Scan, dwell, decide: strategies for detecting abnormalities in diabetic retinopathy
Abstract:
Diabetic retinopathy (DR) is a disease which is widely diagnosed using (colour fundus) images. Efficiency and accuracy are critical in diagnosing DR as lack of timely intervention can lead to irreversible visual impairment. In this paper, we examine strategies for scrutinizing images which affect diagnostic performance of medical practitioners via an eye-tracking study. A total of 56 subjects with 0 to 18 years of experience participated in the study. Every subject was asked to detect DR from 40 images. The findings indicate that practitioners use mainly two types of strategies characterized by either higher dwell duration or longer track length. The main findings of the study are that higher dwell-based strategy led to higher average accuracy (>85%) in diagnosis, irrespective of the expertise of practitioner; whereas, the average obtained accuracy with a long-track length-based strategy was dependent on the expertise of the practitioner. In the second part of the paper, we use the experimental findings to recommend a scanning strategy for fast and accurate diagnosis of DR that can be potentially used by image readers. This is derived by combining the eye-tracking gaze maps of medical experts in a novel manner based on a set of rules. This strategy requires scrutiny of images in a manner which is consistent with spatial preferences found in human perception in general and in the domain of fundus images in particular. The Levenshtein distance-based assessment of gaze patterns also establish the effectiveness of the derived scanning pattern and is thus recommended for image readers.About the dataset:
Data collection is done as follows:
Stimuli Images
A dataset of 40 images with approximately equal number of normal and abnormal cases was selected by senior retina consultant from this experiment. This set contains images with various level of diagnostic difficulties.
Participants
In order to maintain diversity among the participants, retina experts were invited from 6 different eye hospitals. A total of 44 retina experts accepted the invitation and participated in the study. Retina experts were classified into 3 categories, namely, consultants, fellows and residents/optometrists. 12 engineering students from our home institute were also included to fulfill the role of novices. They were given a short training with 10 example fundus images.
Material
Experiments were carried out in a dark room with constant environment condition. Tobii X2-30 eye tracker (sampling rate 30Hz) with a 15.6'' display screen were used. Tobii X2-30 eye-tracker allows free head movement of 20'' x 14'' (Width x Height) at the distance of 70m (distance may vary from person to person). This range is compatible with the size of used display. Stimuli (images) were down-scaled to fit the display size. A software tool was developed in MATLAB 8.2 for use in the study to help participants view images and give their decisions. A small tutorial was also given to all participants to ensure easy usage of the software.
Experiment Design
Check paper for details on experiment design and results.
Download
Below is the link for the download of our DR eye tracking dataset
Please cite the following publication when using this results:
Rangrej SB, Sivaswamy J, Srivastava P - Scan, dwell, decide: strategies for detecting abnormalities in diabetic retinopathy, PLOS ONE (To appear).