Classification Of Gastric Cancer Research Articles

MIEN1 on the 17q12 amplicon facilitates the malignant behaviors of gastric cancer via activating IL-6/JAK2/STAT3 pathway

Oncogene amplification is a significant factor contributing to poor prognosis and limited treatment in patients with advanced gastric cancer. Therefore, identifying amplified oncogenes and elucidating their oncogenic mechanisms will provide reliable therapeutic targets for the clinical treatment of gastric cancer. In this study, we identify a high amplification of 17q12, which includes five oncogenes that are co-amplified and co-overexpressed with ERBB2 using array comparative genomic hybridization, with migration and invasion enhancer 1 (MIEN1) being particularly highlighted for its clinical significance, function, and role in gastric cancer progression. By detecting MIEN1 copy number and expression level across eight gastric cancer cell lines and in tissue microarrays from 543 primary gastric cancer tissues, we found that MIEN1 amplification and overexpression correlated with sex and Lauren’s intestinal type classification of gastric cancer. Besides that, elevated MIEN1 expression was associated with poorer patient survival. In vitro experiments have shown that MIEN1 overexpression enhanced cell proliferation, invasion, and migration, whereas MIEN1 knockdown reversed these malignant phenotypes in vitro. Furthermore, MIEN1 knockdown inhibited tumorigenesis and metastasis of gastric cancer cells in nude mice. Mechanistically, MIEN1 activates the IL-6/JAK2/STAT3 signaling pathway, which drives the proliferation, invasion, and migration of gastric cancer cells. This study demonstrates that MIEN1 contributes to the malignant behavior of gastric cancer through the IL-6/JAK2/STAT3 pathway, suggesting that MIEN1 could serve as a valuable therapeutic target for gastric cancer.

Enhanced gastric cancer classification and quantification interpretable framework using digital histopathology images

Recent developments have highlighted the critical role that computer-aided diagnosis (CAD) systems play in analyzing whole-slide digital histopathology images for detecting gastric cancer (GC). We present a novel framework for gastric histology classification and segmentation (GHCS) that offers modest yet meaningful improvements over existing CAD models for GC classification and segmentation. Our methodology achieves marginal improvements over conventional deep learning (DL) and machine learning (ML) models by adaptively focusing on pertinent characteristics of images. This contributes significantly to our study, highlighting that the proposed model, which performs well on normalized images, is robust in certain respects, particularly in handling variability and generalizing to different datasets. We anticipate that this robustness will lead to better results across various datasets. An expectation-maximizing Naïve Bayes classifier that uses an updated Gaussian Mixture Model is at the heart of the suggested GHCS framework. The effectiveness of our classifier is demonstrated by experimental validation on two publicly available datasets, which produced exceptional classification accuracies of 98.87% and 97.28% on validation sets and 98.47% and 97.31% on test sets. Our framework shows a slight but consistent improvement over previously existing techniques in gastric histopathology image classification tasks, as demonstrated by comparative analysis. This may be attributed to its ability to capture critical features of gastric histopathology images better. Furthermore, using an improved Fuzzy c-means method, our study produces good results in GC histopathology picture segmentation, outperforming state-of-the-art segmentation models with a Dice coefficient of 65.21% and a Jaccard index of 60.24%. The model’s interpretability is complemented by Grad-CAM visualizations, which help understand the decision-making process and increase the model’s trustworthiness for end-users, especially clinicians.

Differentially expressed microRNA in prognosis of gastric cancer with Lauren classification.

Gastric cancer (GC) is one of the most common tumors. There were several classifications of GC recently. The value of Lauren classification in evaluating the prognosis after radical gastrectomy was still unclear and the prognosis of gastric cancer remained relatively poor in the absence of prognostic biomarkers. This study aimed to explore microRNA (miRNA) in the prognosis of GC with different Lauren classification. A retrospective study of 1144patients was performed in this study. Quantificational reverse transcription-PCR (qRT-PCR) was used to examine the expression of miRNAs. Univariate and multivariate analysis were performed to evaluate prognosis value of Lauren classification. Total 1144 GC patients were recruited in this cohort, including 302 diffuse type (26.4%), 436 intestinal type (38.1%) and 406 mixed type (35.5%) GC. Multivariate analysis showed that Lauren classification, patients' age, tumor size, tumor infiltrating depth, vascular nerve infiltrating and metastatic lymph nodes ration were significantly correlated with GC patients' OS and DFS. The miR-141-3p, miR-200b-3p and miR-133a-5p were significantly down-regulated in diffuse type compared to intestinal type GC tissues, the miR-105-5p had significant lower expression in diffuse type compared with intestinal type and mixed type GC tissues. As a consequence of univariate analysis, low miR-141-3p in diffuse type GC showed significant worse OS and DFS than high miR-141-3p. Lauren classification was an independent prognostic factor in GC. MiR-141-3p was an independent prognostic factor and a promising prognostic biomarker in Lauren classification GC.

Chicken swarm algorithm with deep convolutional neural network based tongue image analysis for gastric cancer classification

Tongue images (the size, shape, and colour of tongue and the thickness, colour, and moisture content of tongue coating), reflecting the medical condition of entire body based on the model of traditional Chinese medicine (TCM) are extremely utilized in China for millions of years. Gastric cancer (GC) is great lethal kind of cancer in countries and societies. The screening and analysis of GC yet depend on gastroscopy, however its application was significantly restricted due to its invasive, maximum rate and the requirement for expert endoscopists. Early recognition in GC patients and direct treatment contribute significantly to safety for health. Consequently, this study introduces a Chicken Swarm Algorithm with Deep learningbased Tongue Image Analysis for Gastric Cancer Classification (CSADL-TIAGCC) system. The projected CSADL-TIAGCC approach studies the input tongue images for the identification and classification of GC. To accomplish this, the CSADL-TIAGCC system uses improved U-Net segmentation approach. Besides, residual network (ResNet-34) model-based feature extractor is used. Furthermore, long short term memory (LSTM) approach was exploited for GC classification and its hyperparameters are selected by the CSA. The simulation outcome of the CSADL-TIAGCC algorithm was examined under tongue image database. The experimental outcomes illustrate the enhanced results of the CSADL-TIAGCC technique with respect of different evaluation measures.

Advanced esophagogastric junction mixed neuroendocrine–non-neuroendocrine neoplasm with long-term recurrence-free survival

BackgroundMixed neuroendocrine–non-neuroendocrine neoplasm (MiNEN) is a rare malignant gastrointestinal tumor. The prognosis of patients with MiNEN is poor because of the high frequency of recurrence and metastases. We report a case of esophagogastric junction MiNEN (EGJ-MiNEN) with a long-term recurrence-free survival of 5.5 years.Case presentationA 58-year-old male patient underwent thoracoscopic esophagectomy for esophagogastric junction adenocarcinoma. The patient’s postoperative course was uneventful. R0 resection was achieved, and the pathological diagnosis of the surgical specimen was pT3N2M0 Stage IIIA (according to the Japanese Classification of Gastric Cancer, 4th edition). Based on the pathology results, the patient was treated with postoperative adjuvant therapy with oral S-1. The patient maintained recurrence-free survival for 5.5 years postoperatively. However, 6 years postoperatively, the patient visited our department with cachexia. Computed tomography (CT) revealed a large amount of ascites and pleural effusion. He rapidly developed a poor circulatory and respiratory status and died 16 days after admission. An autopsy revealed severe bloody ascites and pleural effusion, as well as numerous nodules on the pleura and mesentery. Immunohistochemistry of the nodules revealed positivity for chromogranin A, Synaptophysin A, neural cell adhesion molecule (NCAM or CD56), and insulinoma-associated protein 1 (INSM1). The specimen showed a mixture of adenocarcinoma and neuroendocrine cell carcinoma and was diagnosed as MiNEN. Retrospective immunostaining of the surgical specimen showed similar results, and we diagnosed the patient with recurrence of EGJ-MiNEN.ConclusionMiNEN has a poor prognosis; however, in some cases, long-term recurrence-free survival is achieved with radical resection and adjuvant chemotherapy.

Subtyping of gastric cancer based on basement membrane genes that stratifies the prognosis, immune infiltration and therapeutic response

Gastric cancer (GC) is highly heterogeneous and prone to metastasis, which are obstacles to the effectiveness of treatment. The basement membrane (BM) acts as a barrier to tumor cell invasion and metastasis. It is critical to investigate the relationship between BM status, metastasis, and patient prognosis. In several large cohorts, we investigated BM gene expression-based molecular classification and risk-prognosis models for GC, examined tumor microenvironment (TME) differences among different molecular subtypes, and developed risk models in predicting prognosis, immunotherapy effectiveness, and chemotherapy resistance. Three GC subtypes (BMclusterA/B/C) based on BM gene expression status were discovered. Each of the three GC subtypes has unique immune infiltration and activated oncogenic signals. Moreover, a 6-gene score (BMscore) predictive model was developed. The low BMscore group had a high tumor mutation burden, high immunogenicity, and low RHOJ expression levels, implying that individuals with GC in this category may be more susceptible to immunotherapy and treatment. The EMT subtype showed a considerably higher BMscore than the other subtypes in the Asian Organization for Research on Cancer (ACRG) molecular classification. Endothelial cells, smooth muscle cells, and fibroblasts may be engaged in regulating BM reorganization in GC progression, according to single-cell transcriptome analyses. In conclusion, we defined a novel molecular classification of GC based on BM genes, developed a prognostic risk model, and elucidated the cell subpopulations involved in BM remodeling at the single-cell level. This study has deepened the understanding of the relationship between GC metastasis and BM alterations, achieved prognostic stratification, and guided therapy.

Chromosomal copy number based stratification of gastric cancer has added prognostic value to Lauren’s histological classification

BackgroundThe Cancer Genome Atlas (TCGA) recognizes four molecular subgroups of gastric cancer: Epstein-Barr virus (EBV) positive, microsatellite instable (MSI), genomically stable (GS), and chromosomal instable (CIN). Since a GS/CIN classifier is lacking, alternative markers such as Lauren’s histopathology or CDH1/p53 immunohistochemistry are commonly applied. Here we compared survival of gastric cancer subgroups determined by four methods.Methods309 EBV negative and microsatellite stable tumors were included from the Dutch D1/D2 trial and assigned to subgroups by: (i) TCGA’s specific chromosomal copy number aberrations, (ii) genome instability index (GII), (iii) Lauren’s classification, and (iv) CDH1/p53 immunohistochemistry. Subgroups were associated with cancer-related survival (CRS).ResultsFive-year CRS was 42.0% for diffuse and 49.5% for patients with intestinal type tumors, and 57.8% for GS and 41.6% for patients with CIN tumors. Classification by GII or CDH1/p53 IHC did not correlate with CRS. The combination of TCGA and Lauren classifications resulted in four distinct subgroups. Five-year CRS for GS-intestinal (n = 24), GS-diffuse (n = 57), CIN-intestinal (n = 142) and CIN-diffuse (n = 86) was 61.4%, 56.5%, 47.6%, and 31.5%, respectively.ConclusionsTCGA’s GS and CIN subgroups have additional prognostic value to Lauren’s classification in resectable gastric cancer. GS-intestinal, GS-diffuse, CIN-intestinal and CIN-diffuse are suggested stratification variables for future studies.

Current development of molecular classifications of gastric cancer based on omics (Review).

Gastric cancer (GC) is a complex and heterogeneous disease with significant phenotypic and genetic variation. Traditional classification systems rely mainly on the evaluation of clinical pathological features and conventional biomarkers and might not capture the diverse clinical processes of individual GCs. The latest discoveries in omics technologies such as next‑generation sequencing, proteomics and metabolomics have provided crucial insights into potential genetic alterations and biological events in GC. Clustering strategies for identifying subtypes of GC might offer new tools for improving GC treatment and clinical trial outcomes by enabling the development of therapies tailored to specific subtypes. However, the feasibility and therapeutic significance of implementing molecular classifications of GC in clinical practice need to addressed. The present review examines the current molecular classifications, delineates the prevailing landscape of clinically relevant molecular features, analyzes their correlations with traditional GC classifications, and discusses potential clinical applications.

Emerging Therapeutic Targets and Future Directions in Advanced Gastric Cancer: A Comprehensive Review.

Metastatic gastric cancer (GC) still represents a critical clinical challenge, with limited treatment options and a poor prognosis. Most patients are diagnosed at advanced stages, limiting the chances of surgery and cure. The identification of molecular targets and the possibility of combining immune checkpoint inhibitors with chemotherapy have recently reshaped the therapeutic landscape of metastatic gastric cancer. The new classification of gastric cancer, mainly based on immunologic and molecular criteria such as programmed cell death 1 (PD-1), microsatellite instability (MSI), and human epidermal growth factor receptor 2 (HER2), has made it possible to identify and differentiate patients who may benefit from immunotherapy, targeted therapy, or chemotherapy alone. All relevant and available molecular and immunological targets in clinical practice for the systemic treatment of this disease are presented. Particular attention is given to possible future approaches, including circulating tumor DNA (ctDNA) for therapeutic monitoring, new targeting agents against molecular pathways such as fibroblast growth factor receptor (FGFR) and MET, chimeric antigen receptor (CAR)-T cells, and cancer vaccines. This review aims to provide a comprehensive understanding of current targets in advanced gastric cancer and to offer valuable insights into future directions of research and clinical practice in this challenging disease.

Implementing an integrated molecular classification for gastric cancer from endoscopic biopsies using on-slide tests.

The availability of more effective biological therapy can improve outcomes of gastric cancer (GC), but most patients do not have access to personalized treatment. GC molecular classification helps identify patients suitable for specific therapies and provides useful prognostic information. To date, only a small number of patients have access to molecular classification. We proposed a working molecular classification that can be delivered using on-slide tests available in most histopathology laboratories. We used eight on-slide tests [in situ hybridization (ISH) for Epstein-Barr virus-encoded small ribonucleic acid (EBER) and immunohistochemistry (IHC) for MutL homolog 1 (MLH1), PMS1 homolog 2 (PMS2), MutS homolog 2 (MSH2), MutS homolog 6 (MSH6), E-cadherin, β-catenin and p53] to classify GC into one of six categories: GC associated with Epstein-Barr virus (GC-EBV), GC mismatch repair deficient (GC-dMMR), GC with epithelial-mesenchymal transition (GC-EMT), GC with chromosomal instability (GC-CIN), GC genomically stable (GC-GS) and GC not otherwise specified (GC-NOS)∕indeterminate. The classification has provision also for current and future on-slide companion diagnostic (CDx) tests necessary to select specific biological therapies and, as proof of principle, in this study we used three CDx tests currently required for the management of GC [human epidermal growth factor receptor 2 (Her2), programmed cell death-ligand 1 (PD-L1) 22C3 and Claudin18.2 (CLDN18.2)]. This paper describes the necessary tissue pathways and laboratory workflow and assesses the feasibility of using this classification prospectively on small endoscopic biopsies of gastric and gastroesophageal junction adenocarcinoma. This work demonstrates that such molecular classification can be implemented in the context of a histopathology diagnostic routine with little impact on turnaround times and laboratory capacity. The widespread adoption of a molecular classification for GC will help refine prognosis and guide the choice of more appropriate biological therapy for these patients.

Machine Learning Driven Feature Extraction and Dimensionality Reduction for Image Classification

Cancer is the leading cause of death globally, affecting various organs in the human body. Early diagnosis of gastric cancer is essential for improving survival rates. However, traditional diagnosis methods are time-consuming, require multiple tests, and rely on specialist availability. This motivates the development of automated techniques for diagnosing gastric cancer using image analysis. While existing computerized techniques have been proposed, challenges remain. These include difficulty distinguishing healthy from cancerous regions in images and extracting irrelevant features during analysis. This research addresses these challenges by proposing a novel deep learning-based method for gastric cancer classification. The method utilizes deep feature extraction, dimensionality reduction, and classification techniques applied to a gastric cancer image dataset. This approach achieves high accuracy (99.32%), sensitivity (99.13%), and specificity (99.64%) in classifying gastric cancer.

A pathway for detection of gastric cancer biomarkers via using a layer-by-layer coated D-shaped grinding long-period fiber grating sensor

Gastric cancer significantly contributes to global cancer mortality, often leading to inoperable stages and high recurrence rates post-surgery. Elevated levels of G-17 and G-gly have been identified as potential risk factors, particularly in patients with duodenal ulcers. This study introduces an innovative D-shaped grinding long-period fiber grating sensor (D-LLPFGs) designed for non-invasive detection of the gastrin G-17 antigen, employing a layer-by-layer chemical self-assembly to bond G-17 antibodies onto the fiber surface through hydrogen bonding. The D-LLPFGs sensor demonstrated significant spectral shifts within 1 min of antigen-antibody interaction, highlighting its rapid detection capability. At an optimized antibody concentration of 4 μg/ml, antigen testing across different concentrations (10, 12.5, 20, 50 μg/ml) showed peak changes at 12.5 μg/ml antigen, with a 1.186 nm shift and 0.503 dB loss. The sensor exhibited a wavelength sensitivity of 0.095 nm/μg/ml, indicating its high sensitivity and potential in gastric cancer classification, diagnosis, and treatment. This research concludes that the D-shaped fiber sensor is an effective and sensitive tool for detecting G-17 antigen levels, presenting a significant advancement in non-invasive gastric cancer diagnosis. Its quick response time and high sensitivity highlight its potential for broad biomedical applications, offering a new avenue for early cancer detection and improving patient prognosis. The success of this study opens the door to further exploration and implementation of fiber optic sensors in clinical settings, marking a significant step forward in the fight against gastric cancer.

Molecular classifications of gastric cancer and their clinical potential

Gastric cancer if the 5th most common oncological disease in the world and one of the leading causes of death associated with this pathology. In clinical practice, the Lauren classification is widely used for gastric cancer characterization, but it does not provide accurate information on tumor progression and does not allow to select the optimal therapeutic approach. More modern tumor typologies, for example proposed by the The Cancer Genome Atlas (TCGA) and the Asian Cancer Research Group (ACRG), are based on profiling of molecular changes in the tumor genome. Currently, several new classifications exist dividing gastric cancer into groups depending on response to different treatment, for example, checkpoint inhibitors or therapy based on activity of pathological pathways associated with immunity, DNA repair, oncogenic and stromal signatures. The proposed typologies improve diagnosis and treatment of this pathology. The review describes currently available classifications of gastric tumors and considers their practical potential.

Fetal gut cell-like differentiation in esophageal adenocarcinoma defines a rare tumor subtype with therapeutically relevant claudin-6 positivity and SWI/SNF gene alteration

Esophageal adenocarcinoma (EAC) is one of the deadliest tumor entities worldwide, with a 5-year survival rate of less than 25%. Unlike other tumor entities, personalized therapy options are rare, partly due to the lack of knowledge about specific subgroups. In this publication, we demonstrate a subgroup of patients with EAC in a large screening cohort of 826 patients, characterized by specific morphological and immunohistochemical features. This subgroup represents approximately 0.7% (6/826) of the total cohort. Morphological features of this subgroup show a striking clear cytoplasm of the tumour cells and the parallel existence of rare growth patterns like yolk sac-like differentiation and enteroblastic differentiation. Immunohistochemistry reveals expression of the fetal gut cell-like proteins Sal-like protein 4 (SALL4), claudin-6, and glypican 3. Interestingly, we find a correlation with alterations of SWI/SNF-complex associated genes, which are supposed to serve as tumor suppressor genes in various tumour entities. Our results suggest a possible implication of rare tumour subtypes in the WHO classification for EACs according to the classification for gastric cancer. Furthermore, claudin-6 positive tumors have shown promising efficacy of CAR T cell therapy in the recently published BNT-211-01 trial (NCT04503278). This represents a personalized therapeutic option for this tumor subtype.

Deep Learning Radiomics Model of Contrast-Enhanced CT for Differentiating the Primary Source of Liver Metastases

Rationale and ObjectivesTo develop and validate a deep learning radiomics (DLR) model based on contrast-enhanced computed tomography (CT) to identify the primary source of liver metastases. Materials and MethodsIn total, 657 liver metastatic lesions, including breast cancer (BC), lung cancer (LC), colorectal cancer (CRC), gastric cancer (GC), and pancreatic cancer (PC), from 428 patients were collected at three clinical centers from January 2018 to October 2023 series. The lesions were randomly assigned to the training and validation sets in a 7:3 ratio. An additional 112 lesions from 61 patients at another clinical center served as an external test set. A DLR model based on contrast-enhanced CT of the liver was developed to distinguish the five pathological types of liver metastases. Stepwise classification was performed to improve the classification efficiency of the model. Lesions were first classified as digestive tract cancer (DTC) and non-digestive tract cancer (non-DTC). DTCs were divided into CRC, GC, and PC and non-DTCs were divided into LC and BC. To verify the feasibility of the DLR model, we trained classical machine learning (ML) models as comparison models. Model performance was evaluated using accuracy (ACC) and area under the receiver operating characteristic curve (AUC). ResultsThe classification model constructed by the DLR algorithm showed excellent performance in the classification task compared with ML models. Among the five categories task, highest ACC and average AUC were achieved at 0.563 and 0.796 in the validation set, respectively. In the DTC and non-DTC and the LC and BC classification tasks, AUC was achieved at 0.907 and 0.809 and ACC was achieved at 0.843 and 0.772, respectively. In the CRC, GC, and PC classification task, ACC and average AUC were the highest, at 0.714 and 0.811, respectively. ConclusionThe DLR model is an effective method for identifying the primary source of liver metastases.

Comparable prognosis of early gastric cancer between intestinal type and diffuse type in patients of age 75 and older: a SEER-based cohort study.

The prognostic significance of Lauren's classification in elderly early gastric cancer (EGC) patients remains largely unknown. We aim to investigate the characteristics and clinical implications of Lauren's classification in elderly EGC patients. Patients were collected from the Surveillance, Epidemiology, and End Results (SEER) database based on the inclusion and exclusion criteria. Univariate and multivariate Cox regression, propensity score matching, inverse-probability-weighted analysis, and propensity-score adjustment were utilized to evaluate the association between Lauren's classification and cancer-specific survival (CSS) in elderly EGC patients. Stratification and interaction analyses were used to reveal the effects of confounding factors on the association between Lauren's classification and CSS. The diffuse type (median, 41.0 months) showed a similar survival (37.0 months), and was mainly distributed in female group (62.5% vs. 42.2%) with poorly differentiated or undifferentiated components (89.1% vs. 27.0%) compared with intestinal type in elderly EGC patients. Analyses of univariate and multivariate Cox regression, propensity score matching, inverse-probability-weighted analysis, and propensity-score adjustment showed that Lauren's classification was not significantly CSS in elderly EGC patients (P>0.05). Subgroup and interaction analyses confirmed the stability of the results. Diffuse type was mainly distributed in female patients with more poorly differentiated/undifferentiated components and similar prognosis compared with intestinal type in age 75 and older EGC patients. No significant association was observed between diffuse type and CSS of the elderly EGC patients.

Gastro-BaseNet: A Specialized Pre-Trained Model for Enhanced Gastroscopic Data Classification and Diagnosis of Gastric Cancer and Ulcer.

Most of the development of gastric disease prediction models has utilized pre-trained models from natural data, such as ImageNet, which lack knowledge of medical domains. This study proposes Gastro-BaseNet, a classification model trained using gastroscopic image data for abnormal gastric lesions. To prove performance, we compared transfer-learning based on two pre-trained models (Gastro-BaseNet and ImageNet) and two training methods (freeze and fine-tune modes). The effectiveness was verified in terms of classification at the image-level and patient-level, as well as the localization performance of lesions. The development of Gastro-BaseNet had demonstrated superior transfer learning performance compared to random weight settings in ImageNet. When developing a model for predicting the diagnosis of gastric cancer and gastric ulcers, the transfer-learned model based on Gastro-BaseNet outperformed that based on ImageNet. Furthermore, the model's performance was highest when fine-tuning the entire layer in the fine-tune mode. Additionally, the trained model was based on Gastro-BaseNet, which showed higher localization performance, which confirmed its accurate detection and classification of lesions in specific locations. This study represents a notable advancement in the development of image analysis models within the medical field, resulting in improved diagnostic predictive accuracy and aiding in making more informed clinical decisions in gastrointestinal endoscopy.

Classification and localization of gastric cancer using Multi Information Fusion Network

Classification and localization of gastric cancer using Multi Information Fusion Network

Analysis of Colorectal and Gastric Cancer Classification: A Mathematical Insight Utilizing Traditional Machine Learning Classifiers

Cancer remains a formidable global health challenge, claiming millions of lives annually. Timely and accurate cancer diagnosis is imperative. While numerous reviews have explored cancer classification using machine learning and deep learning techniques, scant literature focuses on traditional ML methods. In this manuscript, we undertake a comprehensive review of colorectal and gastric cancer detection specifically employing traditional ML classifiers. This review emphasizes the mathematical underpinnings of cancer detection, encompassing preprocessing techniques, feature extraction, machine learning classifiers, and performance assessment metrics. We provide mathematical formulations for these key components. Our analysis is limited to peer-reviewed articles published between 2017 and 2023, exclusively considering medical imaging datasets. Benchmark and publicly available imaging datasets for colorectal and gastric cancers are presented. This review synthesizes findings from 20 articles on colorectal cancer and 16 on gastric cancer, culminating in a total of 36 research articles. A significant focus is placed on mathematical formulations for commonly used preprocessing techniques, features, ML classifiers, and assessment metrics. Crucially, we introduce our optimized methodology for the detection of both colorectal and gastric cancers. Our performance metrics analysis reveals remarkable results: 100% accuracy in both cancer types, but with the lowest sensitivity recorded at 43.1% for gastric cancer.

Deep Learning and Gastric Cancer: Systematic Review of AI-Assisted Endoscopy.

Gastric cancer (GC), a significant health burden worldwide, is typically diagnosed in the advanced stages due to its non-specific symptoms and complex morphological features. Deep learning (DL) has shown potential for improving and standardizing early GC detection. This systematic review aims to evaluate the current status of DL in pre-malignant, early-stage, and gastric neoplasia analysis. A comprehensive literature search was conducted in PubMed/MEDLINE for original studies implementing DL algorithms for gastric neoplasia detection using endoscopic images. We adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The focus was on studies providing quantitative diagnostic performance measures and those comparing AI performance with human endoscopists. Our review encompasses 42 studies that utilize a variety of DL techniques. The findings demonstrate the utility of DL in GC classification, detection, tumor invasion depth assessment, cancer margin delineation, lesion segmentation, and detection of early-stage and pre-malignant lesions. Notably, DL models frequently matched or outperformed human endoscopists in diagnostic accuracy. However, heterogeneity in DL algorithms, imaging techniques, and study designs precluded a definitive conclusion about the best algorithmic approach. The promise of artificial intelligence in improving and standardizing gastric neoplasia detection, diagnosis, and segmentation is significant. This review is limited by predominantly single-center studies and undisclosed datasets used in AI training, impacting generalizability and demographic representation. Further, retrospective algorithm training may not reflect actual clinical performance, and a lack of model details hinders replication efforts. More research is needed to substantiate these findings, including larger-scale multi-center studies, prospective clinical trials, and comprehensive technical reporting of DL algorithms and datasets, particularly regarding the heterogeneity in DL algorithms and study designs.