Using a Deep Learning Algorithm and Integrated Gradients Explanation to Assist Grading for Diabetic Retinopathy.

By Rory Sayres, Ankur Taly, Ehsan Rahimy, Katy Blumer, David Coz, N. Hammel, Jonathan Krause, Arunachalam Narayanaswamy, Zahra Rastegar, Derek J. Wu and 9 more. Ophthalmology (Rochester, Minn.), 2019

379 citations recorded by the source corpus.

PURPOSE To understand the impact of deep learning diabetic retinopathy (DR) algorithms on physician readers in computer-assisted settings. DESIGN Evaluation of diagnostic technology. PARTICIPANTS One thousand seven hundred ninety-six retinal fundus images from 1612 diabetic patients. METHODS Ten ophthalmologists (5 general ophthalmologists, 4 retina specialists, 1 retina fellow) read images for DR severity based on the International Clinical Diabetic Retinopathy disease severity scale in each of 3 conditions: unassisted, grades only, or grades plus heatmap. Grades-only assistance comprised a histogram of DR predictions (grades) from a trained deep-learning model. For grades plus heatmap, we additionally showed explanatory heatmaps. MAIN OUTCOME MEASURES For each experiment arm, we computed sensitivity and specificity of each reader and the algorithm for different levels of DR severity against an adjudicated reference standard. We also measured accuracy (exact 5-class level agreement and Cohen's quadratically weighted κ), reader-reported confidence (5-point Likert scale), and grading time. RESULTS Readers graded more accurately with model assistance than without for the grades-only condition (P < 0.001). Grades plus heatmaps improved accuracy for patients with DR (P < 0.001), but reduced accuracy for patients without DR (P = 0.006). Both forms of assistance increased readers' sensitivity moderate-or-worse DR: unassisted: mean, 79.4% [95% confidence interval (CI), 72.3%-86.5%]; grades only: mean, 87.5% [95% CI, 85.1%-89.9%]; grades plus heatmap: mean, 88.7% [95% CI, 84.9%-92.5%] without a corresponding drop in specificity (unassisted: mean, 96.6% [95% CI, 95.9%-97.4%]; grades only: mean, 96.1% [95% CI, 95.5%-96.7%]; grades plus heatmap: mean, 95.5% [95% CI, 94.8%-96.1%]). Algorithmic assistance increased the accuracy of retina specialists above that of the unassisted reader or model alone; and increased grading confidence and grading time across all readers. For most cases, grades plus heatmap was only as effective as grades only. Over the course of the experiment, grading time decreased across all conditions, although most sharply for grades plus heatmap. CONCLUSIONS Deep learning algorithms can improve the accuracy of, and confidence in, DR diagnosis in an assisted read setting. They also may increase grading time, although these effects may be ameliorated with experience.

Subfields: Ophthalmology, Radiology, Nuclear Medicine and Imaging

View Paper: https://www.semanticscholar.org/paper/73fb26f9b9386704d3fb25ca9ec39966776ce95e

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Radiology, Nuclear Medicine and Imaging

Citations in the corpus

Key challenges for delivering clinical impact with artificial intelligence (2019, background): “It has also been shown that humans assisted by AI performed better than either alone in a study of diabetic retinopathy screening [96, 97].”
Explainable Deep Learning Models in Medical Image Analysis (2020, methodology): “More studies like [36] need to be conducted to observe the effect of the explainability models on the decision time and accuracy of the clinical experts.”
Assessing the (Un)Trustworthiness of Saliency Maps for Localizing Abnormalities in Medical Imaging (2020, background): “A recent study also showed that saliency maps did not provide additional performance improvement in assisted clinician interpretation compared to only providing a model prediction[34].”
Opportunities and Challenges in Explainable Artificial Intelligence (XAI): A Survey (2020, background): “in healthcare [1]–[3], ophthalmology [4]–[6], developmental disorders [7]–[9], in autonomous robots and vehicles [10]–[12], in image processing classification and detection [13], [14], in speech and audio processing [15], [16], cyber-security [17], [18], and many more indicate the reach of DL algori”
Learning Explainable Models Using Attribution Priors (2019, background): “Previous literature about such methods has focused on the axioms they should satisfy [18, 37, 35, 6], and how attribution methods can give us insight into model behavior [19, 20, 29, 41].”