MedCC: Interpreting Medical Images Using Clinically Significant Concepts and Descriptions

Xuwen Wang, Zhen Guo, Ziyang Wang, Jiao Li

2022

Abstract

This paper aims to identify valuable semantic concepts and predict descriptions automatically for medical images to assist doctors in image reading. A simple framework called MedCC is proposed for medical image concept detection and caption prediction. MedCC employed multiple fine-grained multi-label classification (MLC) models trained on manually annotated datasets, which contain image-concept pairs of different semantic types, such as Imaging Type, Anatomic Structure, and Findings. We validate the performance of MedCC based on the open sourced concept detection dataset and achieved the best F1 score of 0.419, which is comparable with the SOTA models. Combining the detected concepts into sentences according to the manually defined sentence patterns resulted in a BLEU score of 0.257, which still has room for improvement.

Paper Citation

in Harvard Style

Wang X., Guo Z., Wang Z. and Li J. (2022). MedCC: Interpreting Medical Images Using Clinically Significant Concepts and Descriptions. In Proceedings of the 3rd International Symposium on Automation, Information and Computing - Volume 1: ISAIC; ISBN 978-989-758-622-4, SciTePress, pages 518-525. DOI: 10.5220/0011954800003612

in Bibtex Style

@conference{isaic22,
author={Xuwen Wang and Zhen Guo and Ziyang Wang and Jiao Li},
title={MedCC: Interpreting Medical Images Using Clinically Significant Concepts and Descriptions},
booktitle={Proceedings of the 3rd International Symposium on Automation, Information and Computing - Volume 1: ISAIC},
year={2022},
pages={518-525},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011954800003612},
isbn={978-989-758-622-4},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 3rd International Symposium on Automation, Information and Computing - Volume 1: ISAIC
TI - MedCC: Interpreting Medical Images Using Clinically Significant Concepts and Descriptions
SN - 978-989-758-622-4
AU - Wang X.
AU - Guo Z.
AU - Wang Z.
AU - Li J.
PY - 2022
SP - 518
EP - 525
DO - 10.5220/0011954800003612
PB - SciTePress