Stomach Deformities Recognition Using Rank-Based Deep Features Selection

Doctor utilizes various kinds of clinical technologies like MRI, endoscopy, CT scan, etc., to identify patient’s deformity during the review time. Among set of clinical technologies, wireless capsule endoscopy (WCE) is an advanced procedures used for digestive track malformation. During this complet...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Journal of medical systems Jg. 43; H. 12; S. 329
Hauptverfasser: Khan, Muhammad Attique, Sharif, Muhammad, Akram, Tallha, Yasmin, Mussarat, Nayak, Ramesh Sunder
Format: Journal Article
Sprache:Englisch
Veröffentlicht: New York Springer US 01.12.2019
Springer Nature B.V
Schlagworte:
Abnormalities
Artificial neural networks
Capsule Endoscopy - methods
Classification
Colorectal cancer
Computed tomography
Computer simulation
Endoscopy
Entropy
Feature extraction
Feature recognition
Health Informatics
Health Informatics and Computer Vision
Health Sciences
Hemorrhage - diagnosis
Hemorrhage - diagnostic imaging
Humans
Image & Signal Processing
Image classification
Image processing
Image Processing, Computer-Assisted - methods
Image segmentation
Magnetic resonance imaging
Medicine
Medicine & Public Health
Neural networks
Neural Networks, Computer
Optimization
Physicians
Recent Advances in Deep Learning for Biomedical Signal Processing
Statistics for Life Sciences
Stomach Diseases - diagnosis
Stomach Diseases - diagnostic imaging
Stomach Ulcer - diagnostic imaging
Transfer learning
WCE
Deep features selection
Saliency estimation
Features fusion
Colorectal cancer
ISSN:0148-5598, 1573-689X, 1573-689X
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Doctor utilizes various kinds of clinical technologies like MRI, endoscopy, CT scan, etc., to identify patient’s deformity during the review time. Among set of clinical technologies, wireless capsule endoscopy (WCE) is an advanced procedures used for digestive track malformation. During this complete process, more than 57,000 frames are captured and doctors need to examine a complete video frame by frame which is a tedious task even for an experienced gastrologist. In this article, a novel computerized automated method is proposed for the classification of abdominal infections of gastrointestinal track from WCE images. Three core steps of the suggested system belong to the category of segmentation, deep features extraction and fusion followed by robust features selection. The ulcer abnormalities from WCE videos are initially extracted through a proposed color features based low level and high-level saliency (CFbLHS) estimation method. Later, DenseNet CNN model is utilized and through transfer learning (TL) features are computed prior to feature optimization using Kapur’s entropy. A parallel fusion methodology is opted for the selection of maximum feature value (PMFV). For feature selection, Tsallis entropy is calculated later sorted into descending order. Finally, top 50% high ranked features are selected for classification using multilayered feedforward neural network classifier for recognition. Simulation is performed on collected WCE dataset and achieved maximum accuracy of 99.5% in 21.15 s.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
ISSN:0148-5598
1573-689X
1573-689X
DOI:10.1007/s10916-019-1466-3