A diagnostic model based on clinical and enhanced computed tomography features to identify gastric schwannoma and gastric stromal tumor of different sizes: a multi-institution retrospective study

Introduction

Gastric stromal tumor (GST) is the most common malignant stromal tumor in the stomach (1). Although complete surgical resection is feasible in most localized gastrointestinal stromal tumors, metastatic relapse occurs in approximately 40% of patients (1,2). The main surgical methods are laparotomy and laparoscopic surgery (3,4). According to the consensus classification of the National Institutes of Health, regardless of the mitotic count, stromal tumors with a diameter ≥5 cm are considered invasive clinical processes and have with medium to high risk, and laparoscopic surgery can be considered only in GST lesions with a diameter <5 cm and in favorable anatomical locations. Therefore, the treatment plans are different in patients with different tumor sizes, with a threshold of 5 cm. Furthermore, high-risk GST should receive imatinib treatment as adjuvant or neoadjuvant therapy (5,6). Gastric schwannoma (GS) accounts for less than 10% of gastric mesenchymal tumors and is a benign tumor with very few malignant transformations (7,8). GS can be completely resected by gastroscopic resection, with the advantages of less trauma, good prognosis, and low cost (9). It is of great clinical significance to accurately identify the two tumors before surgery.

Different from mucosal gastric tumors, in patients with GST or GS, endoscopy typically reveals undamaged mucosa and suggests extrinsic compression of the gastrointestinal lumen, making them difficult to definitively diagnose with endoscopy prior to surgery. On account of the similar affected population, clinical symptoms, and computed tomography (CT) findings of GST and GS, it is still challenging to distinguish between the two tumors, particularly in large (≥5 cm) tumors, and misdiagnosis occurs frequently before surgery (10). Previous investigations have indicated the diagnostic value of multiple parameters. Specifically, demographic variables (patient age and gender), tumor site, density, enhancement pattern, growth pattern, lymph node, and density, may assist in differentiating GS from GST. Numerous previous studies (11-14) have directly compared tumor sizes rather than categorizing, potentially overlooking the possible impact of differing tumor sizes on the CT features and process of surgical decision-making. In addition, most of the previous studies of GS have comprised small sample sizes.

The radiomics features of tumors enable a more comprehensive, objective, and precise characterization of lesions and facilitate the creation of a radiomics database to support in-depth quantitative analyses (15). A radiomics model on contrast-enhanced CT images has been reported and demonstrated predictive performance comparable to that of senior radiologists in the differentiation of GS from GST (16). However, most models constructed based on radiomics methods are still in the scientific research stage and have not yet been clinically applied. The main reasons for the difficulty in implementing radiomics research are poor biological interpretability and clinical usability; in addition, the difficulty of obtaining a large number of samples of the two types of tumors and the challenge of effectively identifying and segmenting tumors are key factors restricting their development (17).

Therefore, our multicenter study aimed to determine which clinical and CT features are effective in differentiating large and small tumors of GST and GS, and to provide a reference for scientific planning of surgical methods. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-24-1606/rc).

Methods

Patients

This study included the complete clinical, imaging, and follow-up data of 493 patients with GST and 102 patients with GS between January 2008 and December 2022. The inclusion criteria were a postoperative pathological diagnosis of GST and GS, an enhanced CT scan completed within 15 days before surgery, complete clinical and pathological data, no treatment before surgery, and that all the lymph nodes found during surgery were completely resected (Figure 1). Patients were sourced from the following institutions: Tongde Hospital of Zhejiang Province, Xin Hua Hospital of Huainan, and Anqing Municipal Hospital. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This retrospective study was approved by the Ethics Committee of Tongde Hospital of Zhejiang Province (No. 2021-040). The other hospitals were informed and agreed to participate in this study. The requirement for individual informed consent for this retrospective analysis was waived.

Figure 1 Flow chart of study enrolment based on the need for differential diagnosis. ①②③ indicate that the order of exclusion from the group. CT, computed tomography; GS, gastric schwannoma; GST, gastric stromal tumor.

CT inspection methods

The CT scans included both non-contrast and contrast-enhanced images, acquired using one of three scanner models: SOMATOM Emotion 16 (Siemens, Erlangen, Germany), Definition AS (Siemens), or Optima CT680 (GE Healthcare, Chicago, IL, USA). Prior to scanning, patients were required to fast for a minimum of 4–6 hours and to consume 600–1,000 mL of water approximately 15 minutes before the procedure.

Image analysis

The CT images were evaluated according to the consensus of two radiologists with over 5 years of experience. Any discrepancies in the initial readings by the two radiologists were resolved through further discussion to reach a consensus. The size, location (cardia, fundus, gastric body, or antrum), contour (round, oval, or irregular), growth pattern (endoluminal, exophytic or mixed), lymph node, margin (clear or definite boundary) calcification, cystic degeneration, hemorrhage, ulceration, necrosis, calcification, surface ulceration, hemorrhage, intratumoral blood vessels, peritumoral exudation, necrosis under the tumor wall, and CT attenuation value were evaluated.

According to the maximum diameter of the axial plane of CT, the tumor size was divided into two categories: (I) larger than or equal to 5 cm and (II) less than 5 cm. If the parameters of the intraluminal tumor exceeded the edge of the gastric structure contour, the growth pattern was defined as exophytic. Tumors located at the edge of the gastric structure were defined as having an endoluminal growth pattern. If this parameter was across the edge of the gastric structure, it was defined as a mixed growth pattern. When the endoluminal surface of the affected digestive tract showed irregular and discontinuous focal defects, it was referred to as surface ulceration (11). Bleeding was defined as a plaque-like high-density area found in the nonenhanced CT scan and when the CT value was 60–80 Hounsfield units (HU) (13). The degree of contrast enhancement was evaluated by three nonoverlapping regions of interest, 10 mm in size, which were manually drawn on the nonenhanced images, arterial phase, and venous phase images of the solid tumor.

The results were averaged for both measurements. The density difference of the mean HU value on the enhanced and nonenhanced CT images of each lesion was obtained. Intratumoral vessels were assessed when enlarged and engorged blood vessels were detectable. Rough edges were considered indicative of peritumoral exudation. The presence of calcification and hemorrhage within the lesion and any exudate around the tumor were also evaluated. Descriptive analysis was performed on all CT parameters. The relationships between clinical characteristics, all CT parameters, and pathological findings were investigated.

Statistical analysis

Statistical analysis was performed using the software SPSS 25.0. (IBM Corp., Armonk, NY, USA). A P value <0.05 was considered statistically significant. Categorical tumor variables were compared between the two groups using the Chi-squared test. For continuous variables, comparisons were performed using either the independent samples t-test or the Mann-Whitney U test, depending on the data distribution, and statistically significant variables were analyzed by multivariate logistic regression and receiver operating characteristic curve (ROC) analysis. The area under the curve (AUC), accuracy (ACC), sensitivity (SEN), and specificity (SPE) of the two groups were analyzed. The judgment criteria were as follows: AUC value of 0.5–0.7, low diagnostic value; AUC 0.7–0.9, medium; AUC >0.9, high. The SEN, SPE, positive predictive value (PPV), negative predictive value (NPV), and ACC of each significant criterion were also determined for differentiating GST from GS.

Results

A total of 177 patients with tumors ≥5 cm were divided into a large GST (GST-L) group (n=164) and a large GS (GS-L) group (n=13), and 418 patients with tumors <5 cm were divided into a small GST (GST-S) group (n=329) and a small GS (GS-S) group (n=89). Among all 595 patients, 282 patients enrolled in Anqing Municipal Hospital (47.4%, 282/595), 133 patients enrolled in Taizhou Municipal Hospital (22.4%, 133/595), and 180 patients enrolled in Tongde Hospital of Zhejiang Province (30.3%, 180/595).

Large tumor groups

The clinical characteristics are shown in Table 1. Of the patients with GST-L, 50.6% (83/164) were males and 49.4% (81/164) were females. There were 15.4% (2/13) male patients and 84.6% (11/13) female patients with GS-L. A statistically significant difference was found regarding sex. No statistically significant differences were observed in age or clinical symptoms (P>0.05).

The qualitative analysis results of the CT findings are presented in Table 2. The lymph node and necrosis were significantly different between the two groups (P<0.001). Lymph nodes of 0–5 mm were more common in the GST-L group (87.2%, 143/164) than in the GS-L group (30.8%, 4/13). Lymph nodes of 5.1–10 mm accounted for 12.8 % (21/164) in the GST-L group and 38.5% (5/13) in the GS-L group. No lymph nodes were larger than 10 mm in the GST-L group, whereas 30.8% (4/13) of lymph nodes were larger than 10 mm in the GS-L group. A statistically significant difference was also observed in tumor necrosis between the two groups (P<0.05). Necrosis was more common in the GST-L group (95.7%, 157/164) than it was in the GS-L group (23.1%, 3/13) (Figures 2-5).

Figure 2 Endophytic growth pattern of GST. CT demonstrating a heterogeneous mass of the gastric body with surface ulceration (arrows) and necrosis (A-C). CT, computed tomography; GST, gastric stromal tumor.

Figure 3 Exophytic growth pattern of GS. CT demonstrating a homogeneous mass of the gastric antrum with complete mucosa, homogeneously light-moderate enhancement and large lymph nodes (arrows) (A-C). CT, computed tomography; GS, gastric schwannoma.

Figure 4 Distribution diagram of the corresponding necrosis numbers between GST and GS groups across different tumor sizes. GS, gastric schwannoma; GST, gastric stromal tumor.

Figure 5 Distribution diagram of the corresponding lymph node numbers between GST and GS groups across different tumor sizes. GS, gastric schwannoma; GST, gastric stromal tumor; LN, lymph node.

The findings from the quantitative analysis of the CT data are presented in Table 3. The tumor long diameter/short diameter (LD/SD) in the GST-L group (1.37±0.25) was larger than in the GS-L group (1.20±0.17). There were no significant differences in CT attenuation value (P>0.05).

In multivariate analysis, lymph node and necrosis were two significant (P<0.05) independent factors for distinguishing GST-L from GS-L lesions. The two meaningful parameters obtained in the multivariate analysis were subjected to ROC curve analysis. The combined diagnosis showed good ACC, with an AUC of 0.983. The ACC, SEN, SPE, PPV, and NPV were 95.5, 92.3, 95.7, 63.2, and 99.4, respectively (Table 4, Figure 6).

Figure 6 The ROC curves of combinations of significant factors in large and small tumor groups. AUC, area under the curve; ROC, receiver operating characteristic.

Small tumor groups

In the univariate analysis of risk factors for different lesions, sex, age, symptom, location, growth pattern, lymph node, necrosis, calcification, surface ulceration, necrosis under the tumor wall, and CT attenuation value were significant (P<0.05) features for differentiating GST-S from GS-S lesions. Lymph nodes of 5.1–10 mm and larger than 10 mm in the GST-S group (2.7%, 9/329 and 0%, 0/329) were fewer than they were in the GS-S group (10.1%, 9/89 and 2.2%, 2/89), respectively. Necrosis was more common in the GST-S group (37.1%, 122/329) than it was in the GS-S group (5.6%, 5/89) (Tables 1-3, Figures 2-5).

In multivariate analysis, sex, location, growth pattern, lymph node, and necrosis were significant (P<0.05) independent factors for distinguishing GST-S from GS-S lesions. The five meaningful parameters obtained in the multivariate analysis were subjected to ROC curve analysis. The combined diagnosis showed good ACC, with an AUC of 0.931. The ACC, SEN, SPE, PPV, and NPV were 77.8, 81.6, 92.2, 48.7, and 94.4, respectively (Table 4, Figure 6).

Discussion

To the best of our knowledge, tumor size has been analyzed as an ungrouped variable in many previous studies on the differential diagnosis value of GST from GS using CT. However, the risk of GST and GS varies greatly, and the treatment plans are different, even for tumors of the same size. Laparoscopic surgery should be considered only in GST lesions less than 5 cm in diameter, and the indication of laparoscopic resection should be agreed on a case-by-case basis following a multidisciplinary evaluation by teams with broad experience in laparoscopic surgery (2). For GST, even less than 5 cm, laparoscopic resection must be carefully performed to avoid tumor rupture, and peritoneal and hepatic surfaces should be carefully examined to exclude tumor spread (2,18). Positive margin and tumor rupture are also associated with a high risk of relapse. Meanwhile, due to the relatively less aggressiveness of GS, laparoscopic surgery and endoscopic resection are usually recommended (19), especially with the rapid development of endoscopic techniques, such as endoscopic submucosal excavation and endoscopic full-thickness resection, which can achieve complete resection through gastroscopic resection, with less trauma, better prognosis, and lower cost (20,21). Therefore, based on a significant impact on surgery, we compared the differences in GST and GS between the large tumor (≥5 cm) group and the small tumor (<5 cm) group individually.

We found that GST and GS usually showed non-specific clinical manifestations, presented as submucosal tumor and mild-to-moderate enhancement. Owing to the overlap, it is difficult to differentiate GS and GST without pathological evidence. In our study, the clinical and CT features showed high diagnostic value in both large groups and small groups, which may provide imaging evidence for rational selection of surgical methods. This CT-based evaluation method is relatively simple and intuitive, and does not require complex image depth-processing.

In our study, the differential CT features in large and small groups were not completely consistent. In large tumor groups, we found that sex, lymph node, necrosis, and LD/SD were meaningful parameters. Multivariate analysis showed that lymph node and necrosis were two significant independent factors. In the small tumor groups, there were more differences in the CT characteristics compared with the large tumor groups, including location, growth pattern, calcification, surface ulcer, necrosis of the tumor wall, and degree of enhancement. Growth pattern, lymph node, and necrosis were significant independent factors.

Necrosis had a high predictive value in the differentiation of two types of tumors, both in the large tumor group and in the small tumor group. Necrosis is more commonly found in GST lesions. It is an imaging sign that indicates the high risk of GST, and careful consideration should be given to the surgical strategies to avoid the risk of rupture, recurrence, and metastasis. Although the blood supply of GST is generally abundant compared with that of normal tissues, the rapid proliferation of these tumor cells leads to the formation of multiple relatively low-density areas in enhanced CT scans, yet the blood supply is relatively insufficient. Therefore, hemorrhage, necrosis, and cystic changes often appear in GST, as shown by their heterogeneous manifestations on CT. GS does not undergo degenerative changes; the growth rate of tumor cells and the formation of new blood vessels are slow, and the tumor is not prone to calcification, ulceration, and cystic necrosis. Surface ulceration of GST may be associated with tolerance of the ischemic mucosa to gastric acid.

According to the results of our study, lymph node was a highly specific CT finding in both the large tumor group and the small tumor group. Some 81.3% (13/16) of GS-S had lymph nodes around the lesion, whereas only 28.6% (16/56) of GST-S had lymph nodes. Compared to our study, the previous report noted a slightly higher incidence of perilesional lymph nodes, which may be attributed to their use of less stringent size (22). The short axis diameter in the previous study was 5 mm, whereas the long axis diameter in our study was 10 mm. The presence of lymph nodes suggests that GS may trigger systemic immune responses and induce the chemotactic recruitment of lymphocytes, a finding that aligns with the well-documented histological observation of lymphocyte activation (23). GS is often accompanied by obvious lymphoid mantle nodes, with a germinal center around the tumor, and diffuse lymphoid cell infiltration in the tumor can be seen in histopathology. The characteristic histological features of GS are widely used as important histological identification points for GS because they have never been observed in other gastric subepithelial tumors, such as GST or leiomyomas.

In our study, GST and GS showed different growth patterns in the small tumor groups. This result did not appear in the large tumor groups. Compared with previous research (22,24), which suggested that the growth pattern of GS was significantly different from that of GST, this result was quite different in the large tumor groups. The differential diagnosis value of growth pattern is limited to small tumors. A study of 81 patients with gastrointestinal stromal tumors (25) found that large tumors had a tendency of exogenously. Therefore, we believe that in our study, a large tumor size (mean diameter ± standard deviation, 8.4±4.1 cm), may be the main reason for the exogenous or mixed growth pattern.

Several limitations of our study should be acknowledged. First, different tumor types have different numbers of tumors, which is inevitable due to variations in the incidence of gastric subepithelial tumors. Second, the retrospective design of this study precluded the use of a standardized CT protocol. In some patients, there is no gastric expansion in CT images, which may affect the growth pattern of the tumor. Therefore, a standardized gastric CT scheme should be used for further prospective studies.

Conclusions

We can efficiently distinguish GST and GS of different sizes using the hierarchical models. Necrosis and lymph nodes in both large and small tumor models exhibit high discriminative weights.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-24-1606/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-24-1606/dss

Funding: This research was supported by the Zhejiang Provincial Natural Science Foundation (No. LGF21H030004).

Conflicts of Interest: All authors completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-24-1606/coif). All authors report the funding from the Zhejiang Provincial Natural Science Foundation (No. LGF21H030004). The authors have no other conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This retrospective study was approved by the Ethics Committee of Tongde Hospital of Zhejiang Province (No. 2021-040). The other hospitals were informed and agreed to participate in this study. Individual informed consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Hirota S, Tateishi U, Nakamoto Y, Yamamoto H, Sakurai S, Kikuchi H, et al. English version of Japanese Clinical Practice Guidelines 2022 for gastrointestinal stromal tumor (GIST) issued by the Japan Society of Clinical Oncology. Int J Clin Oncol 2024;29:647-80. [Crossref] [PubMed]
2. Serrano C, Martín-Broto J, Asencio-Pascual JM, López-Guerrero JA, Rubió-Casadevall J, Bagué S, García-Del-Muro X, Fernández-Hernández JÁ, Herrero L, López-Pousa A, Poveda A, Martínez-Marín V. 2023 GEIS Guidelines for gastrointestinal stromal tumors. Ther Adv Med Oncol 2023;15:17588359231192388. [Crossref] [PubMed]
3. von Mehren M, Kane JM, Bui MM, Choy E, Connelly M, Dry S, et al. NCCN Guidelines Insights: Soft Tissue Sarcoma, Version 1.2021. J Natl Compr Canc Netw 2020;18:1604-12. [Crossref] [PubMed]
4. Tsuboi K, Yano F, Omura N, Misawa T, Kashiwagi H. Reduced-port surgery with the cowboy technique for a gastric submucosal tumor. Asian J Endosc Surg 2021;14:154-7. [Crossref] [PubMed]
5. Casali PG, Blay JY, Abecassis N, Bajpai J, Bauer S, Biagini R, et al. Gastrointestinal stromal tumours: ESMO-EURACAN-GENTURIS Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol 2022;33:20-33. [Crossref] [PubMed]
6. Serrano C, Álvarez R, Carrasco JA, Marquina G, Martínez-García J, Martínez-Marín V, Sala MÁ, Sebio A, Sevilla I, Martín-Broto J. SEOM-GEIS clinical guideline for gastrointestinal stromal tumors (2022). Clin Transl Oncol 2023;25:2707-17. [Crossref] [PubMed]
7. Zhong Z, Xu Y, Liu J, Zhang C, Xiao Z, Xia Y, Wang Y, Wang J, Xu Q, Lu Y. Clinicopathological study of gastric schwannoma and review of related literature. BMC Surg 2022;22:159. [Crossref] [PubMed]
8. Li J, Ye Y, Wang J, Zhang B, Qin S, Shi Y, He Y, Liang X, Liu X, Zhou Y, Wu X, Zhang X, Wang M, Gao Z, Lin T, Cao H, Shen LChinese Society Of Clinical Oncology Csco Expert Committee On Gastrointestinal Stromal Tumor. Chinese consensus guidelines for diagnosis and management of gastrointestinal stromal tumor. Chin J Cancer Res 2017;29:281-93. [Crossref] [PubMed]
9. Jiang X, Zhao M, Wu J, Ding Y, Wang J. Laparoscopic resection for gastric schwannoma larger than 30 mm with long-term outcomes. BMC Surg 2023;23:284. [Crossref] [PubMed]
10. Wang J, Zhang W, Zhou X, Xu J, Hu HJ. Simple Analysis of the Computed Tomography Features of Gastric Schwannoma. Can Assoc Radiol J 2019;70:246-53. [Crossref] [PubMed]
11. Yue L, Sun Y, Wang X, Hu W. Advances of endoscopic and surgical management in gastrointestinal stromal tumors. Front Surg 2023;10:1092997. [Crossref] [PubMed]
12. Gu X, Li Y, Shi G. Identification of gastric schwannoma and non-metastatic gastric stromal tumor by CT: a single-institution retrospective diagnostic test. J Gastrointest Oncol 2023;14:922-31. [Crossref] [PubMed]
13. He MY, Zhang R, Peng Z, Li Y, Xu L, Jiang M, Li ZP, Feng ST. Differentiation between gastrointestinal schwannomas and gastrointestinal stromal tumors by computed tomography. Oncol Lett 2017;13:3746-52. [Crossref] [PubMed]
14. Choi YR, Kim SH, Kim SA, Shin CI, Kim HJ, Kim SH, Han JK, Choi BI. Differentiation of large (≥5 cm) gastrointestinal stromal tumors from benign subepithelial tumors in the stomach: radiologists' performance using CT. Eur J Radiol 2014;83:250-60. [Crossref] [PubMed]
15. Gillies RJ, Kinahan PE, Hricak H. Radiomics: Images Are More than Pictures, They Are Data. Radiology 2016;278:563-77. [Crossref] [PubMed]
16. Zhang C, Wang C, Mao G, Cheng G, Ji H, He L, Yang Y, Hu H, Wang J. Radiomics analysis of contrast-enhanced computerized tomography for differentiation of gastric schwannomas from gastric gastrointestinal stromal tumors. J Cancer Res Clin Oncol 2024;150:87. [Crossref] [PubMed]
17. Tomaszewski MR, Gillies RJ. The Biological Meaning of Radiomic Features. Radiology 2021;298:505-16. [Crossref] [PubMed]
18. Park SH, Lee HJ, Kim MC, Yook JH, Sohn TS, Hyung WJ, Ryu SW, Kurokawa Y, Kim YW, Han SU, Kim HH, Park DJ, Kim W, Lee SI, Cho H, Cho GS, Kim JJ, Kim KH, Yoo MW, Yang HK. Early experience of laparoscopic resection and comparison with open surgery for gastric gastrointestinal stromal tumor: a multicenter retrospective study. Sci Rep 2022;12:2290. [Crossref] [PubMed]
19. Kikuchi S, Nishizaki M, Kuroda S, Tanabe S, Noma K, Kagawa S, Shirakawa Y, Kato H, Okada H, Fujiwara T. Nonexposure laparoscopic and endoscopic cooperative surgery (closed laparoscopic and endoscopic cooperative surgery) for gastric submucosal tumor. Gastric Cancer 2017;20:553-7. [Crossref] [PubMed]
20. Li B, Chen T, Qi ZP, Yao LQ, Xu MD, Shi Q, Cai SL, Sun D, Zhou PH, Zhong YS. Efficacy and safety of endoscopic resection for small submucosal tumors originating from the muscularis propria layer in the gastric fundus. Surg Endosc 2019;33:2553-61. [Crossref] [PubMed]
21. Hu J, Liu X, Ge N, Wang S, Guo J, Wang G, Sun S. Role of endoscopic ultrasound and endoscopic resection for the treatment of gastric schwannoma. Medicine (Baltimore) 2017;96:e7175. [Crossref] [PubMed]
22. Choi JW, Choi D, Kim KM, Sohn TS, Lee JH, Kim HJ, Lee SJ. Small submucosal tumors of the stomach: differentiation of gastric schwannoma from gastrointestinal stromal tumor with CT. Korean J Radiol 2012;13:425-33. [Crossref] [PubMed]
23. Ji JS, Lu CY, Mao WB, Wang ZF, Xu M. Gastric schwannoma: CT findings and clinicopathologic correlation. Abdom Imaging 2015;40:1164-9. [Crossref] [PubMed]
24. Wang L, Wang Q, Yang L, Ma C, Shi G. Computed tomographic imaging features to differentiate gastric schwannomas from gastrointestinal stromal tumours: a matched case-control study. Sci Rep 2023;13:17568. [Crossref] [PubMed]
25. Kim HC, Lee JM, Kim KW, Park SH, Kim SH, Lee JY, Han JK, Choi BI. Gastrointestinal stromal tumors of the stomach: CT findings and prediction of malignancy. AJR Am J Roentgenol 2004;183:893-8. [Crossref] [PubMed]