English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

To reduce doctors' workload, deep-learning-based automatic medical reportgeneration has recently attracted more and more research efforts, where deepconvolutional neural networks (CNNs) are employed to encode the input images,and recurrent neural networks (RNNs) are used to decode the visual featuresinto medical reports automatically. However, these state-of-the-art methodsmainly suffer from three shortcomings: (i) incomprehensive optimization, (ii)low-order and unidimensional attention mechanisms, and (iii) repeatedgeneration. In this article, we propose a hybrid reinforced medical reportgeneration method with m-linear attention and repetition penalty mechanism(HReMRG-MR) to overcome these problems. Specifically, a hybrid reward withdifferent weights is employed to remedy the limitations of single-metric-basedrewards. We also propose a search algorithm with linear complexity toapproximate the best weight combination. Furthermore, we use m-linear attentionmodules to explore high-order feature interactions and to achieve multi-modalreasoning, while a repetition penalty applies penalties to repeated termsduring the model's training process. Extensive experimental studies on twopublic datasets show that HReMRG-MR greatly outperforms the state-of-the-artbaselines in terms of all metrics. We also conducted a series of ablationexperiments to prove the effectiveness of all our proposed components. We alsoperformed a reward search toy experiment to give evidence that our proposedsearch approach can significantly reduce the search time while approximatingthe best performance.

Hybrid Reinforced Medical Report Generation with M-Linear Attention and Repetition Penalty