Esophageal Cancer: Priorities for Prevention

Ensembled CNN with artificial bee colony optimization method for esophageal cancer stage classification using SVM classifier

BACKGROUND

Esophageal cancer (EC) is aggressive cancer with a high fatality rate and a rapid rise of the incidence globally. However, early diagnosis of EC remains a challenging task for clinicians.

OBJECTIVE

To help address and overcome this challenge, this study aims to develop and test a new computer-aided diagnosis (CAD) network that combines several machine learning models and optimization methods to detect EC and classify cancer stages.

METHODS

The study develops a new deep learning network for the classification of the various stages of EC and the premalignant stage, Barrett's Esophagus from endoscopic images. The proposed model uses a multi-convolution neural network (CNN) model combined with Xception, Mobilenetv2, GoogLeNet, and Darknet53 for feature extraction. The extracted features are blended and are then applied on to wrapper based Artificial Bee Colony (ABC) optimization technique to grade the most accurate and relevant attributes. A multi-class support vector machine (SVM) classifies the selected feature set into the various stages. A study dataset involving 523 Barrett's Esophagus images, 217 ESCC images and 288 EAC images is used to train the proposed network and test its classification performance.

RESULTS

The proposed network combining Xception, mobilenetv2, GoogLeNet, and Darknet53 outperforms all the existing methods with an overall classification accuracy of 97.76% using a 3-fold cross-validation method.

CONCLUSION

This study demonstrates that a new deep learning network that combines a multi-CNN model with ABC and a multi-SVM is more efficient than those with individual pre-trained networks for the EC analysis and stage classification.

Study on image data cleaning method of early esophageal cancer based on VGG_NIN neural network

In order to clean the mislabeled images in the esophageal endoscopy image data set, we designed a new neural network VGG_NIN. Based on the new neural network structure, we developed a method to clean the mislabeled images in the esophageal endoscopy image data set. To verify the effectiveness of the proposed method, we designed two experiments using 3835 esophageal endoscopy images provided by West China Hospital of Sichuan University. The experimental results showed that the proposed method could clean about 93% of the mislabeled images in the data set, which was the first time in the cleaning of esophageal endoscopy image data set. Finally, in order to verify the generalization ability of this method, we cleaned the Kaggle open cat and dog data set, and cleaned out about 167 mislabeled images. Therefore, the proposed method can effectively screen the mislabeled images in the esophageal endoscopy image data set and has good generalization ability, which can provide great help for the development of high-performance gastrointestinal endoscopy image analysis model.

Annotating Early Esophageal Cancers Based on Two Saliency Levels of Gastroscopic Images

Early diagnoses of esophageal cancer can greatly improve the survival rate of patients. At present, the lesion annotation of early esophageal cancers (EEC) in gastroscopic images is generally performed by medical personnel in a clinic. To reduce the effect of subjectivity and fatigue in manual annotation, computer-aided annotation is required. However, automated annotation of EEC lesions using images is a challenging task owing to the fine-grained variability in the appearance of EEC lesions. This study modifies the traditional EEC annotation framework and utilizes visual salient information to develop a two saliency levels-based lesion annotation (TSL-BLA) for EEC annotations on gastroscopic images. Unlike existing methods, the proposed framework has a strong ability of constraining false positive outputs. What is more, TSL-BLA is also placed an additional emphasis on the annotation of small EEC lesions. A total of 871 gastroscopic images from 231 patients were used to validate TSL-BLA. 365 of those images contain 434 EEC lesions and 506 images do not contain any lesions. 101 small lesion regions are extracted from the 434 lesions to further validate the performance of TSL-BLA. The experimental results show that the mean detection rate and Dice similarity coefficients of TSL-BLA were 97.24 and 75.15%, respectively. Compared with other state-of-the-art methods, TSL-BLA shows better performance. Moreover, it shows strong superiority when annotating small EEC lesions. It also produces fewer false positive outputs and has a fast running speed. Therefore, The proposed method has good application prospects in aiding clinical EEC diagnoses.

Analysis of salivary microRNA expression profiles and identification of novel biomarkers in esophageal cancer

MicroRNAs (miRNAs/miRs) regulate the expression of target genes and are considered to be associated with human cancer. The aim of the present study was to screen novel miRNA biomarkers in esophageal cancer (EC). The miRNA expression profile GSE41268 was extracted from Gene Expression Omnibus database, and differentially expressed miRNAs between whole saliva samples from patients with EC and healthy controls were identified using the Linear Models for Microarray Data package. Then, the targets of these miRNAs were screened using the miRecords database and used to construct the regulatory network. Gene ontology and pathway enrichment analyses were performed for the target genes of differentially expressed miRNAs to predict their potential functions. A total of 18 differentially expressed miRNAs were identified in saliva samples from patients with EC, and 43 validated target genes corresponding to 7 upregulated miRNAs were identified. Then, 6 miRNAs (miR-144, miR-451, miR-98, miR-10b, miR-486-5p and miR-363) and their target genes were used to construct a regulatory network. Within the network, miR-144 may target Notch homolog 1, fibrinogen α chain and fibrinogen β chain; miR-451 may regulate murine thymoma viral oncogene homolog 1, matrix metalloproteinase (MMP)9 and MMP2; miR-98 may directly target E2F transcription factor (E2F) 1, E2F2 and v-myc avian myelocytomatosis viral oncogene homolog (MYC); miR-10b may modulate peroxisome proliferator-activated receptor α and Kruppel-like factor 4; miR-485-5p and miR-363 may regulate TNF receptor superfamily member 5 and cyclin-dependent kinase inhibitor 1A. In addition, E2F1, E2F2 and MYC were associated with the cell cycle, which was the most significantly enriched function and pathway in EC. The results of the present study suggested that miR-144, miR-451, miR-98, miR-10b and miR-363 may be involved in EC by regulating their target genes, and may be used as biomarkers for EC.

Association between alcohol dehydrogenase-2 gene polymorphism and esophageal cancer risk: a meta-analysis

BACKGROUND

It has been shown that gene polymorphisms may play an important role in the carcinogenesis of esophageal cancer. This study is to investigate the role of alcohol dehydrogenase 1B (ADH1B) gene Arg47His polymorphism in esophageal cancer susceptibility.

METHODS

Case-control studies published between January 2000 and June 2015 were searched to retrieve relevant articles. The pooled odds ratio (OR) and 95 % confidence interval (CI) were employed to calculate the strength of association.

RESULTS

A total of 23 relevant articles were finally selected for the analysis, including 9338 esophageal cancer patients and 14,896 matched controls. Overall, we found that the 47His allele was significant associated with the decreased risk of esophageal cancer when compared with the 47Arg allele in total populations (A vs. G: OR = 0.67, 95 % CI = 0.59-0.76, P < 0.00001). This protective relationship was observed under other genetic models as well (P < 0.00001). Subgroup analysis by ethnicity showed that ADH1B Arg47His variant was associated with the decreased esophageal cancer risk under all the genetic models (P < 0.00001) among Asians, especially in Chinese and Japanese; while in non-Asians, no significant correlation was detected in any genetic models (P > 0.05). Furthermore, Arg/Arg genotype of ADH1B Arg47His variant combined with drinking, smoking and males appeared to show a high risk in patients with esophageal cancer.

CONCLUSIONS

Our results suggested that ADH1B gene Arg47His variant was associated with the decreased esophageal cancer risk. Genetic-environmental interaction should be further considered in the future researches.