Imaging pediatric gastrointestinal stromal tumor (GIST)

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.

Use of a Convolutional Neural Network to Predict the Malignant Potential of Gastrointestinal Stromal Tumors in Transabdominal Ultrasound Images: Visualization of the Focus of the Prediction Model

OBJECTIVE

We established a deep convolutional neural network (CNN) model based on ultrasound images (US-CNN) for predicting the malignant potential of gastrointestinal stromal tumors (GISTs).

METHODS

A total of 980 ultrasound images from 245 pathology-confirmed GIST patients after surgical operation were retrospectively collected and divided into a low (very-low-risk, low-risk) and a high (medium-risk, high-risk) malignant potential group. Eight pre-trained CNN models were used to extract the features. The CNN model with the highest accuracy in the test set was selected. The model's performance was evaluated by calculating accuracy, sensitivity, specificity, positive-predictive value (PPV), negative-predictive value (NPV) and the F1 score. Three radiologists with different experience levels also predicted the malignant potential of GISTs in the same test set. US-CNN and human assessments were compared. Subsequently, gradient-weighted class activation diagrams (Grad-CAMs) were used to visualize the model's final classification decisions.

RESULTS

Among the eight transfer learning-based CNNs, ResNet18 performed best. The accuracy, sensitivity, specificity, PPV, NPV and F1 score were 0.88, 0.86, 0.89, 0.82, 0.92 and 0.90, respectively, which were significantly better than those achieved by radiologists (resident doctor: 0.66, 0.55, 0.79, 0.74, 0.62 and 0.69; attending doctor: 0.68, 0.59, 0.78, 0.70, 0.69 and 0.73; professor: 0.69, 0.63, 0.72, 0.51, 0.80 and 0.76). Model interpretation with Grad-CAMs revealed that the activated areas mainly focused on cystic necrosis and margins.

CONCLUSION

The US-CNN model predicts GIST malignant potential well, which can assist in clinical treatment decision-making.

Imaging of pediatric abdominal soft tissue tumors: A COG Diagnostic Imaging Committee/SPR Oncology Committee White Paper

This paper provides imaging recommendations for pediatric abdominal tumors that arise outside of the solid viscera. These tumors are rare in children and have been categorized in two groups: abdominal wall and peritoneal tumors (desmoid tumor and desmoplastic small round cell tumor) and tumors that arise from the gastrointestinal tract (gastrointestinal stromal tumor and gastrointestinal neuroendocrine tumor). Authors offer consensus recommendations for imaging assessment of these tumors at diagnosis, during follow-up, and when off-therapy.

Endoscopic Ultrasound for Diagnosis and Management of Pediatric Gastrointestinal Stromal Tumor

Gastrointestinal (GI) stromal tumors arise from the interstitial cells of Cajal and are rare in the pediatric population. The most common clinical manifestation is anemia secondary to GI bleeding. Endoscopy is commonly used for diagnostic and therapeutic interventions of an obstructing mass or gastrointestinal bleed, while experience with endoscopic ultrasound (EUS) and EUS fine needle aspiration (EUS-FNA) for pediatric patients with suspected gastric tumors is limited. We report 2 cases, a 14-year-old male and an 11-year-old female, who presented with symptomatic anemia. Both patients were diagnosed with GI stromal tumors of the stomach using EUS and EUS-FNA. This report shows that EUS and EUS-FNA are safe and effective diagnostic tools for pediatric patients.

Latest Advances in the Management of Pediatric Gastrointestinal Stromal Tumors

Gastrointestinal stromal tumor is the most common mesenchymal neoplasm of the gastrointestinal tract, usually found in elderly adults. It is infrequent among pediatric patients and usually differs biologically from adult-type diseases presenting mutations of KIT and PDGFR genes. In this population, more frequent is the wild-type GIST possessing SDH, TRK, RAS, NF1 mutations, among others. Both tumor types require individualized treatment with kinase inhibitors that are still being tested in the pediatric population due to the different neoplasm biology. We review the latest updates to the management of pediatric gastrointestinal tumors with a particular focus on the advances in molecular biology of the disease that enables the definition of possible resistance. Emerging treatment with kinase inhibitors that could serve as targeted therapy is discussed, especially with multikinase inhibitors of higher generation, the effectiveness of which has already been confirmed in the adult population.

Synchronous occurrence of gastrointestinal stromal tumor, pancreatic intraductal papillary mucinous neoplasm, and intrahepatic cholangiocarcinoma: Case report

RATIONALE

Gastrointestinal stromal tumor (GIST) is the most common primary mesenchymal tumors in gastrointestinal tract. Synchronous occurrence of GIST and tumors in other organs is rare. We first report an exceedingly rare case of synchronous occurrence of gastric GIST, pancreatic intraductal papillary mucinous neoplasm (IPMN) and intrahepatic cholangiocarcinoma.

PATIENT CONCERNS

A 70-year-old male presented to our hospital because of abdominal pain and dyspepsia. Tumor markers and liver function were abnormal. Abdomen computed tomography showed concurrent tumors in stomach, pancreas, and liver.

DIAGNOSIS

Pathology confirmed synchronous occurrence of gastric GIST, pancreatic IPMN and intrahepatic cholangiocarcinoma.

INTERVENTIONS

Mass excision, partly gastrectomy, wedge resection of VIII liver segments, and pancreatic-oduodenectomy were performed.

OUTCOMES

During the 18-month follow-up, both laboratory tests and computed tomography examination revealed no sign of recurrence or metastasis. Currently, the patient is free of clinical symptoms such as abdominal discomfort, jaundice, and fever.

CONCLUSION

As yet, no cases simultaneously with gastric GIST, pancreatic IPMN and intrahepatic cholangiocarcinoma have been described in literatures. This report increases the knowledge to avoid misdiagnosis and delayed therapy for coexistence of the described 3 types of neoplasm.

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.

Adoption of Two-Dimensional Ultrasound Gastrointestinal Filling Contrast on Artificial Intelligence Algorithm in Clinical Diagnosis of Gastric Cancer

This research was aimed to explore the value of gastrointestinal filling contrast-enhanced ultrasound (CEUS) and computed tomography (CT-)-enhanced scanning based on artificial intelligence (AI) algorithm in the evaluation of gastric cancer staging. 102 patients with gastric cancer were selected as the research objects. All of them underwent CEUS of gastrointestinal filling and 64-slice spiral CT before surgery. In addition, an improved mean shift algorithm was proposed based on differential optical flow and deep convolutional neural network (D-CNN), which was applied in image processing. The predicted positive rate (PPR), sensitivity, specificity, and accuracy of gastric cancer in different stages by CEUS and CT were calculated using pathological diagnosis results as the gold standard. 17 patients with T1 stage, 41 patients with T2-T3 stage, and 35 patients with T4 stage were detected by CEUS. 13 patients with T1 stage, 34 patients with T2-T3 stage, and 30 patients with T4 stage were detected by CT enhanced examination. The PPRs of CEUS for T1, T2-T3, and T4 stages of gastric cancer were higher than those of CT enhanced (P < 0.05). The PPR of CEUS for N0 staging of gastric cancer was higher than that of CT enhanced (P < 0.05), and it for N3 staging of gastric cancer was lower than that of CT enhanced (P < 0.05). From the analysis of M staging of gastric cancer, the PPRs of CEUS for M0 and M1 staging of gastric cancer were not statistically different from the PPRs of CT enhanced (P > 0.05). The sensitivity (95.6%), specificity (81.82%), and accuracy (94.12%) of CEUS in assessing resectability were significantly higher than those of CT enhancement (89.01%, 63.67%, and 86.27%, respectively), and the differences were statistically significant (P < 0.05). In summary, CEUS gastrointestinal filling based on the D-CNN algorithm could better improve the display rate of the tissue lesions around the stomach. It also helped to judge the lesion progress, the depth of infiltration, and lymph node metastasis of the lesion. In addition, it had excellent performance in evaluating the resectability of gastric cancer before surgery and had clinical promotion value.

Application Effect Analysis of Operating Room Detailed Nursing Based on Medical Big Data in Patients Undergoing Gastrointestinal Tumor Surgery

With the continuous development of internet information computing, the continuous improvement of medical and health systems, and the continuous increase of medical big data, traditional operating room care also needs to be further optimized. Medical big data is a forum data set for medical industry healthcare, electronic medical record information, clinical case record information, medical financial data, remote patient monitoring data, clinical decision support data, medical insurance data set, online consulting platform, and so on. Gastrointestinal tumors are currently one of the largest malignant tumors. Compared with ordinary patients, the presence of fear, depression, irritability, and other unhealthy emotions in patients with gastrointestinal tumors will reduce the therapeutic effect. Without careful care, the use of chemotherapy and other treatments makes patients vulnerable to various side effects. This article aims to study the use of medical big data intelligent algorithms to perform detailed care for patients during gastrointestinal tumor surgery and analyze the effects of care. This paper proposes an improved DNN algorithm; the DNN algorithm is to use several weight coefficient matrices and bias vectors to perform a series of linear operations and activation operations with the input value vector, starting from the input layer, backward calculation layer by layer, until the operation reaches the output layer, and the output result is obtained. This algorithm is used to study the theory, use mathematical formulas for method calculation and model design, and use the model to carry out detailed nursing experiments in the relevant operating room. The results of the experiment show that patients who have performed detailed care have a 27.2% improvement in treatment and rehabilitation effects than those who have not, and the level of detailed care has an obvious positive relationship with the rate of condition conversion. In the end, the hospital's detailed care quality evaluation index, which is QEI, increases by 1 point, which can increase the rate of condition conversion by 0.4.

International consensus on initial screening and follow-up of asymptomatic SDHx mutation carriers

Approximately 20% of patients diagnosed with a phaeochromocytoma or paraganglioma carry a germline mutation in one of the succinate dehydrogenase (SDHx) genes (SDHA, SDHB, SDHC and SDHD), which encode the four subunits of the SDH enzyme. When a pathogenic SDHx mutation is identified in an affected patient, genetic counselling is proposed for first-degree relatives. Optimal initial evaluation and follow-up of people who are asymptomatic but might carry SDHx mutations have not yet been agreed. Thus, we established an international consensus algorithm of clinical, biochemical and imaging screening at diagnosis and during surveillance for both adults and children. An international panel of 29 experts from 12 countries was assembled, and the Delphi method was used to reach a consensus on 41 statements. This Consensus Statement covers a range of topics, including age of first genetic testing, appropriate biochemical and imaging tests for initial tumour screening and follow-up, screening for rare SDHx-related tumours and management of elderly people who have an SDHx mutation. This Consensus Statement focuses on the management of asymptomatic SDHx mutation carriers and provides clinicians with much-needed guidance. The standardization of practice will enable prospective studies in the near future.

Pediatric Gastrointestinal Stromal Tumors and Neuroendocrine Tumors: Advances in Surgical Management

Gastrointestinal stromal tumors and neuroendocrine tumors in adult and pediatric populations differ immensely. Despite these established differences, the extreme rarity of gastrointestinal stromal tumors and neuroendocrine tumors in the pediatric population has resulted in the lack of consensus management guidelines, making optimal surgical approaches unclear. Comprehensive management principles to guide surgical approaches in adult literature are extensive. However, these are still lacking for pediatric patients. International cooperation to develop standardized pediatric-specific guidelines is urgently warranted in the future. This article highlights the vast differences between adult and pediatric parameters and provides recommendations on optimal and novel surgical approaches in children.

Ultrasonography in infectious and neoplastic diseases of the bowel and peritoneum

Ultrasonography is not the most cited imaging technique for the evaluation of infectious and neoplastic diseases of the gastrointestinal tract and the peritoneum, but it is often the initial technique used in the initial workup for nonspecific clinical syndromes. Despite its limitations, ultrasonography's strengths enable it to provide meaningful diagnostic information. To discuss the most important ultrasonographic, clinical, and epidemiological findings for infectious disease, we follow a topographical approach: stomach (Anisakis), proximal small bowel (Giardia lamblia, Strongyloides stercoralis, Mycobacterium avium-intracellulare complex, and Cryptosporidium), distal small bowel (Yersinia, Salmonella, and Campylobacter), terminal ileum and cecum (tuberculosis), right colon (Entamoeba histolytica), left colon (Shigella), sigmoid colon and rectum, pancolitis (Clostridium difficile, Cytomegalovirus, and Escherichia coli), and peritoneum. To discuss the ultrasonographic and clinical findings of the most common neoplastic diseases, we follow a nosological approach: polyploid lesions as precursors of tumors, carcinomas, neuroendocrine tumors, hematological tumors, mesenchymal tumors, and metastases. We briefly discuss tumors of the peritoneum and the use of ultrasonography to guide percutaneous biopsy procedures.

Differential diagnosis of liver metastases of gastrointestinal stromal tumors from colorectal cancer based on combined tumor biomarker with features of conventional ultrasound and contrast-enhanced ultrasound

PURPOSE

To explore the value of tumor marker CA125 and CEA linked with conventional ultrasound (US) and contrast-enhanced ultrasound (CEUS) features in differentiating gastrointestinal stromal tumors liver metastases (GISTLM) from colorectal cancer liver metastases (CRCLM).

MATERIALS AND METHODS

From December 2005 to February 2019, eighty patients with pathologically proven GISTLM together with 80 CRCLM patients were retrospectively evaluated with contrast-enhanced ultrasound (CEUS). Clinical characteristics such as CA125 and CEA were documented to compare the difference between GISTLM and CRCLM. Univariate analysis was performed to determine significant features in ultrasound and then these features were entered into multivariate logistic regression model to determine diagnostic criteria. By analyzing the tumor marker and imaging features, diagnostic performance was evaluated via receiver-operating characteristic (ROC) analysis. The sensitivity, specificity and accuracy were calculated for the diagnosis of GISTLM.

RESULTS

Multiple logistic regression analysis showed that increased CA125 and normal CEA were the independent variables of GISTLM. On conventional US, the features of hypo- or mix-echogenicity and anechoic area were associated with GISTLM. On CEUS, capsule enhancement, starting time of washout > 40 s and proportion of non-enhancement area > 20% were the features indicating GISTLM. All of the p values were < 0.05. When linking tumor marker with imaging features, the diagnostic sensitivity improved from 36.3-57.5% to 70.0%, and the area under the ROC (AUROC) curve improved from 0.681-0.750 to 0.838, with a specificity of 97.5%.

CONCLUSIONS

Combining the imaging features of conventional US and CEUS with serum tumor markers provides a potentially effective diagnostic method in differentiation of GISTLM and CRCLM.

The pediatric stomach - masses and mass-like pathology

Diagnostic imaging of pediatric gastric masses often provides a challenge for the practicing radiologist. Radiologists should be aware of this relatively unusual pathology, particularly in cross-sectional imaging findings. We will review pediatric gastric masses and mass-like lesions, focusing on neoplastic and inflammatory etiologies.

Multiparametric MRI and 18F-FDG PET features for differentiating gastrointestinal stromal tumors from benign gastric subepithelial lesions

OBJECTIVES

To investigate whether multiparametric magnetic resonance imaging (MRI) and ¹⁸F-fluorodeoxyglucose positron emission tomography (PET) can be helpful in differentiating gastrointestinal stromal tumors (GISTs) from non-GISTs and high-risk GISTs from low-risk GISTs.

METHODS

This retrospective study included 56 patients with pathologically confirmed GISTs (n = 39), leiomyoma (n = 8), schwannoma (n = 5), heterotopic pancreas (n = 3), and glomus tumor (n = 1) who underwent MRI and/or PET examinations. Two radiologists reviewed MRI regarding location, shape, contour, growth pattern, margin, signal intensity (SI) on T1- (T1WI) and T2-weighted images (T2WI), degree and pattern of enhancement, hemorrhage, and necrosis. Mean apparent diffusion coefficient (ADC) and maximum standardized uptake value (SUV_max) were measured. Imaging features were compared among non-GISTs, low-risk GISTs, and high-risk GISTs using uni- and multivariate statistical analyses.

RESULTS

Size, longitudinal location, shape, contour, growth pattern, SI on T1- and T2WI, enhancement pattern, hemorrhage, necrosis, ADC, and SUV_max were significantly different among non-GISTs, low-risk GISTs, and high-risk GISTs (p < 0.05). On multivariate analysis, SI on T2WI (hazard ratio [HR], 66.0; p = 0.002) was the only independent variable for differentiating GISTs from non-GISTs whereas enhancement pattern (HR, 56.0; p = 0.041), ADC (HR, 0.997; p = 0.01), and SUV_max (HR, 2.08; p = 0.027) were significant features for differentiating between high-risk and low-risk GISTs.

CONCLUSIONS

Several qualitative and quantitative MRI and PET features including ADC and SUV_max values are significantly different among non-GISTs, low-risk GISTs, and high-risk GISTs. Multiparametric information obtained from MRI with or without PET can be useful for differentiation of gastric subepithelial tumors as well as for determining patients' management and prognosis.

KEY POINTS

• Several qualitative MRI features are helpful in distinguishing gastrointestinal stromal tumors (GISTs) from non-GISTs as well as high-risk GISTs from low-risk GISTs. • Apparent diffusion coefficient value on diffusion-weighted imaging can be useful in distinguishing GISTs from non-GISTs as well as high-risk GISTs from low-risk GISTs. • PET has the potential to distinguish between high-risk and low-risk GISTs.

Pediatric gastrointestinal stromal tumors-a review of diagnostic modalities

Gastrointestinal stromal tumors are exceedingly rare tumors in the pediatric population, as a result many clinicians either may never see this diagnosis or will encounter it only a few times throughout their careers. It is imperative in the pediatric population to follow appropriate steps to ensure a swift diagnosis and referral to specialized centers that are equipped with the multidisciplinary teams accustomed to treating rare diseases. This review aims to discuss the most recent data available on the diagnostic modalities utilized in cases of suspected Pediatric GIST.