Development of a deep learning model for early gastric cancer diagnosis using preoperative computed tomography images

Background

Gastric cancer is a highly prevalent and fatal disease. Accurate differentiation between early gastric cancer (EGC) and advanced gastric cancer (AGC) is essential for personalized treatment. Currently, the diagnostic accuracy of computerized tomography (CT) for gastric cancer staging is insufficient to meet clinical requirements. Many studies rely on manual marking of lesion areas, which is not suitable for clinical diagnosis.

Methods

In this study, we retrospectively collected data from 341 patients with gastric cancer at the First Affiliated Hospital of Wenzhou Medical University. The dataset was randomly divided into a training set (n=273) and a validation set (n=68) using an 8:2 ratio. We developed a two-stage deep learning model that enables fully automated EGC screening based on CT images. In the first stage, an unsupervised domain adaptive segmentation model was employed to automatically segment the stomach on unlabeled portal phase CT images. Subsequently, based on the results of the stomach segmentation model, the image was cropped out of the stomach area and scaled to a uniform size, and then the EGC and AGC classification models were built based on these images. The segmentation accuracy of the model was evaluated using the dice index, while the classification performance was assessed using metrics such as the area under the curve (AUC) of the receiver operating characteristic (ROC), accuracy, sensitivity, specificity, and F1 score.

Results

The segmentation model achieved an average dice accuracy of 0.94 on the hand-segmented validation set. On the training set, the EGC screening model demonstrated an AUC, accuracy, sensitivity, specificity, and F1 score of 0.98, 0.93, 0.92, 0.92, and 0.93, respectively. On the validation set, these metrics were 0.96, 0.92, 0.90, 0.89, and 0.93, respectively. After three rounds of data regrouping, the model consistently achieved an AUC above 0.9 on both the validation set and the validation set.

Conclusion

The results of this study demonstrate that the proposed method can effectively screen for EGC in portal venous CT images. Furthermore, the model exhibits stability and holds promise for future clinical applications.

Research trends in endoscopic applications in early gastric cancer: A bibliometric analysis of studies published from 2012 to 2022

Background

Endoscopy is the optimal method of diagnosing and treating early gastric cancer (EGC), and it is therefore important to keep up with the rapid development of endoscopic applications in EGC. This study utilized bibliometric analysis to describe the development, current research progress, hotspots, and emerging trends in this field.

Methods

We retrieved publications about endoscopic applications in EGC from 2012 to 2022 from Web of Science™ (Clarivate™, Philadelphia, PA, USA) Core Collection (WoSCC). We mainly used CiteSpace (version 6.1.R3) and VOSviewer (version 1.6.18) to perform the collaboration network analysis, co-cited analysis, co-occurrence analysis, cluster analysis, and burst detection.

Results

A total of 1,333 publications were included. Overall, both the number of publications and the average number of citations per document per year increased annually. Among the 52 countries/regions that were included, Japan contributed the most in terms of publications, citations, and H-index, followed by the Republic of Korea and China. The National Cancer Center, based in both Japan and the Republic of Korea, ranked first among institutions in terms of number of publications, citation impact, and the average number of citations. Yong Chan Lee was the most productive author, and Ichiro Oda had the highest citation impact. In terms of cited authors, Gotoda Takuji had both the highest citation impact and the highest centrality. Among journals, Surgical Endoscopy and Other Interventional Techniques had the most publications, and Gastric Cancer had the highest citation impact and H-index. Among all publications and cited references, a paper by Smyth E C et al., followed by one by Gotoda T et al., had the highest citation impact. Using keywords co-occurrence and cluster analysis, 1,652 author keywords were categorized into 26 clusters, and we then divided the clusters into six groups. The largest and newest clusters were endoscopic submucosal dissection and artificial intelligence (AI), respectively.

Conclusions

Over the last decade, research into endoscopic applications in EGC has gradually increased. Japan and the Republic of Korea have contributed the most, but research in this field in China, from an initially low base, is developing at a striking speed. However, a lack of collaboration among countries, institutions, and authors, is common, and this should be addressed in future. The main focus of research in this field (i.e., the largest cluster) is endoscopic submucosal dissection, and the topic at the frontier (i.e., the newest cluster) is AI. Future research should focus on the application of AI in endoscopy, and its implications for the clinical diagnosis and treatment of EGC.

Development and validation of a predictive model combining clinical, radiomics, and deep transfer learning features for lymph node metastasis in early gastric cancer

Background

This study aims to develop and validate a predictive model combining deep transfer learning, radiomics, and clinical features for lymph node metastasis (LNM) in early gastric cancer (EGC).

Materials and methods

This study retrospectively collected 555 patients with EGC, and randomly divided them into two cohorts with a ratio of 7:3 (training cohort, n = 388; internal validation cohort, n = 167). A total of 79 patients with EGC collected from the Second Affiliated Hospital of Soochow University were used as external validation cohort. Pre-trained deep learning networks were used to extract deep transfer learning (DTL) features, and radiomics features were extracted based on hand-crafted features. We employed the Spearman rank correlation test and least absolute shrinkage and selection operator regression for feature selection from the combined features of clinical, radiomics, and DTL features, and then, machine learning classification models including support vector machine, K-nearest neighbor, random decision forests (RF), and XGBoost were trained, and their performance by determining the area under the curve (AUC) were compared.

Results

We constructed eight pre-trained transfer learning networks and extracted DTL features, respectively. The results showed that 1,048 DTL features extracted based on the pre-trained Resnet152 network combined in the predictive model had the best performance in discriminating the LNM status of EGC, with an AUC of 0.901 (95% CI: 0.847–0.956) and 0.915 (95% CI: 0.850–0.981) in the internal validation and external validation cohorts, respectively.

Conclusion

We first utilized comprehensive multidimensional data based on deep transfer learning, radiomics, and clinical features with a good predictive ability for discriminating the LNM status in EGC, which could provide favorable information when choosing therapy options for individuals with EGC.

Risk Factors of Lymph Node Metastasis and Its Prognostic Significance in Early Gastric Cancer: A Multicenter Study

Background

Early gastric cancer (EGC) is invasive gastric cancer that invades no deeper than the submucosa, regardless of lymph node metastasis (LNM). It is mainly treated by surgery. Recently, the resection range of EGC has been minimized, but cancer recurrence and overall survival in some patients should be given high status. LNM is an important indicator of prognosis and treatment in gastric cancer. The law of the number and location of metastatic lymph nodes in EGC is not yet clear. Therefore, we aimed to identify the risk factors of LNM in radically resected EGC and guide treatment.

Methods

The clinicopathological factors of 611 patients with EGC were retrospectively analyzed in six hospitals between January 2010 and December 2016. The relationship between clinicopathological factors and LNM, as well as their prognostic significance, were analyzed by univariate and multivariate analyses.

Results

The rate of LNM was 20.0% in the 611 EGC patients. The depth of invasion, differentiation type, tumor diameter, morphological ulceration, and lymphovascular invasion were independent risk factors for LNM (P<0.05) by logistic regression analysis. Tumor location in the proximal third of the stomach and morphological ulceration were significant factors for group 2 LNM. Moreover, the 5-year survival rate was 94.9% for patients with no positive nodes, 88.5% for patients with 1-2 positive nodes, 64.3% for patients with 3-6 positive nodes, and 41.8% for patients with >6 metastatic nodes. Interestingly, the 7-year risk of relapse diminished for patients with no LNM or retrieved no less than 15 lymph nodes.

Conclusions

Fifteen lymph node dissection and D2 radical operation are the surgical options in case of high risk factors for LNM. Extended lymph node dissection (D2+) is recommended for morphological ulceration or disease located in the proximal third of the stomach due to their high rate of group 2 LNM. Furthermore, LNM is a significant prognostic factor of EGC. Moreover, lymph nodes can also play a significant role in the chemotherapeutic and radiotherapy approach for non-surgical patients with EGC.

Mucosal microbial microenvironment in early gastric neoplasia and non-neoplastic gastric disease

BACKGROUND AND AIM

The biological characterization of microbial environment in early gastric cancer (EGC), other than Helicobacter pylori, is limited. This study aimed to explore the microbial microenvironment in chronic gastritis (CG), fundic gland polyps (FGPs), low-grade intraepithelial neoplasia (LGIN), and EGC.

METHODS

16S-rRNA gene sequencing and bioinformatic analysis were performed on 63 individuals with 252 mucosal biopsies or endoscopic submucosal dissection margin samples from endoscopy.

RESULTS

The microbiota in gastric LGIN functions analogously to EGC in terms of functional prediction. Neoplastic lesions showed a significant difference to CG or FGPs in beta diversity of the microbiota. Bacteria genera including Paracoccus, Blautia, Barnesiella, Lactobacillus, Thauera, Collinsella were significantly enriched in gastric neoplastic mucosa (LGIN and EGC) compared with non-neoplastic tissues (CG and FGPs). While Pseudomonas and Kingella were depleted in neoplastic tissues. FGPs showed a distinctive microbial network system that negatively interacted with Helicobacter.

CONCLUSIONS

In terms of the mucosal microbial microenvironment, gastric LGIN and EGC showed no significant difference as early neoplastic lesions. We observed a coordinated microbial microenvironment that correlated negatively with Helicobacter.

Nomograms for Predicting the Lymph Node Metastasis in Early Gastric Cancer by Gender: A Retrospective Multicentric Study

Background

Lymph node metastasis (LNM) has a significant impact on the prognosis of patients with early gastric cancer (EGC). Our aim was to identify the independent risk factors for LNM and construct nomograms for male and female EGC patients, respectively.

Methods

Clinicopathological data of 1,742 EGC patients who underwent radical gastrectomy and lymphadenectomy in the First Affiliated Hospital, Second Affiliated Hospital, and Fourth Affiliated Hospital of Anhui Medical University between November 2011 and April 2021 were collected and analyzed retrospectively. Male and female patients from the First Affiliated Hospital of Anhui Medical University were assigned to training sets and then from the Second and Fourth Affiliated Hospitals of Anhui Medical University were enrolled in validation sets. Based on independent risk factors for LNM in male and female EGC patients from the training sets, the nomograms were established respectively, which was also verified by internal validation from the training sets and external validation from the validation sets.

Results

Tumor size (odd ratio (OR): 1.386, p = 0.030), depth of invasion (OR: 0.306, p = 0.001), Lauren type (OR: 2.816, p = 0.000), lymphovascular invasion (LVI) (OR: 0.160, p = 0.000), and menopause (OR: 0.296, p = 0.009) were independent risk factors for female EGC patients. For male EGC patients, tumor size (OR: 1.298, p = 0.007), depth of invasion (OR: 0.257, p = 0.000), tumor location (OR: 0.659, p = 0.002), WHO type (OR: 1.419, p = 0.001), Lauren type (OR: 3.099, p = 0.000), and LVI (OR: 0.131, p = 0.000) were independent risk factors. Moreover, nomograms were established to predict the risk of LNM for female and male EGC patients, respectively. The area under the ROC curve of nomograms for female and male training sets were 87.7% (95% confidence interval (CI): 0.8397–0.914) and 94.8% (95% CI: 0.9273–0.9695), respectively. For the validation set, they were 92.4% (95% CI: 0.7979–1) and 93.4% (95% CI: 0.8928–0.9755), respectively. Additionally, the calibration curves showed good agreements between the bias-corrected prediction and the ideal reference line for both training sets and validation sets in female and male EGC patients.

Conclusions

Nomograms based on risk factors for LNM in male and female EGC patients may provide new insights into the selection of appropriate treatment methods.

Minimally invasive approaches for early gastric cancer in East Asia: current status and future perspective

East Asia is well known as a region with higher incidences of gastric cancer compared with the rest of the world. This region has also experienced a rise in the detection of early gastric cancer (EGC) for the past three decades. The success of nationwide screening programs conducted in Japan or South Korea, increasing public awareness of gastric cancer, improved diagnostic ability of gastroenterologists and the aging population of modern societies may all contribute to the increase in EGC detection rates. Along with the increasing diagnosis of EGC, several key advances in the minimally invasive approach to EGC treatment have been made. Endoscopic resection is an ideal procedure for lesions without lymph node involvement, and its indications have expanded based on the results of prospective studies. Laparoscopic surgery with lymph node dissection has been becoming a standard treatment for EGC patients. Additionally, robot-assisted surgery is penetrating the field of gastric cancer surgery as surgeons pursue a more accurate minimally invasive approach with reduced morbidity rates. However, prolonged operation time and high cost remain problems to be solved for robot-assisted surgery. In this context, function-preserving surgery has become ever more important and should be considered as a method to enhance patients' quality of life after a gastrectomy for EGC. Pylorus-preserving gastrectomy or proximal gastrectomy is more frequently employed in East Asia and strategies that employ sentinel node (SN) navigation to personalize function-preserving surgery in clinical practice are emerging as several prospective studies investigate its efficacy.

Endoscopic submucosal dissection versus endoscopic mucosal resection for patients with early gastric cancer: a meta-analysis

OBJECTIVES

To investigate the treatment effects of endoscopic submucosal dissection (ESD) versus endoscopic mucosal resection (EMR) for early gastric cancer (EGC).

DESIGN

Meta-analysis.

METHODS

We systematically searched three electronic databases, including PubMed, EmBase and the Cochrane library for studies published with inception to January 2018. The eligible studies should be evaluated for the efficacy and safety of ESD versus EMR for patients with EGC. The summary ORs and standard mean differences (SMDs) with 95% CIs were employed as effect estimates. Sensitivity analyses were conducted to evaluate the impact of single study on overall analysis. Subgroup analyses were performed for investigated outcomes to evaluate the treatment effects of ESD versus EMR for patients with EGC with specific subsets.

RESULTS

Eighteen studies, with a total of 6723 patients with EGC, were included in final analysis. The summary ORs indicated that patients with EGC who received ESD were associated with an increased incidence of en bloc resection (OR: 9.00; 95% CI: 6.66 to 12.17; p<0.001), complete resection (OR: 8.43; 95% CI: 5.04 to 14.09; p<0.001) and curative resection (OR: 2.92; 95% CI: 1.85 to 4.61; p<0.001) when compared with EMR. Furthermore, ESD was associated with lower risk of local recurrence (OR: 0.18; 95% CI: 0.09 to 0.34; p<0.001). In addition, there was no significant difference between ESD and EMR for the risk of bleeding (OR: 1.26; 95% CI: 0.88 to 1.80; p=0.203). Though, ESD was correlated with greater risk of perforation (OR: 2.55; 95% CI: 1.48 to 4.39; p=0.001), and longer operation time (SMD: 1.12; 95% CI: 0.13 to 2.10; p=0.026) as compared with EMR. Additionally, several different features observed in included studies and patients could bias the effectiveness of ESD versus EMR in patients with EGC.

CONCLUSIONS

ESD is superior than EMR for en bloc resection, complete resection, curative resection and local recurrence, while it increased perforation risk and longer operation time.

Sentinel node navigation in gastric cancer: new horizons for personalized minimally invasive surgical oncology?

Complete (R0) resection and regional lymph nodes (LNs) dissection represent undoubtedly the basic surgical tools for patients with gastric cancer. It is reported that the LN metastasis rate in patients with early gastric cancer (EGC) is approximately 15-20%. Therefore, the innovative clinical application of sentinel node navigation surgery (SNNS) for EGC might be able to prevent unnecessary LN dissection as well as to reduce significantly the volume of gastric resection. Recent evidence suggests that double tracer methods appear superior compared to single tracer techniques. However, the researchers' interest is now focused on the identification of new LN detection methods utilizing sophisticated technology such as infrared ray endoscopy, fluorescence imaging and near-infrared technology. Despite its notable limitations, hematoxylin-eosin is still considered the mainstay staining for assessing the metastatic status of LNs. In this review, we summarize the current evidences and we provide the latest scientific information assessing safety, efficacy and potential limitations of the innovative sentinel node (SN) navigation technique for gastric cancer. We try also to provide a "view" towards a future potential application of personalized minimally invasive surgery in gastric cancer field.

Endoscopic resection of early gastric cancer: current status and new approaches

Endoscopic resection (ER) of early gastric cancer (EGC) has been an optimal treatment for selected patients. As endoscopic submucosal dissection (ESD) has been widely used for treatment of EGC, concerns have been asked to achieve curative resection for EGC while guaranteeing precise prediction of lymph node metastasis (LNM). Moreover, a new microscopic imaging for precise endoscopic diagnosis of EGC is introduced. This review covers the current status and new approaches of ER of EGC.

Outcomes and precautions of endoscopic submucosal dissection for undifferentiated-type early gastric cancer

OBJECTIVES

Since the development of techniques for endoscopic submucosal dissection (ESD), the indication range of endoscopic resection (ER) has been extended in early gastric cancer (EGC) treatment. For undifferentiated-type EGC, tumors with an intramucosal depth of invasion, no ulceration and a diameter of 20 mm or less were included in the expanded indications for ER in the Japanese Gastric Cancer Treatment Guidelines 2010. Nonetheless, because of difficulty in detecting lesions that meet the criteria for ER, the number of endoscopically resected cases of undifferentiated-type EGC is less than that of differentiated-type EGC.

METHODS

We retrospectively investigated the outcomes of ESD in 38 patients with 40 lesions of EGC in which the dominant pathological type was confirmed to be undifferentiated (signet ring cell carcinoma, poorly differentiated adenocarcinoma, mucinous adenocarcinoma) on histological examination of resected specimens.

RESULTS

Margin involvement and submucosal infiltration were common noncurative factors. Precise evaluation of the area and depth of lesions is a problem to be solved. Among a total of five patients with involved or uncertain horizontal margins, one of two patients who underwent additional surgery had residual cancer, and one of three patients who were observed had recurrence.

CONCLUSIONS

Undifferentiated-type EGC with a positive horizontal margin may relapse after ESD. It is therefore essential to precisely evaluate the area of the lesion and to perform resection with an adequate safety margin to decrease the risk of recurrence.

Future perspective of gastric cancer endotherapy

Endoscopic resection of early gastric cancer (EGC) has proven safety and efficacy, and is the established standard of care in Japan. In the past decade, it is increasingly established worldwide. The endoscopic submucosal dissection (ESD) is superior to endoscopic mucosal resection (EMR) technique as it is designed to provide adequate staging and long-term curative therapy-based on the en bloc R0 specimen irrespective of the size and/or location of the tumor coupled with the reliable pathological specimen. However, ESD is still requiring skilled and experienced endoscopist to perform because of complex procedures, higher complication and causing long-time consuming. The learning and application of these relatively complex endoscopic techniques for EGC has been shown across the world. Thus, a standardized ESD training system is urgently needed to disseminate safe and effective ESD technique to practices with limited ESD experience. In recent years, several innovations providing solutions to easier and safer performance of ESD have emerged. Those increase control of surgical effectors manipulating the target tissue, and enhance performance in complex surgical tasks. Very recently, the use of the laparoscopic and endoscopic cooperative surgery (LECS) procedure is indicated for EGC that would be difficult to treat with ESD. As an ultimate gastric cancer endotherapy with a reasonable surgical time, LECS might be promising method at this stage. The indications for LECS for EGC could be expanded in the future, which could result in increasingly successful gastric cancer treatment.

Factors predictive of inaccurate determination of horizontal extent of intestinal-type early gastric cancers during endoscopic submucosal dissection: a retrospective analysis

BACKGROUND

Certain tumor characteristics may pose challenges when endoscopically determining the horizontal extent of early gastric cancers (EGC). In the present study, clinicopathological features related to inaccurate endoscopic evaluation of horizontal extent of intestinal-type EGC were analyzed.

PATIENTS AND METHODS

We analyzed 431 lesions with intestinal-type EGC treated by endoscopic submucosal dissection (ESD) at our hospital. We focused on whether pretreatment demarcation was accurate by comparing positional relationships between marking dots and tumor edges in resected specimens, and factors related to inaccurate evaluation were analyzed. Gender, age, tumor size, location, circumference, depth, ulceration, macroscopic type, presence of a flat (0-IIb) component, predominant histological type, mixture of diffuse-type adenocarcinoma, mixed histology, and use of magnification endoscopy with narrow band imaging were analyzed. Reasons for inaccurate evaluation were also investigated by re-examining endoscopic images and prepared histological slides.

RESULTS

Rate of inaccurate evaluation of horizontal extent was 7.4% (32/431 lesions). Multivariate analysis revealed the following significant independent variables contributing to inaccurate endoscopic evaluation: presence of a flat component, large size, and predominant histological findings of moderately differentiated adenocarcinoma. Re-examination of prepared histological slides of inaccurately evaluated cases revealed a marginal flat spreading area in 28 of the 32 lesions (87.5%). In 14 of the 32 lesions (43.8%), tumor margins were composed of moderately differentiated adenocarcinoma.

CONCLUSIONS

For lesions with a flat component, large lesions, and moderately differentiated adenocarcinoma, determination of the horizontal extent can be challenging in EGC indicated for ESD, even with the best available endoscopic tools.