Multi-stage domain-specific pretraining for improved detection and localization of Barrett's neoplasia: A comprehensive clinically validated study

Putten, Joost van der; Groof, Jeroen de; Struyvenberg, Maarten R.; Boers, Tim; Fockens, Kiki; Curvers, Wouter L.; Schoon, Erik J.; Bergman, Jacques J.G.H.M.; Sommen, Fons van der; With, Peter H. N. de

doi:10.1016/j.artmed.2020.101914

Cited by 16 publications

(18 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the AI model, we used the decoder of a CNN architecture based on ResNet, which has been used in previous oesophageal CAD work [ 24 ]. The model was pretrained on the ImageNet dataset [ 25 ], and then trained and tested using data that were extracted from endocytoscopic BE videos ( Figure 3 B).…”

Section: Methodsmentioning

confidence: 99%

Optical Biopsy of Dysplasia in Barrett’s Oesophagus Assisted by Artificial Intelligence

Laan

Putten

Zhao

et al. 2023

Cancers

Self Cite

View full text Add to dashboard Cite

Optical biopsy in Barrett’s oesophagus (BE) using endocytoscopy (EC) could optimize endoscopic screening. However, the identification of dysplasia is challenging due to the complex interpretation of the highly detailed images. Therefore, we assessed whether using artificial intelligence (AI) as second assessor could help gastroenterologists in interpreting endocytoscopic BE images. First, we prospectively videotaped 52 BE patients with EC. Then we trained and tested the AI pm distinct datasets drawn from 83,277 frames, developed an endocytoscopic BE classification system, and designed online training and testing modules. We invited two successive cohorts for these online modules: 10 endoscopists to validate the classification system and 12 gastroenterologists to evaluate AI as second assessor by providing six of them with the option to request AI assistance. Training the endoscopists in the classification system established an improved sensitivity of 90.0% (+32.67%, p < 0.001) and an accuracy of 77.67% (+13.0%, p = 0.020) compared with the baseline. However, these values deteriorated at follow-up (−16.67%, p < 0.001 and -8.0%, p = 0.009). Contrastingly, AI-assisted gastroenterologists maintained high sensitivity and accuracy at follow-up, subsequently outperforming the unassisted gastroenterologists (+20.0%, p = 0.025 and +12.22%, p = 0.05). Thus, best diagnostic scores for the identification of dysplasia emerged through human–machine collaboration between trained gastroenterologists with AI as the second assessor. Therefore, AI could support clinical implementation of optical biopsies through EC.

show abstract

Section: Methodsmentioning

confidence: 99%

Optical Biopsy of Dysplasia in Barrett’s Oesophagus Assisted by Artificial Intelligence

Laan

Putten

Zhao

et al. 2023

Cancers

Self Cite

View full text Add to dashboard Cite

show abstract

“…Most studies used locally collected datasets, and only three studies used known datasets in the eld of clinical images such as MICCAI version 2015, Kvasir Dataset, and ImageNet (11,34,46). Various ML techniques were employed for data recognition and classi cation, with the maximum number of images used for early detection of EC through ML algorithms being 494,356 images (33), and the least used image being 80 images (34,36). On average, 28,939 images were used in the eld of EC detection.…”

Section: Ec Image Segmentationmentioning

confidence: 99%

“…In the eld of early detection of EC, different imaging modalities such as gastroscopy, WLI, and narrow-band imaging (NBI) have been used in various studies (19)(20)(21). A review of the literature showed that WLI images were used in 35% of studies (11,17,(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33), followed by a combination of WLI and NBI images in 10% (18, 23,34,35), CT images in 13% (36-39), NBI images in 3% (40), images of other modalities in 13% (41)(42)(43)(44), and the type of imaging was not mentioned in 26% of studies (Fig. 4) (32,42,(45)(46)(47)(48)(49)(50).…”

Section: Ec Image Segmentationmentioning

confidence: 99%

See 1 more Smart Citation

Machine Learning Applications for Early Detection of Esophageal Cancer: A Systematic Review

Hosseini

Asadi

Emami

et al. 2023

Preprint

View full text Add to dashboard Cite

Introduction: Esophageal cancer (EC) is a significant global health problem, with an estimated 7th highest incidence and 6th highest mortality rate. Timely diagnosis and treatment are critical for improving patients' outcomes, as over 40% of patients with EC are diagnosed after metastasis. Recent advances in machine learning (ML) techniques, particularly in computer vision, have demonstrated promising applications in medical image processing, assisting clinicians in making more accurate and faster diagnostic decisions. Given the significance of early detection of EC, this systematic review aims to summarize and discuss the current state of research on ML-based methods for the early detection of EC. Methods: We conducted a comprehensive systematic search of five databases (PubMed, Scopus, Web of Science, Wiley, and IEEE) using search terms such as "ML", "Deep Learning (DL(", "Neural Networks (NN)", "Esophagus”, “EC" and “Early Detection”. After applying inclusion and exclusion criteria, 31 articles were retained for full review. Results: The results of this review highlight the potential of ML-based methods in the early detection of EC. The average accuracy of the reviewed methods in the analysis of endoscopic and computed tomography (CT (images of the esophagus was over 89%, indicating a high impact on early detection of EC. Additionally, the highest percentage of clinical images used in the early detection of EC with the use of ML was related to white light imaging (WLI) images. Among all ML techniques, methods based on convolutional neural networks (CNN) achieved higher accuracy and sensitivity in the early detection of EC compared to other methods. Conclusion: Our findings suggest that ML methods may improve accuracy in the early detection of EC, potentially supporting radiologists, endoscopists, and pathologists in diagnosis and treatment planning. However, the current literature is limited, and more studies are needed to investigate the clinical applications of these methods in early detection of EC. Furthermore, many studies suffer from class imbalance and biases, highlighting the need for validation of detection algorithms across organizations in longitudinal studies.

show abstract

“…More specifically, our hypothesis is that domain-specific self-supervised pretraining will help make predictions more accurate and stable, thereby contributing to the calibration of the system. For this purpose, we employ a pretrained ResNet50 architecture -in both supervised and unsupervised manner-on ImageNet (≈1.5M natural images) and an unsupervised ResNet50 architecture pretrained on a large-scale dataset of 5M endoscopic images (GastroNet), a ten times larger version of the dataset used before by Van der Putten et al 8 The calibration performance of each encoder is evaluated in two different cases, followed by either deterministic or Bayesian classifier layers. In our experiments, we observe that self-supervised pretraining on domain-specific data outperforms the supervised pretraining on natural images, in terms of characterization performance.…”

Section: Introductionmentioning

confidence: 99%

Effect of domain-specific self-supervised pretraining on predictive uncertainty for colorectal polyp characterization

Dehghani¹,

Scheeve²,

Boers³

et al. 2023

Medical Imaging 2023: Computer-Aided Diagnosis

Self Cite

View full text Add to dashboard Cite

Colorectal polyps (CRPs) are potential precursors of colorectal cancer (CRC), one of the most common types of cancer worldwide. Computer-Aided Diagnosis (CADx) systems can play a crucial role as a second opinion for endoscopists in characterizing CRPs and contribute to the diagnostic performance of colonoscopies. Despite their potential, deep neural network-based systems often tend to overestimate the confidence about their decisions and provide predictive probabilities that are poorly related to their classification accuracy. Quantifying uncertainty of such supportive systems is crucial for optimal clinical workflow integration and physician's acceptance. Thus, a trustworthy CADx system is expected to provide accurate and well-calibrated classification confidence. Transfer learning from either natural image datasets, such as ImageNet, or other datasets with similar modalities, has been widely used for improving the accuracy of deep learning-based systems in medical image classification. In this paper, we study the impact of domain-specific pretraining on the calibration and the overall performance of a CADx system for CRP characterization. We evaluate our hypothesis on a fully deterministic and a hybrid Bayesian version of each approach using a generic ResNet50 architecture. Experimental results demonstrate the effectiveness of domain-specific pretraining in achieving a higher overall characterization AUC. Additionally, the in-domain and out-of-domain pretrained models portray similar calibration error rates, however, their corresponding hybrid Bayesian models offer higher robustness with improved calibration performance. A hybrid Bayesian version of a domain-specific pretraining approach has shown to significantly improve the accuracy and reliability of CADx systems used for CRP characterization and similar positive effects may be expected for other medical imaging applications.

show abstract

Multi-stage domain-specific pretraining for improved detection and localization of Barrett's neoplasia: A comprehensive clinically validated study

Cited by 16 publications

References 18 publications

Optical Biopsy of Dysplasia in Barrett’s Oesophagus Assisted by Artificial Intelligence

Optical Biopsy of Dysplasia in Barrett’s Oesophagus Assisted by Artificial Intelligence

Machine Learning Applications for Early Detection of Esophageal Cancer: A Systematic Review

Effect of domain-specific self-supervised pretraining on predictive uncertainty for colorectal polyp characterization

Contact Info

Product

Resources

About