PromptMRG: Diagnosis-Driven Prompts for Medical Report Generation

Authors: Haibo Jin, Haoxuan Che, Yi Lin, Hao Chen

AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on two MRG benchmarks show the effectiveness of the proposed method, where it obtains state-of-the-art clinical efficacy performance on both datasets.
Researcher Affiliation Academia 1Department of Computer Science and Engineering, Hong Kong University of Science and Technology, Hong Kong, China 2Department of Chemical and Biological Engineering, Hong Kong University of Science and Technology, Hong Kong, China
Pseudocode No The paper describes its methods but does not include a dedicated pseudocode or algorithm block.
Open Source Code No The paper does not provide an explicit statement about releasing source code or a link to a code repository.
Open Datasets Yes 1) MIMIC-CXR (Johnson et al. 2019) is the largest MRG dataset with chest X-rays and paired reports. 2) IU X-Ray (Demner-Fushman et al. 2016) is also widely used for MRG evaluation
Dataset Splits Yes We follow the official split and the preprocessing from Chen et al. (2020), where the processed dataset has 270,790, 2,130, and 3,858 samples for training, validation, and test, respectively.
Hardware Specification Yes The model was implemented with Py Torch 2.0 and trained with one RTX 3090 GPU for about 24 hours.
Software Dependencies Yes The model was implemented with Py Torch 2.0 and trained with one RTX 3090 GPU for about 24 hours.
Experiment Setup Yes The coefficient λ is set to 4 and k = 21 is used for CFE. Adam W (Loshchilov and Hutter 2017) is used as the optimizer with a weight decay of 0.05. The initial learning rate is set to 5e-5 with a cosine learning rate schedule. The number of training epochs is 10, batch size is 16, and image size is 224.