Abstract
Diagnostic support tools based on artificial intelligence (AI) have exhibited high performance in various medical fields. However, their clinical application remains challenging because of the lack of explanatory power in AI decisions (black box problem), making it difficult to build trust with medical professionals. Nevertheless, visualizing the internal representation of deep neural networks will increase explanatory power and improve the confidence of medical professionals in AI decisions. We propose a novel deep learning-based explainable representation “graph chart diagram” to support fetal cardiac ultrasound screening, which has low detection rates of congenital heart diseases due to the difficulty in mastering the technique. Screening performance improves using this representation from 0.966 to 0.975 for experts, 0.829 to 0.890 for fellows, and 0.616 to 0.748 for residents in the arithmetic mean of area under the curve of a receiver operating characteristic curve. This is the first demonstration wherein examiners used deep learning-based explainable representation to improve the performance of fetal cardiac ultrasound screening, highlighting the potential of explainable AI to augment examiner capabilities.
Highlights
IntroductionWith the rapid development of tools to support medical diagnosis using artificial intelligence (AI), expectations from AI have been increasing continuously [1,2,3,4,5]
We show the evaluation of abnormality detection performance of artificial intelligence (AI) only by abnormality scores calculated from graph chart diagrams
We proposed a deep learning-based explainable representation that compresses and represents the information in fetal cardiac ultrasound screening videos and introduced two factors to realize the depiction of the graph chart, including a cascade graph encoder and view-proxy loss
Summary
With the rapid development of tools to support medical diagnosis using artificial intelligence (AI), expectations from AI have been increasing continuously [1,2,3,4,5]. One of the major obstacles is regarded as the “black box problem” of AI [4,6,7,8]. The black box problem is a problem in which the relationship between input and output obtained from data is so complicated that any human, including the developer, cannot determine the rationale for the AI decision [9]. There are three major approaches for achieving explainable AI using a deep neural network (DNN), a machine learning technology typically used in medical
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
