Efficient disease detection in gastrointestinal videos – global features versus neural networks

Abstract

Analysis of medical videos from the human gastrointestinal (GI) tract for detection and localization of abnormalities like lesions and diseases requires both high precision and recall. Additionally, it is important to support efficient, real-time processing for live feedback during (i) standard colonoscopies and (ii) scalability for massive population-based screening, which we conjecture can be done using a wireless video capsule endoscope (camera-pill). Existing related work in this field does neither provide the necessary combination of accuracy and performance for detecting multiple classes of abnormalities simultaneously nor for particular disease localization tasks. In this paper, a complete end-toend multimedia system is presented where the aim is to tackle automatic analysis of GI tract videos. The system includes an entire pipeline ranging from data collection, processing and analysis, to visualization. The system combines deep learning neural networks, information retrieval, and analysis of global and local image features in order to implement multi-class classification, detection and localization. Furthermore, it is built in a modular way, so that it can be easily extended to deal with other types of abnormalities. Simultaneously, the system is developed for efficient processing in order to provide real-time feedback to the doctors and for scalability reasons when potentially applied for massive population-based algorithmic screenings in the future. Initial experiments show that our system has multiclass detection accuracy and polyp localization precision at least as good as state-of-the-art systems, and provides additional novelty in terms of real-time performance, low resource consumption and ability to extend with support for new classes of diseases.

Bleeding detection in wireless capsule endoscopy videos — Color versus texture features

“…the feature values used. The machine-learning approaches used are RT, RF, and LMT; all have been proven 31,32,34 to be able to provide a good two-class classification performance for GI tract frames. RT is the simplest one that therefore has the lowest computation complexity with good enough performance.…”

Performance of Data Enhancements and Training Optimization for Neural Network: A Polyp Detection Case Study

“…A machine learning approach that has been reborn and lately gained a lot of interest is neural networks [14] which is a type of machine learning which loosely mimics how a biological brain learns, i.e., being able to learn general concepts from concrete examples. Neural networks are also used in the medical domain, e.g., for micro-calcification detection in mammogram images [15], detecting breast cancer [3], colonic polyp detection [5], [6], [7], [8], [9] and lung cancer [4]. To perform the analysis of the data, deep neural networks (deep learning) contain multiple network layers where each layer can learn different abstraction levels of the data using the input of previous layers.…”

“…Thus, detection and removal of polyps therefore prevents the development of cancer, and our detection challenge is shown in figure 1, where the polyps must be detected and identified among images and video frames also containing normal mucosa and various other anatomical landmarks and abnormalities. With respect to using neural networks for this type of abnormality detection, research have already proven that neural networks can be suitable, e.g., for similar problems detecting breast cancer [3], lung cancer [4], and for polyp detection in particular [5], [6], [7], [8], [9]. In this paper, we enhance our neural network-based EIR system (named after a Norse goddess associated with medical skills) with data enhancement methods and assess their effect on detection performance.…”

On medical application of neural networks trained with various types of data

“…A method of detecting gastrointestinal disease in endoscopy videos has been proposed (35). This method can classify multiple types of gastrointestinal diseases using a convolutional neural network.…”