Correlating medical-dependent query features with image retrieval models using association rules

Abstract

The increasing quantities of available medical resources have motivated the development of effective search tools and medical decision support systems. Medical image search tools help physicians in searching medical image datasets for diagnosing a disease or monitoring the stage of a disease given previous patient's image screenings. Image retrieval models are classified into three categories : contentbased (visual), textual and combined models. In most of previous work, a unique image retrieval model is applied for any user formulated query independently of what retrieval model best suits the information need behind the query. The main challenge in medical image retrieval is to cope the semantic gap between user information needs and retrieval models. In this paper, we propose a novel approach for finding correlations between medical query features and retrieval models based on association rule mining. We define new medical-dependent query features such as image modality and presence of specific medical image terminology and make use of existing generic query features such as query specificity, ambiguity and cohesiveness. The proposed query features are then exploited into association rule mining for discovering rules which correlate query features to visual, textual or combined image retrieval models. Based on the discovered rules, we propose to use an associative classifier that finds the best suitable rule with a maximum feature coverage for a new query. Experiments are performed on Image CLEF queries from 2008 to 2012 where we evaluate the impact of our proposed query features on the classification performance. Results show that combining our proposed specific and generic query features is effective for classifying queries. A comparative study between our classifier, CBA, Naïve Bayes, Bayes Net and decision trees showed that our best coverage associative classifier outperforms existing classifiers where it achieves an improvement of 30%.

Mining correlations between medically dependent features and image retrieval models for query classification

“…For the 2011 data set, the number of queries annotated by the default class is 13 of 30 queries. Although the results show that our classifier (NaiveClass) outperforms the existing classifiers in terms of accuracy rate (Ayadi et al, 2013), it has a major limitation when it uses the same default class textual "T" during classification. Figure 5 shows the error distribution by changing the default class.…”

“…Several information retrieval (IR) studies (Hauff, Azzopardi, & Hiemstra, ; Tamine, Chouquet, & Palmer, ) have adopted features such as term frequency and query length to predict the effectiveness of query and retrieval systems. Ayadi et al () and Bashir and Rauber () used these features to predict a correlation between query and retrieval function; Burges et al (), Can, Croft, and Manmatha (), Cao, Qin, Liu, Tsai, and Li (), and Ye and Huang () used them to learn to rank, and Xu, Xu, Wang, and Wang () used them to re‐rank. In this section, we present existing query and re‐ranking features for improving image retrieval performance.…”

“…Our previous work (Ayadi et al, ) addresses our query classification approach: A new classifier based on both specific and generic query features is proposed. Each query has a set of specific features that depend on their domain.…”

“…In this section, we briefly introduce the principle of association rule mining using features. We then detail the “ NaiveClass ” classifier proposed in Ayadi et al ().…”

“…In our previous work (Ayadi et al, ), we proposed a naive algorithm for query classification based on best feature coverage criterion. This algorithm ( NaiveClass ) is based on the selection of the rule that contains the maximum of common features with the query to label the query.…”