Dataset Information

A Novel Deep Learning System for Diagnosing Early Esophageal Squamous Cell Carcinoma: A Multicenter Diagnostic Study.

ABSTRACT:

Introduction

This study aims to construct a real-time deep convolutional neural networks (DCNNs) system to diagnose early esophageal squamous cell carcinoma (ESCC) with white light imaging endoscopy.

Methods

A total of 4,002 images from 1,078 patients were used to train and cross-validate the DCNN model for diagnosing early ESCC. The performance of the model was further tested with independent internal and external validation data sets containing 1,033 images from 243 patients. The performance of the model was then compared with endoscopists. The accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and Cohen kappa coefficient were measured to assess performance.

Results

The DCNN model had excellent performance in diagnosing early ESCC with a sensitivity of 0.979, a specificity of 0.886, a positive predictive value of 0.777, a negative predictive value of 0.991, and an area under curve of 0.954 in the internal validation data set. The model also depicted a tremendously generalized performance in 2 external data sets and exhibited superior performance compared with endoscopists. The performance of the endoscopists was markedly elevated after referring to the predictions of the DCNN model. An open-accessed website of the DCNN system was established to facilitate associated research.

Discussion

A real-time DCNN system, which was constructed to diagnose early ESCC, showed good performance in validation data sets. However, more prospective validation is needed to understand its true clinical significance in the real world.

SUBMITTER: Tang D

PROVIDER: S-EPMC8341371 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:The aim of this study was to develop a diagnostic strategy for esophageal squamous cell carcinoma (ESCC) that combines plasma metabolomics with machine learning algorithms. Plasma-based untargeted metabolomics analysis was performed with samples derived from 88 ESCC patients and 52 healthy controls. The dataset was split into a training set and a test set. After identification of differential metabolites in training set, single-metabolite-based receiver operating characteristic (ROC) curves and multiple-metabolite-based machine learning models were used to distinguish between ESCC patients and healthy controls. Kaplan-Meier survival analysis and Cox proportional hazards regression analysis were performed to investigate the prognostic significance of the plasma metabolites. Finally, twelve differential plasma metabolites (six up-regulated and six down-regulated) were annotated. The predictive performance of the six most prevalent diagnostic metabolites through the diagnostic models in the test set were as follows: arachidonic acid (accuracy: 0.887), sebacic acid (accuracy: 0.867), indoxyl sulfate (accuracy: 0.850), phosphatidylcholine (PC) (14:0/0:0) (accuracy: 0.825), deoxycholic acid (accuracy: 0.773), and trimethylamine N-oxide (accuracy: 0.653). The prediction accuracies of the machine learning models in the test set were partial least-square (accuracy: 0.947), random forest (accuracy: 0.947), gradient boosting machine (accuracy: 0.960), and support vector machine (accuracy: 0.980). Additionally, survival analysis demonstrated that acetoacetic acid was an unfavorable prognostic factor (hazard ratio (HR): 1.752), while PC (14:0/0:0) (HR: 0.577) was a favorable prognostic factor for ESCC. This study devised an innovative strategy for ESCC diagnosis by combining plasma metabolomics with machine learning algorithms and revealed its potential to become a novel screening test for ESCC. Highlights • Six most prevalent diagnostic plasma metabolites were identified in ESCC.• Plasma-metabolite-based machine learning models (PLS, RF, GBM, and SVM) for ESCC diagnosis.• Acetoacetic acid was an unfavorable prognostic factor, while PC (14:0/0:0) was a favorable prognostic factor for ESCC.

Project description:Background Deep learning has presented considerable potential and is gaining more importance in computer assisted diagnosis. As the gold standard for pathologically diagnosing cervical intraepithelial lesions and invasive cervical cancer, colposcopy-guided biopsy faces challenges in improving accuracy and efficiency worldwide, especially in developing countries. To ease the heavy burden of cervical cancer screening, it is urgent to establish a scientific, accurate and efficient method for assisting diagnosis and biopsy. Methods The data were collected to establish three deep-learning-based models. For every case, one saline image, one acetic image, one iodine image and the corresponding clinical information, including age, the results of human papillomavirus testing and cytology, type of transformation zone, and pathologic diagnosis, were collected. The dataset was proportionally divided into three subsets including the training set, the test set and the validation set, at a ratio of 8:1:1. The validation set was used to evaluate model performance. After model establishment, an independent dataset of high-definition images was collected to further evaluate the model performance. In addition, the comparison of diagnostic accuracy between colposcopists and models weas performed. Results The sensitivity, specificity and accuracy of the classification model to differentiate negative cases from positive cases were 85.38%, 82.62% and 84.10% respectively, with an AUC of 0.93. The recall and DICE of the segmentation model to segment suspicious lesions in acetic images were 84.73% and 61.64%, with an average accuracy of 95.59%. Furthermore, 84.67% of high-grade lesions were detected by the acetic detection model. Compared to colposcopists, the diagnostic system performed better in ordinary colposcopy images but slightly unsatisfactory in high-definition images. Implications The deep learning-based diagnostic system could help assist colposcopy diagnosis and biopsy for HSILs.

Project description:BACKGROUND:After neoadjuvant chemoradiotherapy (nCRT) for esophageal cancer, high pathologically complete response (pCR) rates are being achieved especially in patients with squamous cell carcinoma (SCC). An active surveillance strategy has been proposed for SCC patients with clinically complete response (cCR) after nCRT. To justify omitting surgical resection, patients with residual disease should be accurately identified. The aim of this study is to assess the accuracy of response evaluations after nCRT based on the preSANO trial, including positron emission tomography with computed tomography (PET-CT), endoscopy with bite-on-bite biopsies and endoscopic ultrasonography (EUS) with fine-needle aspiration (FNA) in patients with potentially curable esophageal SCC. METHODS:Operable esophageal SCC patients who are planned to undergo nCRT according to the CROSS regimen and are planned to undergo surgery will be recruited from four Asian centers. Four to 6 weeks after completion of nCRT, patients will undergo a first clinical response evaluation (CRE-1) consisting of endoscopy with bite-on-bite biopsies. In patients without histological evidence of residual tumor (i.e. without positive biopsies), surgery will be postponed another 6 weeks. A second clinical response evaluation (CRE-2) will be performed 10-12?weeks after completion of nCRT, consisting of PET-CT, endoscopy with bite-on-bite biopsies and EUS with FNA. Immediately after CRE-2 all patients without evidence of distant metastases will undergo esophagectomy. Results of CRE-1 and CRE-2 as well as results of the three single diagnostic modalities will be correlated to pathological response in the resection specimen (gold standard) for calculation of sensitivity, specificity, negative predictive value and positive predictive value. DISCUSSION:If the current study shows that major locoregional residual disease (>?10% residual carcinoma or any residual nodal disease) can be accurately (i.e. with sensitivity of 80.5%) detected in patients with esophageal SCC, a prospective trial will be conducted comparing active surveillance with standard esophagectomy in patients with a clinically complete response after nCRT (SINO trial). TRIAL REGISTRATION:The preSINO trial has been registered at ClinicalTrials.gov as NCT03937362 (May 3, 2019).