0.753 (95% confidence interval [CI]: 0.597-0.909), 0.919 (95%CI: 0.828-1.000), and 0.916 (95%CI: 0.801-1.000) in the training group vs 0.642 (95%CI: 0.379-0.870), 0.881 (95%CI: 0.772-0.990), and 0.894 (95%CI: 0.773-1.000) in the test group, respectively. The AUC of the nomogram model was significantly larger than that of the radiological model in both the training group (Z = 2.795, P = 0.0052) and test group (Z = 2.785, P = 0.0054). The decision curve of analysis showed that the nomogram model produced increased benefit across the entire risk threshold range.

CONCLUSION

Radiomics may be an effective tool to predict the malignant potential of 2-5 cm gastric GISTs and assist preoperative clinical decision making.

Comparison of Computed Tomography Features of Gastric and Small Bowel Gastrointestinal Stromal Tumors With Different Risk Grades

OBJECTIVE

This study aimed to compare the computed tomography (CT) features of gastric and small bowel gastrointestinal stromal tumors (GISTs) and further identify the predictors for risk stratification of them, respectively.

METHODS

According to the modified National Institutes of Health criteria, patients were classified into low-malignant potential group and high-malignant potential group. Two experienced radiologists reviewed the CT features including the difference of CT values between arterial phase and portal venous phase (PVPMAP) by consensus. The CT features of gastric and small bowel GISTs were compared, and the association of CT features with risk grades was analyzed, respectively. Determinant CT features were used to construct corresponding models.

RESULTS

Univariate analysis showed that small bowel GISTs tended to present with irregular contour, mixed growth pattern, ill-defined margin, severe necrosis, ulceration, tumor vessels, heterogeneous enhancement, larger size, and marked enhancement compared with gastric GISTs. According to multivariate analysis, tumor size (P < 0.001; odds ratio [OR], 3.279), necrosis (P = 0.008; OR, 2.104) and PVPMAP (P = 0.045; OR, 0.958) were the independent influencing factors for risk stratification of gastric GISTs. In terms of small bowel GISTs, the independent predictors were tumor size (P < 0.001; OR, 3.797) and ulceration (P = 0.031; OR, 4.027). Receiver operating characteristic curve indicated that the CT models for risk stratification of gastric and small bowel GISTs both achieved the best predictive performance.

CONCLUSIONS

Computed tomography features of gastric and small bowel GISTs are different. Furthermore, the qualitative and quantitative CT features of GISTs may be favorable for preoperative risk stratification.

Differential Diagnosis and Molecular Stratification of Gastrointestinal Stromal Tumors on CT Images Using a Radiomics Approach

Treatment planning of gastrointestinal stromal tumors (GISTs) includes distinguishing GISTs from other intra-abdominal tumors and GISTs’ molecular analysis. The aim of this study was to evaluate radiomics for distinguishing GISTs from other intra-abdominal tumors, and in GISTs, predict the c-KIT, PDGFRA, BRAF mutational status, and mitotic index (MI). Patients diagnosed at the Erasmus MC between 2004 and 2017, with GIST or non-GIST intra-abdominal tumors and a contrast-enhanced venous-phase CT, were retrospectively included. Tumors were segmented, from which 564 image features were extracted. Prediction models were constructed using a combination of machine learning approaches. The evaluation was performed in a 100 × random-split cross-validation. Model performance was compared to that of three radiologists. One hundred twenty-five GISTs and 122 non-GISTs were included. The GIST vs. non-GIST radiomics model had a mean area under the curve (AUC) of 0.77. Three radiologists had an AUC of 0.69, 0.76, and 0.84, respectively. The radiomics model had an AUC of 0.52 for c-KIT, 0.56 for c-KIT exon 11, and 0.52 for the MI. The numbers of PDGFRA, BRAF, and other c-KIT mutations were too low for analysis. Our radiomics model was able to distinguish GISTs from non-GISTs with a performance similar to three radiologists, but less observer dependent. Therefore, it may aid in the early diagnosis of GIST, facilitating rapid referral to specialized treatment centers. As the model was not able to predict any genetic or molecular features, it cannot aid in treatment planning yet.

Building Radiomics Models Based on Triple-Phase CT Images Combining Clinical Features for Discriminating the Risk Rating in Gastrointestinal Stromal Tumors

We aimed to build radiomics models based on triple-phase CT images combining clinical features to predict the risk rating of gastrointestinal stromal tumors (GISTs). A total of 231 patients with pathologically diagnosed GISTs from July 2012 to July 2020 were categorized into a training data set (82 patients with high risk, 80 patients with low risk) and a validation data set (35 patients with high risk, 34 patients with low risk) with a ratio of 7:3. Four diagnostic models were constructed by assessing 20 clinical characteristics and 18 radiomic features that were extracted from a lesion mask based on triple-phase CT images. The receiver operating characteristic (ROC) curves were applied to calculate the diagnostic performance of these models, and ROC curves of these models were compared using Delong test in different data sets. The results of ROC analyses showed that areas under ROC curves (AUC) of model 4 [Clinic + CT value of unenhanced (CTU) + CT value of arterial phase (CTA) + value of venous phase (CTV)], model 1 (Clinic + CTU), model 2 (Clinic + CTA), and model 3 (Clinic + CTV) were 0.925, 0.894, 0.909, and 0.914 in the training set and 0.897, 0.866, 0,892, and 0.892 in the validation set, respectively. Model 4, model 1, model 2, and model 3 yielded an accuracy of 88.3%, 85.8%, 86.4%, and 84.6%, a sensitivity of 85.4%, 84.2%, 76.8%, and 78.0%, and a specificity of 91.2%, 87.5%, 96.2%, and 91.2% in the training set and an accuracy of 88.4%, 84.1%, 82.6%, and 82.6%, a sensitivity of 88.6%, 77.1%, 74.3%, and 85.7%, and a specificity of 88.2%, 91.2%, 91.2%, and 79.4% in the validation set, respectively. There was a significant difference between model 4 and model 1 in discriminating the risk rating in gastrointestinal stromal tumors in the training data set (Delong test, p < 0.05). The radiomic models based on clinical features and triple-phase CT images manifested excellent accuracy for the discrimination of risk rating of GISTs.

Gastrointestinal stromal tumors: associations between contrast-enhanced CT images and KIT exon 11 gene mutation

Background

Mutation screening for gastrointestinal stromal tumor (GIST) is crucial and the c kit gene (KIT) exon 11 mutation is the most common type. This study aimed to explore the associations between GIST with KIT exon 11 mutation and contrast-enhanced computed tomography (CT) images.

Methods

Pathologically proven GISTs with definitive genotype testing results in our hospital were retrospectively included. Abdominal contrast-enhanced CT images were analyzed. Conventional CT image features and radiomic features were recorded and extracted to build the following models: model _[CT], model _{[radiomic + clinic]} and model _{[CT + radiomic + clinic]}. The diagnostic performances of GISTs with KIT exon 11 mutation and KIT exon 11 deletion involving codons 557–558 were evaluated.

Results

In total, 327 GISTs (255 with KIT exon 11 mutation, and 73 with KIT exon 11 mutation deletion involving codons 557–558) were included. Significant CT features were found for GISTs with KIT exon 11 mutation. The area under curves (AUCs) of the models for KIT exon 11 mutation were 0.7158, 0.7530, and 0.8375 in the training cohort, and 0.6777, 0.7349, and 0.8105 in validation cohort, respectively. The AUCs of the models for KIT exon 11 mutation deletion involving codons 557–558 were 0.7155, 8621, and 0.8691 in the training cohort, and 0.7099, 0.8355, and 0.8488 in the validation cohort, respectively. The model _{[CT + radiomic + clinic]} demonstrated the highest AUCs for prediction of KIT exon 11 mutation and those with deletion involving codons 557–558 (P<0.05), respectively. The model _{[radiomic + clinic]} showed higher diagnostic performance than model _[CT] significantly.

Conclusions

Our results demonstrated the associations between GIST with KIT exon 11 mutation and contrast-enhanced CT images. We found combing conventional image analysis and texture analysis is a useful tool to distinguish GIST with KIT exon 11 mutation. CT radiogenomics exhibited good application potential in predict the KIT exon 11 mutation of GIST.

Preoperative CT-Based Deep Learning Model for Predicting Risk Stratification in Patients With Gastrointestinal Stromal Tumors

Objective

To develop and evaluate a deep learning model (DLM) for predicting the risk stratification of gastrointestinal stromal tumors (GISTs).

Methods

Preoperative contrast-enhanced CT images of 733 patients with GISTs were retrospectively obtained from two centers between January 2011 and June 2020. The datasets were split into training (n = 241), testing (n = 104), and external validation cohorts (n = 388). A DLM for predicting the risk stratification of GISTs was developed using a convolutional neural network and evaluated in the testing and external validation cohorts. The performance of the DLM was compared with that of radiomics model by using the area under the receiver operating characteristic curves (AUROCs) and the Obuchowski index. The attention area of the DLM was visualized as a heatmap by gradient-weighted class activation mapping.

Results

In the testing cohort, the DLM had AUROCs of 0.90 (95% confidence interval [CI]: 0.84, 0.96), 0.80 (95% CI: 0.72, 0.88), and 0.89 (95% CI: 0.83, 0.95) for low-malignant, intermediate-malignant, and high-malignant GISTs, respectively. In the external validation cohort, the AUROCs of the DLM were 0.87 (95% CI: 0.83, 0.91), 0.64 (95% CI: 0.60, 0.68), and 0.85 (95% CI: 0.81, 0.89) for low-malignant, intermediate-malignant, and high-malignant GISTs, respectively. The DLM (Obuchowski index: training, 0.84; external validation, 0.79) outperformed the radiomics model (Obuchowski index: training, 0.77; external validation, 0.77) for predicting risk stratification of GISTs. The relevant subregions were successfully highlighted with attention heatmap on the CT images for further clinical review.

Conclusion

The DLM showed good performance for predicting the risk stratification of GISTs using CT images and achieved better performance than that of radiomics model.

Preoperative prediction of gastrointestinal stromal tumors with high Ki-67 proliferation index based on CT features

BACKGROUND

To determine whether preoperative computed tomography (CT) features can be used for the prediction of gastrointestinal stromal tumors (GISTs) with a high Ki-67 proliferation index (Ki-67 PI).

METHODS

A total of 198 patients with surgically and pathologically proven GISTs were retrospectively included. All GISTs were divided into a low Ki-67 PI group (<10%) and a high Ki-67 PI group (≥10%). All imaging features were blindly interpreted by two radiologists. Receiver operating characteristic (ROC) curve analyses were conducted to evaluate the predictive performance of the imaging features.

RESULTS

Imaging features were found to be significantly different between the low and the high Ki-67 PI groups (P<0.05). Wall thickness of necrosis showed the highest predictive ability, with an area under the curve (AUC) of 0.838 [95% confidence interval (CI): 0.627-0.957], followed by necrosis, necrosis degree, hyperenhancement of the overlying mucosa (HYOM), and long diameter (LD) (AUC >0.7, P<0.05). HYOM was the strongest predictive feature for the high Ki-67 PI GISTs group, with an odds ratio (OR) value of 30.037 (95% CI: 5.707-158.106).

CONCLUSIONS

Imaging features, including the presence of necrosis, high necrosis degree, thick wall of necrosis, and HYOM were significant predictive indicators for the high Ki-67 PI GISTs group.

A CT-based nomogram for predicting the malignant potential of primary gastric gastrointestinal stromal tumors preoperatively

PURPOSE

To develop and validate a computerized tomography (CT)-based nomogram for predicting the malignant potential of primary gastric gastrointestinal stromal tumors (GISTs).

METHODS

The primary and validation cohorts consisted of 167 and 39 patients (single center, different time periods) with histologically confirmed primary gastric GISTs. Clinical data and preoperative CT images were reviewed. The association of CT characteristics with malignant potential was analyzed using univariate and stepwise logistic regression analyses. A nomogram based on significant CT findings was developed for predicting malignant potential. The predictive accuracy of the nomogram was determined by the concordance index (C-index) and calibration curves. External validation was performed with the validation cohort.

RESULTS

CT imaging features including tumor size, tumor location, tumor necrosis, growth pattern, ulceration, enlarged vessels feeding or draining the mass (EVFDM), tumor contour, mesenteric fat infiltration, and direct organ invasion showed significant differences between the low- and high-grade malignant potential groups in univariate analysis (P < 0.05). Only tumor size (> 5 cm vs ≤ 5 cm), location (cardiac/pericardial region vs other), EVFDM, and mesenteric fat infiltration (present vs absent) were significantly associated with high malignant potential in multivariate logistic regression analysis. Incorporating these four independent factors into the nomogram model achieved good C-indexes of 0.946 (95% confidence interval [CI] 0.899-0.975) and 0.952 (95% CI 0.913-0.977) in the primary and validation cohorts, respectively. The cutoff point was 0.33, with sensitivity, specificity, and diagnostic accuracy of 0.865, 0.915, and 0.780, respectively.

DISCUSSION

Primary gastric GISTs originating in the cardiac/pericardial region appear to be associated with higher malignant potential. The nomogram consisting of CT features, including size, location, EVFDM, and mesenteric fat infiltration, could be used to accurately predict the high malignant potential of primary gastric GISTs.

MRI-Based Radiomics Models for Predicting Risk Classification of Gastrointestinal Stromal Tumors

Background

We conduct a study in developing and validating four MRI-based radiomics models to preoperatively predict the risk classification of gastrointestinal stromal tumors (GISTs).

Methods

Forty-one patients (low-risk = 17, intermediate-risk = 13, high-risk = 11) underwent MRI before surgery between September 2013 and March 2019 in this retrospective study. The Kruskal–Wallis test with Bonferonni correction and variance threshold was used to select appropriate features, and the Random Forest model (three classification model) was used to select features among the high-risk, intermediate-risk, and low-risk of GISTs. The predictive performance of the models built by the Random Forest was estimated by a 5-fold cross validation (5FCV). Their performance was estimated using the receiver operating characteristic (ROC) curve, summarized as the area under the ROC curve (AUC). Area under the curve (AUC), accuracy, sensitivity, and specificity for risk classification were reported. Linear discriminant analysis (LDA) was used to assess the discriminative ability of these radiomics models.

Results

The high-risk, intermediate-risk, and low-risk of GISTs were well classified by radiomics models, the micro-average of ROC curves was 0.85, 0.81, 0.87 and 0.94 for T1WI, T2WI, ADC and combined three MR sequences. And ROC curves achieved excellent AUCs for T1WI (0.85, 0.75 and 0.82), T2WI (0.69, 0.78 and 0.78), ADC (0.85, 0.77 and 0.80) and combined three MR sequences (0.96, 0.92, 0.81) for the diagnosis of high-risk, intermediate-risk, and low-risk of GISTs, respectively. In addition, LDA demonstrated the different risk of GISTs were correctly classified by radiomics analysis (61.0% for T1WI, 70.7% for T2WI, 83.3% for ADC, and 78.9% for the combined three MR sequences).

Conclusions

Radiomics models based on a single sequence and combined three MR sequences can be a noninvasive method to evaluate the risk classification of GISTs, which may help the treatment of GISTs patients in the future.

Differentiation of gastric schwannomas from gastrointestinal stromal tumors by CT using machine learning

OBJECTIVE

To identify schwannomas from gastrointestinal stromal tumors (GISTs) by CT features using Logistic Regression (LR), Decision Trees (DT), Random Forest (RF), and Gradient Boosting Decision Tree (GBDT).

METHODS

This study enrolled 49 patients with schwannomas and 139 with GISTs proven by pathology. CT features with P < 0.1 derived from univariate analysis were inputted to four models. Five machine learning (ML) versions, multivariate analysis, and radiologists' subjective diagnostic performance were compared to evaluate diagnosis performance of all the traditional and advanced methods.

RESULTS

The CT features with P < 0.1 were as follows: (1) CT attenuation value of unenhancement phase (CTU), (2) portal venous enhancement (CTV), (3) degree of enhancement in the portal venous phase (DEPP), (4) CT attenuation value of portal venous phase minus arterial phase (CTV-CTA), (5) enhanced potentiality (EP), (6) location, (7) contour, (8) growth pattern, (9) necrosis, (10) surface ulceration, (11) enlarged lymph node (LN). LR (M1), RF, DT, and GBDT models contained all of the above 11 variables, while LR (M2) was developed using six most predictive variables derived from (M1). LR (M2) model with AUC of 0.967 in test dataset was thought to be optimal model in differentiating the two tumors. Location in gastric body, exophytic and mixed growth pattern, lack of necrosis and surface ulceration, enlarged lymph nodes, and larger EP were the most important CT features suggestive of schwannomas.

CONCLUSION

LR (M2) provided the optimal diagnostic potency among other ML versions, multivariate analysis, and radiologists' performance on differentiation of schwannomas from GISTs.

Computed-Tomography-Based Radiomics Model for Predicting the Malignant Potential of Gastrointestinal Stromal Tumors Preoperatively: A Multi-Classifier and Multicenter Study

Background: Our goal was to establish and verify a radiomics risk grading model for gastrointestinal stromal tumors (GISTs) and to identify the optimal algorithm for risk stratification. Methods: We conducted a retrospective analysis of 324 patients with GISTs, the presence of which was confirmed by surgical pathology. Patients were treated at three different hospitals. A training cohort of 180 patients was collected from the largest center, while an external validation cohort of 144 patients was collected from the other two centers. To extract radiomics features, regions of interest (ROIs) were outlined layer by layer along the edge of the tumor contour on CT images of the arterial and portal venous phases. The dimensionality of radiomic features was reduced, and the top 10 features with importance value above 5 were selected before modeling. The training cohort used three classifiers [logistic regression, support vector machine (SVM), and random forest] to establish three GIST risk stratification prediction models. The receiver operating characteristic curve (ROC) was used to compare model performance, which was validated by external data. Results: In the training cohort, the average area under the curve (AUC) was 0.84 ± 0.07 of the logistic regression, 0.88 ± 0.06 of the random forest, and 0.81 ± 0.08 of the SVM. In the external validation cohort, the AUC was 0.85 of the logistic regression, 0.90 of the random forest, and 0.80 of the SVM. The random forest model performed the best in both the training and the external validation cohorts and could be generalized. Conclusion: Based on CT radiomics, there are multiple machine-learning models that can predict the risk of GISTs. Among them, the random forest algorithm had the highest prediction efficiency and could be readily generalizable. Through external validation data, we assume that the random forest model may be used as an effective tool to guide preoperative clinical decision-making.

Assessment of morphological CT imaging features for the prediction of risk stratification, mutations, and prognosis of gastrointestinal stromal tumors

OBJECTIVES

To investigate the correlation between CT imaging features and risk stratification of gastrointestinal stromal tumors (GISTs), prediction of mutation status, and prognosis.

METHODS

This retrospective dual-institution study included patients with pathologically proven GISTs meeting the following criteria: (i) preoperative contrast-enhanced CT performed between 2008 and 2019; (ii) no treatments before imaging; (iii) available pathological analysis. Tumor risk stratification was determined according to the National Institutes of Health (NIH) 2008 criteria. Two readers evaluated the CT features, including enhancement patterns and tumor characteristics in a blinded fashion. The differences in distribution of CT features were assessed using univariate and multivariate analyses. Survival analyses were performed by using the Cox proportional hazard model, Kaplan-Meier method, and log-rank test.

RESULTS

The final population included 88 patients (59 men and 29 women, mean age 60.5 ± 11.1 years) with 45 high-risk and 43 low-to-intermediate-risk GISTs (median size 6.3 cm). At multivariate analysis, lesion size ≥ 5 cm (OR: 10.52, p = 0.009) and enlarged feeding vessels (OR: 12.08, p = 0.040) were independently associated with the high-risk GISTs. Hyperenhancement was significantly more frequent in PDGFRα-mutated/wild-type GISTs compared to GISTs with KIT mutations (59.3% vs 23.0%, p = 0.004). Ill-defined margins were associated with shorter progression-free survival (HR 9.66) at multivariate analysis, while ill-defined margins and hemorrhage remained independently associated with shorter overall survival (HR 44.41 and HR 30.22). Inter-reader agreement ranged from fair to almost perfect (k: 0.32-0.93).

CONCLUSIONS

Morphologic contrast-enhanced CT features are significantly different depending on the risk status or mutations and may help to predict prognosis.

KEY POINTS

• Lesions size ≥ 5 cm (OR: 10.52, p = 0.009) and enlarged feeding vessels (OR: 12.08, p = 0.040) are independent predictors of high-risk GISTs. • PDGFRα-mutated/wild-type GISTs demonstrate more frequently hyperenhancement compared to GISTs with KIT mutations (59.3% vs 23.0%, p = 0.004). • Ill-defined margins (hazard ratio 9.66) were associated with shorter progression-free survival at multivariate analysis, while ill-defined margins (hazard ratio 44.41) and intralesional hemorrhage (hazard ratio 30.22) were independently associated with shorter overall survival.

Gastrointestinal stromal tumors: relationship between preoperative CT features and pathologic risk stratification

OBJECTIVE

To investigate a relationship between contrast-enhanced computed tomography (CECT) features of gastrointestinal stromal tumors (GISTs) and risk of relapse according to Miettinen stratified risk classifications.

METHODS

After ethical committee approval, a retrospective analysis was conducted on the preoperative CECT of patients with pathologically proven GIST undergoing surgery between June 2009 and December 2019. Chi-square analysis was used to evaluate the correlation between Miettinen stratified risk categories and the following imaging features: tumor size and location, growth pattern, margins, type and degree of contrast enhancement, presence of calcifications, necrosis, signs of ulceration/fistulation, internal hemorrhagic foci, enlarged feeding or draining vessels (EFDV), ascites, peritoneal implants, lymphadenopathy, or metastasis.

RESULTS

A total of 54 patients (mean age 65 ± 11, 29 men) were included in the study with a total of 56 GISTs. Necrosis, ulceration/fistulation, hemorrhage, margins, enlarged vessels, type of contrast enhancement, and metastasis turned out to be associated with Miettinen risk categories (p < 0.005). Logistic regression analysis identified the presence of necrosis and EFDV as predictors of pathologic risk of relapse (overall accuracy of 89.3%).

CONCLUSION

Preoperative CECT may be helpful in predicting pathologic risk categories of GISTs, as determined by the Miettinen classification system.

Preoperative CT feature of incomplete overlying enhancing mucosa as a high-risk predictor in gastrointestinal stromal tumors of the stomach

OBJECTIVES

To determine whether the CT finding of overlying enhancing gastric mucosa (OEGM) can be used to predict risk stratifications by observing CT features of gastrointestinal stromal tumors (GISTs) of the stomach.

METHODS

Clinical characteristics and CT features within pathologically demonstrated GISTs were retrospectively reviewed. Risk stratifications were classified into non-high group and high-risk group according to the modified National Institutes of Health criteria. Univariate analysis and multivariate logistic regression analysis were performed in order to determine significant predictors for high-risk stratification. Receiver operating characteristic (ROC) curve analysis, subgroup analysis, and pathologic-radiologic correlation analysis were all executed.

RESULTS

A total of 147 patients were finally enrolled as test subjects. Within the univariate analysis, high-risk tumors tended to have a larger diameter, irregular shape, exophytic growth pattern, present necrosis, incomplete OEGM, tumor vessels, heterogeneous enhancement, and present rupture. According to ROC curve analysis, incomplete OEGM showed the largest area under curve values for diagnosing lesions (0.835; 95% CI, 0.766-0.904; p < 0.001). Multivariate analysis showed that the incomplete OEGM was the strongest independent predictor for high-risk stratification of gastric GISTs (OR = 21.944; 95% CI, 4.344-110.863; p < 0.001). Within the subgroup analysis, incomplete OEGM was more frequently associated with tumors size > 10 cm, irregular shape, exophytic growth pattern, high mitotic count, and disrupted mucosa on pathology.

CONCLUSIONS

The CT feature of incomplete OEGM is an independent predictive factor for high-risk stratification of gastric GISTs and strongly correlated with pathological mucosal changes.

KEY POINTS

• Preoperative CT features can be helpful in assessment of risk stratifications of gastric GISTs. • OEGM is an independent predictor for high-risk stratification of gastric GISTs. • Incomplete OEGM likely indicates high-risk stratification of gastric GISTs.

Development of a Malignancy Potential Binary Prediction Model Based on Deep Learning for the Mitotic Count of Local Primary Gastrointestinal Stromal Tumors

Objective

The mitotic count of gastrointestinal stromal tumors (GIST) is closely associated with the risk of planting and metastasis. The purpose of this study was to develop a predictive model for the mitotic index of local primary GIST, based on deep learning algorithm.

Materials and Methods

Abdominal contrast-enhanced CT images of 148 pathologically confirmed GIST cases were retrospectively collected for the development of a deep learning classification algorithm. The areas of GIST masses on the CT images were retrospectively labelled by an experienced radiologist. The postoperative pathological mitotic count was considered as the gold standard (high mitotic count, > 5/50 high-power fields [HPFs]; low mitotic count, ≤ 5/50 HPFs). A binary classification model was trained on the basis of the VGG16 convolutional neural network, using the CT images with the training set (n = 108), validation set (n = 20), and the test set (n = 20). The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated at both, the image level and the patient level. The receiver operating characteristic curves were generated on the basis of the model prediction results and the area under curves (AUCs) were calculated. The risk categories of the tumors were predicted according to the Armed Forces Institute of Pathology criteria.

Results

At the image level, the classification prediction results of the mitotic counts in the test cohort were as follows: sensitivity 85.7% (95% confidence interval [CI]: 0.834–0.877), specificity 67.5% (95% CI: 0.636–0.712), PPV 82.1% (95% CI: 0.797–0.843), NPV 73.0% (95% CI: 0.691–0.766), and AUC 0.771 (95% CI: 0.750–0.791). At the patient level, the classification prediction results in the test cohort were as follows: sensitivity 90.0% (95% CI: 0.541–0.995), specificity 70.0% (95% CI: 0.354–0.919), PPV 75.0% (95% CI: 0.428–0.933), NPV 87.5% (95% CI: 0.467–0.993), and AUC 0.800 (95% CI: 0.563–0.943).

Conclusion

We developed and preliminarily verified the GIST mitotic count binary prediction model, based on the VGG convolutional neural network. The model displayed a good predictive performance.

New advances in radiomics of gastrointestinal stromal tumors

Gastrointestinal stromal tumors (GISTs) are uncommon neoplasms of the gastrointestinal tract with peculiar clinical, genetic, and imaging characteristics. Preoperative knowledge of risk stratification and mutational status is crucial to guide the appropriate patients’ treatment. Predicting the clinical behavior and biological aggressiveness of GISTs based on conventional computed tomography (CT) and magnetic resonance imaging (MRI) evaluation is challenging, unless the lesions have already metastasized at the time of diagnosis. Radiomics is emerging as a promising tool for the quantification of lesion heterogeneity on radiological images, extracting additional data that cannot be assessed by visual analysis. Radiomics applications have been explored for the differential diagnosis of GISTs from other gastrointestinal neoplasms, risk stratification and prediction of prognosis after surgical resection, and evaluation of mutational status in GISTs. The published researches on GISTs radiomics have obtained excellent performance of derived radiomics models on CT and MRI. However, lack of standardization and differences in study methodology challenge the application of radiomics in clinical practice. The purpose of this review is to describe the new advances of radiomics applied to CT and MRI for the evaluation of gastrointestinal stromal tumors, discuss the potential clinical applications that may impact patients’ management, report limitations of current radiomics studies, and future directions.

Differentiation of gastric glomus tumor from small gastric stromal tumor by computed tomography

OBJECTIVE

This study was performed to investigate the value of computed tomography (CT) in the differentiation of gastric glomus tumors (GGTs) and small gastric stromal tumors (GSTs).

METHODS

Fifty-nine patients with pathologically confirmed GGTs (n = 11) and GSTs (n = 48) from 2006 to 2019 were retrospectively evaluated. All patients' preoperative CT imaging features were analyzed.

RESULTS

The following features were significantly different between GGTs and small GSTs: location in the antrum, endophytic growth, heterogeneous enhancement in the arterial phase, CT value in the arterial phase of ≥60.7 Hounsfield units (HU), CT value in the portal phase of ≥87.6 HU, degree of enhancement in the arterial phase of ≥29.9 HU, and degree of enhancement in the portal phase of ≥49.0 HU. A model including four randomly selected features among these seven criteria was built to differentiate GGTs from small GSTs with a sensitivity and specificity of 90.9% (10/11) and 100% (48/48), respectively.

CONCLUSION

We identified seven features that are useful for differentiating GGTs from small GSTs. A combination of four of these seven criteria may increase the diagnostic accuracy.

Comparison of malignancy-prediction efficiency between contrast and non-contract CT-based radiomics features in gastrointestinal stromal tumors: A multicenter study

This work seeks the development and validation of radiomics signatures from nonenhanced computed tomography (CT, NE-RS) to preoperatively predict the malignancy degree of gastrointestinal stromal tumors (GISTs) and the comparison of these signatures with those from contrast-enhanced CT. A dataset for 370 GIST patients was collected from four centers. This dataset was divided into cohorts for training, as well as internal and external validation. The minimum-redundancy maximum-relevance algorithm and the least absolute shrinkage and selection operator (LASSO) algorithm were used to filter unstable features. (a) NE-RS and radiomics signature from contrast-enhanced CT (CE-RS) were built and compared for the prediction of malignancy potential of GIST based on the area under the receiver operating characteristic curve (AUC). (b) The radiomics model was also developed with both the tumor size and NE-RS. The AUC values were comparable between NE-RS and CE-RS in the training (.965 vs .936; P = .251), internal validation (.967 vs .960; P = .801), and external validation (.941 vs .899; P = .173) cohorts in diagnosis of high malignancy potential of GISTs. We next focused on the NE-RS. With 0.185 selected as the cutoff of NE-RS for diagnosis of the malignancy potential of GISTs, accuracy, sensitivity, and specificity for diagnosis high-malignancy potential GIST was 90.0%, 88.2%, and 92.3%, respectively, in the training cohort. For the internal validation set, the corresponding metrics are 89.1%, 94.9%, and 80.0%, respectively. The corresponding metrics for the external cohort are 84.6%, 76.1%, and 91.0%, respectively. Compared with only NE-RS, the radiomics model increased the sensitivity in the diagnosis of GIST with high-malignancy potential by 5.9% (P = .025), 2.5% (P = .317), 10.5% (P = .008) for the training set, internal validation set, and external validation set, respectively. The NE-RS had comparable prediction efficiency in the diagnosis of high-risk GISTs to CE-RS. The NE-RS and radiomics model both had excellent accuracy in predicting malignancy potential of GISTs.

Difference of computed tomographic characteristic findings between gastric and intestinal gastrointestinal stromal tumors

PURPOSE

We aimed to compare the computed tomography (CT) imaging differences between gastric and intestinal gastrointestinal stromal tumors (GISTs).

MATERIALS AND METHODS

Thirty-eight patients with 38 gastric GISTs and 27 with 31 intestinal GISTs were enrolled. Tumors were classified as small (< 5 cm) or large (≥ 5 cm). Qualitative and quantitative CT imaging characteristics on non-contrast and contrast-enhanced CT were evaluated by two radiologists independently and statistically compared.

RESULTS

Early venous return and higher CT number of the draining vein in the arterial phase were more frequent in small-sized intestinal GISTs than in small-sized gastric GISTs (p < 0.001). Small-sized intestinal GISTs demonstrated a wash-out pattern, whereas small-sized gastric GISTs showed a plateau pattern. Contrast enhancement was higher in small-sized intestinal GISTs than in small-sized gastric GISTs (p < 0.001). CT number was inversely proportional to the diameter of intestinal GISTs in both arterial and venous phases but not to that of gastric GISTs.

CONCLUSION

Strong enhancement with wash-out pattern and early venous return are characteristic findings of small-sized intestinal GISTs. Radiologists should be aware that CT findings of GISTs have a wide spectrum and may differ according to size and site of origin.

A multivariate prediction model for high malignancy potential gastric GI stromal tumors before endoscopic resection

BACKGROUND AND AIMS

Endoscopic resection is becoming an option in the management of gastric GI stromal tumors (GISTs). Although no consensus has been reached, patients with high malignancy potential GISTs are generally considered to be surgical candidates. However, no systematic preoperative evaluation strategy has yet been developed. The current study was performed to develop a preoperative multivariate model to predict the malignant potential of gastric GISTs.

METHODS

This study consisted of 2 stages. First, a multivariate prediction model for gastric GISTs smaller than 5 cm was developed using a multivariate logistic regression analysis in a retrospective cohort. Next, the prediction model was validated further in a validation cohort of gastric GISTs.

RESULTS

In the developing stage, 275 patients were included. The multivariate analysis demonstrated that independent risk factors for high malignancy potential gastric GISTs smaller than 5 cm were tumor size ≥2 cm (according to cutoff value), an irregular tumor shape, and mucosal ulceration (P < .05). Based on accordant regression coefficients, 3 risk factors were weighted with point values: 1 point for mucosal ulceration, 2 points for an irregular tumor shape, and 3 points for tumor size ≥2 cm. In the validation stage, 186 patients were included. The area under the curve of the prediction model was .80 (95% confidence interval, .73-.85), which was significantly higher than that of tumor size alone (P = .034).

CONCLUSIONS

The independent risk factors for high malignancy potential gastric GISTs smaller than 5 cm were tumor size larger than 2 cm, an irregular tumor shape, and mucosal ulceration. These factors could be used to predict malignancy potential of gastric GISTs in a simple combination.

Computed tomography-based radiomics model for discriminating the risk stratification of gastrointestinal stromal tumors

PURPOSE

The pathological risk degree of gastrointestinal stromal tumors (GISTs) has become an issue of great concern. Computed tomography (CT) is beneficial for showing adjacent tissues in detail and determining metastasis or recurrence of GISTs, but its function is still limited. Radiomics has recently shown a great potential in aiding clinical decision-making. The purpose of our study is to develop and validate CT-based radiomics models for GIST risk stratification.

METHODS

Three hundred and sixty-six patients clinically suspected of primary GISTs from January 2013 to February 2018 were retrospectively enrolled, among which data from 140 patients were eventually analyzed after exclusion. Data from patient CT images were partitioned based on the National Institutes of Health Consensus Classification, including tumor segmentation, radiomics feature extraction and selection. A radiomics model was then proposed and validated.

RESULTS

The radiomics signature demonstrated discriminative performance for advanced and nonadvanced GISTs with an area under the curve (AUC) of 0.935 [95% confidence interval (CI) 0.870-1.000] and an accuracy of 90.2% for validation cohort. The radiomics signature demonstrated favorable performance for the risk stratification of GISTs with an AUC of 0.809 (95% CI 0.777-0.841) and an accuracy of 67.5% for the validation cohort. Radiomics analysis could capture features of the four risk categories of GISTs. Meanwhile, this CT-based radiomics signature showed good diagnostic accuracy to distinguish between nonadvanced and advanced GISTs, as well as the four risk stratifications of GISTs.

CONCLUSION

Our findings highlight the potential of a quantitative radiomics analysis as a complementary tool to achieve an accurate diagnosis for GISTs.

Predictive Features of Thymic Carcinoma and High-Risk Thymomas Using Random Forest Analysis

PURPOSE

To determine the predictive features of thymic carcinomas and high-risk thymomas using random forest algorithm.

METHODS

A total of 137 patients with pathologically confirmed high-risk thymomas and thymic carcinomas were enrolled in this study. Three clinical features and 20 computed tomography features were reviewed. The association between computed tomography features and pathological patterns was analyzed by univariate analysis and random forest. The predictive efficiency of the random forest algorithm was evaluated by receiver operating characteristic curve analysis.

RESULTS

There were 92 thymic carcinomas and 45 high-risk thymomas in this study. In univariate analysis, patient age, presence of myasthenia gravis, lesion shape, enhancement pattern, presence of necrosis or cystic change, mediastinal invasion, vessel invasion, lymphadenopathy, pericardial effusion, and distant organ metastasis were found to be statistically different between high-risk thymomas and thymic carcinomas (all P < 0.01). Random forest suggested that tumor shape, lymphadenopathy, and the presence of pericardial effusion were the key features in tumor differentiation. The predictive accuracy for the test data and whole data was 94.73% and 96.35%, respectively. Further receiver operating characteristic curve analysis showed the area under the curve was 0.957 (95% confidence interval, 0.986-0.929).

CONCLUSIONS

The random forest model in the present study has high efficiency in predictive diagnosis of thymic carcinomas and high-risk thymomas. Tumor shape, lymphadenopathy, and pericardial effusion are the key features for tumor differentiation. Thymic tumors with irregular shape, the presence of lymphadenopathy, and pericardial effusion are highly indicative of thymic carcinomas.

Risk stratification in GIST: shape quantification with CT is a predictive factor

Background

Tumor shape is strongly associated with some tumor’s genomic subtypes and patient outcomes. Our purpose is to find the relationship between risk stratification and the shape of GISTs.

Methods

A total of 101 patients with primary GISTs were confirmed by pathology and immunohistochemistry and underwent enhanced CT examination. All lesions’ pathologic sizes were 1 to 10 cm. Points A and B were the extremities of the longest diameter (LD) of the tumor and points C and D the extremities of the small axis, which was the longest diameter perpendicular to AB. The four angles of the quadrangle ABCD were measured and each angle named by its summit (A, B, C, D). For regular lesions, we took angles A and B as big angle (BiA) and small angle (SmA). For irregular lesions, we compared A/B ratio and D/C ratio and selected the larger ratio for analysis. The chi-square test, t test, ROC analysis, and hierarchical or binary logistic regression analysis were used to analyze the data.

Results

The BiA/SmA ratio was an independent predictor for risk level of GISTs (p = 0.019). With threshold of BiA at 90.5°, BiA/SmA ratio at 1.35 and LD at 6.15 cm, the sensitivities for high-risk GISTs were 82.4%, 85.3%, and 83.8%, respectively; the specificities were 87.1%, 71%, and 77.4%, respectively; and the AUCs were 0.852, 0.818, and 0.844, respectively. LD could not effectively distinguish between intermediate-risk and high-risk GISTs, but BiA could (p < 0.05). Shape and Ki-67 were independent predictors of the mitotic value (p = 0.036 and p < 0.001, respectively), and the accuracy was 87.8%.

Conclusions

Quantifying tumor shape has better predictive efficacy than LD in predicting the risk level and mitotic value of GISTs, especially for high-risk grading and mitotic value > 5/50HPF.

Key Points

• The BiA/SmA ratio was an independent predictor affecting the risk level of GISTs. LD could not effectively distinguish between intermediate-risk and high-risk GISTs, but BiA could.

• Shape and Ki-67 were independent predictors of the mitotic value.

• The method for quantifying the tumor shape has better predictive efficacy than LD in predicting the risk level and mitotic value of GISTs.