Table 7.

Comparative assessment of limitations and future perspectives for research in diagnosis interpretability.

References	Specific limitation	Specific future perspective
Monajatipoor et al (2022)	Data annotation; visual encoder choices; evaluation scope	Larger annotated datasets; visual encoder flexibility; diverse medical imaging tasks
Wang et al (2023b)	Non-human-like language generation; limited prompts; impact of LLM Size	Language model refinement; exploration of more complex information; investigation of vision classifiers; quantitative analysis of prompt design; engagement with clinical professionals
Yan et al (2023a)	Potential biases; accuracy of concept scores	Improved language models; robustness enhancements; reducing model dependency; biased data mitigation
Patrício et al (2023)	Interpretable concept	Expanding to other imaging modalities; enhanced interpretability