Deep learning analysis for differential diagnosis and risk classification of gastrointestinal tumors.

OBJECTIVES: Recently, artificial intelligence (AI) has been applied to clinical diagnosis. Although AI has already been developed for gastrointestinal (GI) tract endoscopy, few studies have applied AI to endoscopic ultrasound (EUS) images. In this study, we used a computer-assisted diagnosis (CAD) system with deep learning analysis of EUS images (EUS-CAD) and assessed its ability to differentiate GI stromal tumors (GISTs) from other mesenchymal tumors and their risk classification performance. MATERIALS AND METHODS: A total of 101 pathologically confirmed cases of subepithelial lesions (SELs) arising from the muscularis propria layer, including 69 GISTs, 17 leiomyomas and 15 schwannomas, were examined. A total of 3283 EUS images were used for training and five-fold-cross-validation, and 827 images were independently tested for diagnosing GISTs. For the risk classification of 69 GISTs, including very-low-, low-, intermediate- and high-risk GISTs, 2,784 EUS images were used for training and three-fold-cross-validation. RESULTS: For the differential diagnostic performance of GIST among all SELs, the accuracy, sensitivity, specificity and area under the receiver operating characteristic (ROC) curve were 80.4%, 82.9%, 75.3% and 0.865, respectively, whereas those for intermediate- and high-risk GISTs were 71.8%, 70.2%, 72.0% and 0.771, respectively. CONCLUSIONS: The EUS-CAD system showed a good diagnostic yield in differentiating GISTs from other mesenchymal tumors and successfully demonstrated the GIST risk classification feasibility. This system can determine whether treatment is necessary based on EUS imaging alone without the need for additional invasive examinations.

Ruptured Duodenal GIST in a Young Female - A Rare Presentation and Comprehensive Review.

Gastrointestinal stromal tumors (GISTs) are uncommon mesenchymal neoplasms primarily found in the gastrointestinal (GI) tract. While they typically occur in older adults, GISTs can manifest in individuals of any age. This publication paper presents a case study of a rare occurrence: a spontaneously ruptured duodenal GIST in a young female. Through an in-depth review of the literature, we aim to provide insights into the clinical presentation, diagnostic challenges, management strategies, and long-term outcomes associated with this unique clinical scenario.

RésuméLes tumeurs stromales gastro-intestinales (GIST) sont des néoplasmes mésenchymateux peu communs principalement trouvés dans le tractus gastro-intestinal (GI). Bien qu'elles surviennent généralement chez les adultes plus âgés, les GIST peuvent se manifester chez des individus de tout âge. Cet article présente une étude de cas d'une occurrence rare: une GIST duodénale spontanément rompue chez une jeune femme. À travers une revue approfondie de la littérature, nous visons à fournir des éclairages sur la présentation clinique, les défis diagnostiques, les stratégies de prise en charge et les résultats à long terme associés à ce scénario clinique unique.

Excision of a gastrointestinal stromal tumour in a dog using short-wave infrared fluorescence imaging and indocyanine green.

A 7-year-old castrated male Golden Retriever weighing 36.8 kg presented to the Veterinary Teaching Hospital with vomiting, anorexia and depression. After blood tests, radiographic, ultrasound and computed tomography examinations, a 7.85 × 5.90 × 8.75 cm mass was identified in the caecum. To visualise the tumour margin and improve the accuracy of tumour resection, intraoperative short-wave infrared imaging using indocyanine green was performed during surgery. An indocyanine green solution was injected intravenously as a bolus of 5 mg/kg 24 h before surgery. Tumour resection was performed with a 0.5 cm margin from the fluorescent-marked tissues. Histopathological examination revealed a diagnosis of a gastrointestinal stromal tumour (GIST) and the absence of neoplastic cells in the surgical margin, indicating a successful surgery. To our knowledge, this is the first case of a GIST resection in a dog using intraoperative short-wave infrared imaging.

Clinical course of asymptomatic duodenal subepithelial lesions.

BACKGROUND/AIMS: There is limited knowledge regarding the management of duodenal subepithelial lesions (SELs) owing to a lack of understanding of their natural course. This study aimed to assess the natural course of asymptomatic duodenal SELs and provide management recommendations. METHODS: Patients diagnosed with duodenal SELs and followed up for a minimum of 6 months were retrospectively investigated. RESULTS: Among the 443,533 patients who underwent esophagogastroduodenoscopy between 2008 and 2020, duodenal SELs were identified in 0.39% (1,713 patients). Among them, 396 duodenal SELs were monitored for a median period of 72.5 months (interquartile range, 37.7-111.3 mo). Of them, 16 SELs (4.0%) showed substantial changes in size or morphology at a median follow-up of 35.1 months (interquartile range, 21.7-51.4 mo). Of these SELs with substantial changes, tissues of two SELs were acquired using endoscopic ultrasound-guided fine needle aspiration biopsy: one was a lipoma and the other was non-diagnostic. Three SELs were surgically or endoscopically removed; two were diagnosed as gastrointestinal stromal tumors, and one was a lipoma. An initial size of 20 mm or larger was associated with substantial changes during follow-up (p = 0.016). CONCLUSION: While the majority of duodenal SELs may not exhibit substantial interval changes, regular follow-up with endoscopy may be necessary for cases with an initial size of 20 mm or larger, considering a possibility of malignancy.

Non-gastrointestinal stromal tumor, mesenchymal neoplasms of the gastrointestinal tract: a review of tumor genetics, pathology, and cross-sectional imaging findings.

There is a diverse group of non-gastrointestinal stromal tumor (GIST), mesenchymal neoplasms of the gastrointestinal (GI) tract that demonstrate characteristic pathology and histogenesis as well as variable imaging findings and biological behavior. Recent advancements in tumor genetics have unveiled specific abnormalities associated with certain tumors, influencing their molecular pathogenesis, biology, response to treatment, and prognosis. Notably, giant fibrovascular polyps of the esophagus, identified through MDM2 gene amplifications, are now classified as liposarcomas. Some tumors exhibit distinctive patterns of disease distribution. Glomus tumors and plexiform fibromyxomas exhibit a pronounced affinity for the gastric antrum. In contrast, smooth muscle tumors within the GI tract are predominantly found in the esophagus and colorectum, surpassing the incidence of GISTs in these locations. Surgical resection suffices for symptomatic benign tumors; multimodality treatment may be necessary for frank sarcomas. This article aims to elucidate the cross-sectional imaging findings associated with a wide spectrum of these tumors, providing insights that align with their histopathological features.

Endoscopic Ultrasound-Guided Tissue Acquisition Using Fork-Tip Needle for Subepithelial Lesions: A Single-Center Validation Study.

BACKGROUND: The cutoff value for stereomicroscopic on-site evaluation (SOSE) in endoscopic ultrasound-guided tissue acquisition (EUS-TA) has high diagnostic sensitivity when a Franseen needle is employed for upper gastrointestinal subepithelial lesions (SELs) (stereomicroscopically visible white core [SVWC] ≥ 4 mm). AIM: We aimed to determine whether high diagnostic sensitivity could be obtained when EUS-TA was performed using a Fork-tip needle. METHODS: Twenty-one patients were prospectively registered. Patients underwent EUS-TA using a Fork-tip needle for upper gastrointestinal SELs at Kitasato University Hospital between January and November 2022. Punctures were made twice using the needle, and SOSE was conducted for each specimen. Blood and physical examination were performed to assess adverse events. Pathological diagnosis was made using hematoxylin and eosin-stained sections and immunohistochemical staining. Statistical comparisons were completed using Fisher's exact tests. RESULTS: The diagnostic rate of EUS-TA was 100% (21/21 cases). The final diagnosis was gastrointestinal stromal tumor in 17 (81.0%) and leiomyoma in 4 (19.0%) patients. SOSE was conducted on all 42 punctures, and the tissue sampling rate was 100% (42/42 punctures). Specimens with SVWC ≥ 4 mm were collected in 97.6% punctures (41/42 punctures) and the diagnostic sensitivity for these specimens was 100% (41/41 punctures), which is significantly higher (p < 0.0238) compared to the absence of cutoff value (diagnostic sensitivity of 0%). No EUS-TA-related adverse events occurred. CONCLUSIONS: EUS-TA combined with SOSE for upper gastrointestinal SEL using a fork-tip needle had a high diagnostic rate, and the cutoff value of SVWC ≥ 4 mm had high diagnostic sensitivity.

Multimode ultrasound imaging of an anal canal gastrointestinal stromal tumour (GIST).

Gastrointestinal stromal tumours (GISTs) can develop throughout the entire gastrointestinal tract, but these tumours are usually found in the stomach and small intestine. In this case, a rare GIST arising from the anal canal was investigated using high-frequency endoanal ultrasound and external three-dimensional ultrasound with tomographic ultrasound imaging. The endoanal approach revealed the inner structure of the tumour. External ultrasound was used to determine the relationship between the lesion and surrounding tissues. In the limited reports of anal canal GISTs, no other lesions have been correctly diagnosed preoperatively or displayed in detail on imaging. The multilayer structure of the anal sphincter and these lesions can be clearly displayed by a variety of ultrasound imaging methods, which are nonradiative, low-cost and easily accessible. Modern ultrasound has the potential for broad application in anal canal tumour diagnosis and surveillance.

The correlation of sarcopenia and adverse events of imatinib therapy postoperatively in gastrointestinal stromal tumor through computed tomography quantitative body composition.

PURPOSE: This study aimed to investigate the correlation between sarcopenia and adverse events (AEs) of postoperative imatinib therapy through computed tomography (CT) quantitative body composition for intermediate- and high-risk gastrointestinal stromal tumors (GISTs). METHODS: The study retrospectively analyzed the clinical data of 208 patients with intermediate- and high-risk GIST treated surgically and treated with imatinib afterward at the First Affiliated Hospital of Wenzhou Medical University between October 2011 and October 2021. Images of preoperative CT scans within 1 month were used to determine the body composition of the patients. On the basis of the L3 skeletal muscle index, patients were classified into sarcopenia and nonsarcopenia groups. In 2 groups, AEs related to imatinib were analyzed. RESULTS: The proportion of AEs related to imatinib in the sarcopenia group was higher, and this disparity had a significant statistical significance (P = .013). Sarcopenia was significantly associated with hemoglobin reduction compared with nonsarcopenia (P = .015). There was a significant difference between the sarcopenia group and the nonsarcopenia group in the ratio of severe AEs (grades 3-4). Hemoglobin content (odds ratio [OR], 0.981; 95% CI, 0.963-1.000; P = .045), sex (OR, 0.416; 95% CI, 0.192-0.904; P = .027), and sarcopenia (OR, 5.631; 95% CI, 2.262-14.014; P < .001) were the influential factors of imatinib severe AEs in patients with intermediate- and high-risk GIST within 1 year after imatinib treatment. CONCLUSION: Patients with preoperative sarcopenia have a higher incidence and severity of AEs during adjuvant imatinib therapy.

Clinical course of small gastric subepithelial lesion less than 20 mm diagnosed by endoscopic ultrasound-guided fine-needle aspiration.

BACKGROUND AND AIM: Gastrointestinal stromal tumors (GISTs) are treated as malignant gastric subepithelial lesions (SELs), and resection is recommended. However, small gastric SELs < 20 mm with no malignant features are monitored without histopathological examination, and the frequency of malignancy is unknown. This study aimed to clarify the clinicopathological findings and clinical course of gastric SELs < 20 mm measured by endoscopic ultrasound (EUS). METHODS: This retrospective cohort study included consecutive patients with small gastric SELs < 20 mm diagnosed using EUS at a tertiary referral center between 2009 and 2021. The clinical course after diagnosis using EUS-guided fine-needle aspiration (EUS-FNA) was reviewed. RESULTS: Among 333 patients with small gastric SELs, 104 patients with 105 lesions underwent EUS-FNA. The pathological diagnosis was confirmed in 87 patients. GISTs were the most common pathology (47%). Among the 87 patients, 43 underwent therapeutic interventions, including tumor resection and chemotherapy. In groups of tumor resection, the pathological tumor size on the resected specimen was significantly larger than the size measured by EUS (19.5 mm vs 15.0 mm, P < 0.001), and 37% of resected SELs were 20 mm or over. No recurrence was observed after tumor resection during a mean follow-up period of 40 months. CONCLUSIONS: Approximately 40% of small gastric SELs were malignant tumors, such as GIST, with most of them requiring treatment. Additionally, considering that the EUS measurement is 5 mm smaller than the pathological tumor diameter, further examinations, such as systematic EUS-FNA, may be required for SEL, including those smaller than 20 mm.

Convolutional neural network applied to preoperative venous-phase CT images predicts risk category in patients with gastric gastrointestinal stromal tumors.

OBJECTIVE: The risk category of gastric gastrointestinal stromal tumors (GISTs) are closely related to the surgical method, the scope of resection, and the need for preoperative chemotherapy. We aimed to develop and validate convolutional neural network (CNN) models based on preoperative venous-phase CT images to predict the risk category of gastric GISTs. METHOD: A total of 425 patients pathologically diagnosed with gastric GISTs at the authors' medical centers between January 2012 and July 2021 were split into a training set (154, 84, and 59 with very low/low, intermediate, and high-risk, respectively) and a validation set (67, 35, and 26, respectively). Three CNN models were constructed by obtaining the upper and lower 1, 4, and 7 layers of the maximum tumour mask slice based on venous-phase CT Images and models of CNN_layer3, CNN_layer9, and CNN_layer15 established, respectively. The area under the receiver operating characteristics curve (AUROC) and the Obuchowski index were calculated to compare the diagnostic performance of the CNN models. RESULTS: In the validation set, CNN_layer3, CNN_layer9, and CNN_layer15 had AUROCs of 0.89, 0.90, and 0.90, respectively, for low-risk gastric GISTs; 0.82, 0.83, and 0.83 for intermediate-risk gastric GISTs; and 0.86, 0.86, and 0.85 for high-risk gastric GISTs. In the validation dataset, CNN_layer3 (Obuchowski index, 0.871) provided similar performance than CNN_layer9 and CNN_layer15 (Obuchowski index, 0.875 and 0.873, respectively) in prediction of the gastric GIST risk category (All P >.05). CONCLUSIONS: The CNN based on preoperative venous-phase CT images showed good performance for predicting the risk category of gastric GISTs.

Radiomics analysis of contrast-enhanced computerized tomography for differentiation of gastric schwannomas from gastric gastrointestinal stromal tumors.

PURPOSE: To assess the performance of radiomics-based analysis of contrast-enhanced computerized tomography (CE-CT) images for distinguishing GS from gastric GIST. METHODS: Forty-nine patients with GS and two hundred fifty-three with GIST were enrolled in this retrospective study. CT features were evaluated by two associate chief radiologists. Radiomics features were extracted from portal venous phase images using Pyradiomics software. A non-radiomics dataset (combination of clinical characteristics and radiologist-determined CT features) and a radiomics dataset were used to build stepwise logistic regression and least absolute shrinkage and selection operator (LASSO) logistic regression models, respectively. Model performance was evaluated according to sensitivity, specificity, accuracy, and receiver operating characteristic (ROC) curve, and Delong's test was applied to compare the area under the curve (AUC) between different models. RESULTS: A total of 1223 radiomics features were extracted from portal venous phase images. After reducing dimensions by calculating Pearson correlation coefficients (PCCs), 20 radiomics features, 20 clinical characteristics + CT features were used to build the models, respectively. The AUC values for the models using radiomics features and those using clinical features were more than 0.900 for both the training and validation groups. There were no significant differences in predictive performance between the radiomic and clinical data models according to Delong's test. CONCLUSION: A radiomics-based model applied to CE-CT images showed comparable predictive performance to senior physicians in the differentiation of GS from GIST.

Development of a multi-phase CT-based radiomics model to differentiate heterotopic pancreas from gastrointestinal stromal tumor.

BACKGROUND: To investigate whether CT-based radiomics can effectively differentiate between heterotopic pancreas (HP) and gastrointestinal stromal tumor (GIST), and whether different resampling methods can affect the model's performance. METHODS: Multi-phase CT radiological data were retrospectively collected from 94 patients. Of these, 40 with HP and 54 with GISTs were enrolled between April 2017 and November 2021. One experienced radiologist manually delineated the volume of interest and then resampled the voxel size of the images to 0.5 × 0.5 × 0.5 mm3, 1 × 1 × 1 mm3, and 2 × 2 × 2 mm3, respectively. Radiomics features were extracted using PyRadiomics, resulting in 1218 features from each phase image. The datasets were randomly divided into training set (n = 66) and validation set (n = 28) at a 7:3 ratio. After applying multiple feature selection methods, the optimal features were screened. Radial basis kernel function-based support vector machine (RBF-SVM) was used as the classifier, and model performance was evaluated using the area under the receiver operating curve (AUC) analysis, as well as accuracy, sensitivity, and specificity. RESULTS: The combined phase model performed better than the other phase models, and the resampling method of 0.5 × 0.5 × 0.5 mm3 achieved the highest performance with an AUC of 0.953 (0.881-1), accuracy of 0.929, sensitivity of 0.938, and specificity of 0.917 in the validation set. The Delong test showed no significant difference in AUCs among the three resampling methods, with p > 0.05. CONCLUSIONS: Radiomics can effectively differentiate between HP and GISTs on CT images, and the diagnostic performance of radiomics is minimally affected by different resampling methods.