Table 2.

The BLEU^a, METEOR^b, and ROUGE-L^c scores of the generated reports by various methods on the MIMIC-CXR data set.

Methods	MIMIC-CXR
	BLEU-1	BLEU-2	BLEU-3	BLEU-4	BLEU-Ad	METEOR	ROUGE-L
R2GEN	0.353	0.218	0.145	0.103	0.155	0.142	0.270
CAe	0.350	0.219	0.152	0.109	0.160	0.151	0.283
CMCLf	0.344	0.217	0.140	0.097	0.151	0.133	0.281
PPKEDg	0.360	0.224	0.149	0.106	0.160	0.149	0.284
M2TRh	0.353	0.218	0.148	0.106	0.157	0.142	0.278
R2GENCMNi	0.378	0.232	0.154	0.107	0.164	0.145	0.272
ALIGNTRANSFORMER	0.378	0.235	0.156	0.112	0.168	N/Aj	0.283
KNOWMATk	0.363	0.228	0.156	0.115	0.166	N/A	0.284
CMM-RL	0.381	0.232	0.155	0.109	0.165	0.151	0.287
CMCAl	0.360	0.227	0.156	0.117	0.167	0.148	0.287
ORGANm	0.407	0.256	0.172	0.123	0.184	0.162	0.293
ClinicalBLIP	0.332	0.219	0.153	0.115	0.162	0.365	0.313

^aBLEU: Bilingual Evaluation Understudy.

^bMETEOR: Metric for Evaluation of Translation With Explicit Ordering.

^cROUGE-L: Recall-Oriented Understudy for Gisting Evaluation-L.

^dBLEU-A: average of the BLEU-2/3/4 scores.

^eCA: contrastive attention.

^fCMCL: Competence-Based Multimodal Curriculum Learning.

^gPPKED: Posterior-and-Prior Knowledge Exploring-and-distilling.

^hM2TR: Meshed-Memory Transformer.

ⁱR2GENCMN: R2GEN enhanced with cross-modal memory networks.

^jN/A: not available.

^kKNOWMAT: Knowledge Matters.

^lCMCA: Cross-Modal Contrastive Attention Model.

^mORGAN: Observation-Guided Radiology Report Generation Framework.