Automatic medical image annotation and keyword-based image retrieval using relevance feedback

Abstract

This paper presents novel multiple keywords annotation for medical images, keyword-based medical image retrieval, and relevance feedback method for image retrieval for enhancing image retrieval performance. For semantic keyword annotation, this study proposes a novel medical image classification method combining local wavelet-based center symmetric-local binary patterns with random forests. For keyword-based image retrieval, our retrieval system use the confidence score that is assigned to each annotated keyword by combining probabilities of random forests with predefined body relation graph. To overcome the limitation of keyword-based image retrieval, we combine our image retrieval system with relevance feedback mechanism based on visual feature and pattern classifier. Compared with other annotation and relevance feedback algorithms, the proposed method shows both improved annotation performance and accurate retrieval results.

Recognizing pedestrian’s unsafe behaviors in far-infrared imagery at night

“…Among the representative classifiers, SVM is one of the most popular learning techniques used in action recognition. Although SVM generalizes well on unseen examples, it is the insufficiency of labelled examples especially the small number of negative examples and is not suitable when a feature has high-dimensionality as a result of computational complexity [16]. RF is an ensemble classifier of a number of decision trees, with each tree grown using some types of randomization.…”

“…Boosted RF adds a bootstrapping phase during the learning step, which is similar to the Adaboost algorithm [28]. The two parameters of the Boosted RF, a depth of tree (D) and the initial number of trees (T), are set as 20 and 120, based on the experimental results of [16].…”

A new approach for image databases design

“…Lu et al [29] integrated the imageguided surgery toolkit and the medical imaging interaction toolkits for the development of an image-guided surgery navigation system. Ko et al [22] developed an automatic approach to create annotations for medical images; correspondingly, images were retrieved using keywords. Dubey et al [14] presented a descriptor of image features based on the local diagonal external pattern; it was mainly applied in CT image retrieval.…”

ImageCLEF 2014: Overview and Analysis of the Results

“…DT divides the data into smaller non-overlapping subsets, according to if-then-rules. Byoung et al [12] used a combination of RFs and wavelet-based center symmetric-local binary patterns for medical image classification to perform multiple keyword annotations. It has been shown that classification using RFs is much faster than SVMs.…”