IBMResearch at MEDIQA 2021: Toward Improving Factual Correctness of Radiology Report Abstractive Summarization

Diwakar Mahajan, Ching-Huei Tsou, Jennifer J Liang


Abstract
Although recent advances in abstractive summarization systems have achieved high scores on standard natural language metrics like ROUGE, their lack of factual consistency remains an open challenge for their use in sensitive real-world settings such as clinical practice. In this work, we propose a novel approach to improve factual correctness of a summarization system by re-ranking the candidate summaries based on a factual vector of the summary. We applied this process during our participation in MEDIQA 2021 Task 3: Radiology Report Summarization, where the task is to generate an impression summary of a radiology report, given findings and background as inputs. In our system, we first used a transformer-based encoder-decoder model to generate top N candidate impression summaries for a report, then trained another transformer-based model to predict a 14-observations-vector of the impression based on the findings and background of the report, and finally, utilized this vector to re-rank the candidate summaries. We also employed a source-specific ensembling technique to accommodate for distinct writing styles from different radiology report sources. Our approach yielded 2nd place in the challenge.
Anthology ID:
2021.bionlp-1.35
Volume:
Proceedings of the 20th Workshop on Biomedical Language Processing
Month:
June
Year:
2021
Address:
Online
Editors:
Dina Demner-Fushman, Kevin Bretonnel Cohen, Sophia Ananiadou, Junichi Tsujii
Venue:
BioNLP
SIG:
SIGBIOMED
Publisher:
Association for Computational Linguistics
Note:
Pages:
302–310
Language:
URL:
https://aclanthology.org/2021.bionlp-1.35
DOI:
10.18653/v1/2021.bionlp-1.35
Bibkey:
Cite (ACL):
Diwakar Mahajan, Ching-Huei Tsou, and Jennifer J Liang. 2021. IBMResearch at MEDIQA 2021: Toward Improving Factual Correctness of Radiology Report Abstractive Summarization. In Proceedings of the 20th Workshop on Biomedical Language Processing, pages 302–310, Online. Association for Computational Linguistics.
Cite (Informal):
IBMResearch at MEDIQA 2021: Toward Improving Factual Correctness of Radiology Report Abstractive Summarization (Mahajan et al., BioNLP 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.bionlp-1.35.pdf