CLASSIFYING FEATURES FOR AN ORAL LESION IMAGE LEARNING TOOL

Abstract

<h3>Objectives</h3> A web-based image classification tool (DiLearn) was developed to facilitate active learning in the oral health profession. Students engage with oral lesion images using swipe gestures to classify each image into predetermined categories (e.g., Left - refer; Right - no intervention). To assemble the training modules and to provide feedback to students, DiLearn requires each oral lesion image to be classified with various features displayed in the image. Collection of accurate meta information is a crucial step for enabling the self-directed active learning approach taken in DiLearn. The purpose of this study is to evaluate the classification consistency of features in oral lesion images by experts and students for use of the learning tool. <h3>Study Design</h3> Twenty oral lesion images from DiLearn's image bank were classified by 3 oral lesion experts and 2 senior dental hygiene students using the same rubric containing 8 features. Fleiss's κ was used to evaluate the classification agreement among oral lesion experts and to evaluate the overall agreement. Cohen's κ was used to compare agreement between student raters. <h3>Results</h3> Classification agreement between the 3 experts ranged from identical (Fleiss's κ = 1) for clinical action to slight agreement for border: regularity (Fleiss's κ = 0.136), with most categories having fair to moderate agreement (Fleiss's κ = 0.332-545). With the exception of morphology, inclusion of the 2 student raters with the experts yielded fair to moderate overall classification agreement (Fleiss's κ = 0.224-0.554). In sum, the feature clinical action can be accurately classified, whereas other anatomic features indirectly related to diagnosis have lower classification consistency. <h3>Conclusions</h3> Findings suggest 1 oral lesion expert or 2 student raters can provide fairly consistent meta information for select features implicated in the creation of image classification tasks in DiLearn.