Samba: Severity-aware Recurrent Modeling for Cross-domain Medical Image Grading

Authors: Qi Bi, Jingjun Yi, Hao Zheng, Wei Ji, Haolan Zhan, Yawen Huang, Yuexiang Li, Yefeng Zheng

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

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments show the proposed Samba outperforms the VMamba baseline by an average accuracy of 23.5%, 5.6% and 4.1% on the cross-domain grading of fatigue fracture, breast cancer and diabetic retinopathy, respectively.
Researcher Affiliation Collaboration Qi Bi1 , Jingjun Yi2, Hao Zheng2 , Wei Ji3, Haolan Zhan4, Yawen Huang2, Yuexiang Li5 , Yefeng Zheng1 1Westlake University, China, 2Jarvis Research Center, Tencent Youtu Lab, China, 3Yale University, United States, 4Monash University, Australia, 5Guangxi Medical University, China
Pseudocode No The paper describes its methodology in detail using prose and diagrams (e.g., Figure 2) but does not include any explicit pseudocode blocks or algorithm listings.
Open Source Code Yes Source code is available at https://github.com/Bi Qi WHU/Samba.
Open Datasets Yes Cross-domain Fatigue Fracture Grading Benchmark [31]... Cross-domain Breast Cancer Grading Benchmark...2https://github.com/YANRUI121/Breast-cancer-grading... Cross-domain Diabetic Retinopathy Grading Benchmark consists of six DR retinal image datasets, namely, Deep DR [33], Messidor [1], IDRID [40], APTOS [3], FGADR [67], and RLDR [49].
Dataset Splits No The paper defines source and target domains for cross-domain generalization and mentions the use of training and testing data. However, it does not specify explicit train/validation/test splits (e.g., percentages or sample counts) within these domains for reproducibility.
Hardware Specification No The paper does not specify any particular hardware components such as GPU models, CPU types, or memory used for running the experiments. It mentions GFLOPs as a model metric, not hardware.
Software Dependencies No The paper does not provide specific version numbers for any software dependencies (e.g., programming languages, libraries, frameworks) used in the experiments.
Experiment Setup Yes In the EM algorithm, the iteration number T plays an important role... we report the results when the iteration number T of the EM algorithm varies from 1 to 8... By default K is set to 64, and we further test the performance when K is set to 16, 32, 48 and 96, respectively... Using moving average to update the severity base µk is able to improve the performance substantially.