Automatic detection of small bowel tumors in wireless capsule endoscopy images using ensemble learning

Abstract

PurposeWireless Capsule Endoscopy (WCE) is a minimally invasive diagnosis tool for lesion detection in the gastrointestinal tract, reaching places where conventional endoscopy is unable to. However, the significant amount of acquired data leads to difficulties in the diagnosis by the physicians; which can be eased with computer assistance. This paper addresses a method for the automatic detection of tumors in WCE by using a two‐step based procedure: region of interest selection and classification.MethodsThe first step aims to separate abnormal from normal tissue by using automatic segmentation based on a Gaussian Mixture Model (GMM). A modified version of the Anderson method for convergence acceleration of the expectation–maximization (EM) algorithm is proposed. The proposed features for both segmentation and classification are based on the CIELab color space, as a way of bypassing lightness variations, where the L component is discarded. Tissue variability among subjects, light inhomogeneities and even intensity differences among different devices can be overcome by using simultaneously features from both regions. In the second step, an ensemble system with partition of the training data with a new training scheme is proposed. At this stage, the gating network is trained after the experts have been trained decoupling the joint maximization of both modules. The partition module is also used at the test step, leading the incoming data to the most likely expert allowing incremental adaptation by preserving data diversity.ResultsThis algorithm outperforms others based on texture features selected from Wavelets and Curvelets transforms, classified by a regular support vector machine (SVM) in more than 5%.ConclusionsThis work shows that simpler features can outperform more elaborate ones if appropriately designed. In the current case, luminance was discarded to cope with saturated tissue, facilitating the color perception. Ensemble systems remain an open research field. In the current case, changes in both topology and training strategy have led to significant performance improvements. A system with this level of performance can be used in current clinical practice.

Artificial intelligence in capsule endoscopy: development status and future expectations

“…Early research in AI-assisted capsule endoscopy for this application includes a study by Li et al in 2011, which utilised colour texture features to differentiate between normal and tumour-containing images with a sensitivity of 92.33% and a specificity of 88.67% [55] . Multiple other machine learning models utilising Binary Classifiers, SVMs, and MLPs have been utilised to varying accuracies and efficacies [56][57][58][59][60][61] . Deep learning was integrated into the field with the study by Yuan and Meng in 2017 [62] , where they utilised a stacked sparse autoencoder method to categorise images into polyps, bubbles, turbid images, and clear images with an overall accuracy of 98.00%.…”

Artificial intelligence within the small bowel: are we lagging behind?

“…Few studies on the delineation of polyps or tumours by deep networks exist [30] (Table 3). Sensitivity to detect such lesions is high but specificity falls possibly because of the variability of such lesions.…”

Classification of Cervical Biopsy Images Based on LASSO and EL-SVM

“…To solve this kind of problem, this paper proposes ensemble learning-support vector machine (EL-SVM). The algorithm flow is shown in Figure 11, in which the integrated learning method uses the Bagging algorithm [25]. The bootstrap sampling of cervical biopsy tissue image samples is used to generate ''good and different'' training and test sets.…”