MEATRD: Multimodal Anomalous Tissue Region Detection Enhanced with Spatial Transcriptomics
Authors: Kaichen Xu, Qilong Wu, Yan Lu, Yinan Zheng, Wenlin Li, Xingjie Tang, Jun Wang, Xiaobo Sun
AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive evaluations across eight real ST datasets reveal MEATRD s superior performance in ATR detection, surpassing various state-of-the-art AD methods. Remarkably, MEATRD also proves adept at discerning ATRs that only show slight visual deviations from normal tissues. Extensive benchmarks on eight breast cancer ST datasets demonstrate MEATRD s superiority over nine state-of-the-art (SOTA) AD methods in accurately detecting ATRs, including those with subtle visual deviations from surrounding normal tissues. Table 1 showcases MEATRD s superiority over baselines in detecting ATRs across datasets, consistently ranking first in AUC scores and six times in F1 scores. It outperforms the second-best performing method with an average leap of 17.45% in AUC scores and 10.31% in F1 scores. Furthermore, Table 4 in supplementary material F indicates that our model performed well in detecting PSCs, demonstrating its generalization to other types of diseases. We conduct ablation studies over the eight human breast cancer ST datasets (i.e., 10x-h BC-{A1-H1}) to investigate the effects of MEATRD s key components in ATR detection. Here, we conduct sensitivity analyses on eight 10x-h BC datasets to examine the effects of MEATRD s key hyperparameters, including α and β. |
| Researcher Affiliation | Collaboration | 1 School of Statistics and Mathematics, Zhongnan University of Economics and Law 2 i Wudao Tech EMAIL, EMAIL, EMAIL |
| Pseudocode | Yes | The workflow of Stage II is illustrated in Figure 2 and Algorithm 1 in supplementary material C. The training workflow of Stage III is also illustrated in Algorithm 2 of supplementary material C. |
| Open Source Code | Yes | Code https://github.com/wqlzuel/MEATRD |
| Open Datasets | Yes | MEATRD is extensively evaluated across eight breast cancer datasets and four primary sclerosing cholangitis (PSC) datasets. (see supplementary material E for data description and preprocessing). By far, a total of 1033 publicly available human ST datasets that span 56 diseases and 35 tissues, providing a rich resource for investigating ATRs at the molecular level (Wang et al. 2024). Wang, G.; Wu, S.; Xiong, Z.; Qu, H.; Fang, X.; and Bao, Y. 2024. CROST: a comprehensive repository of spatial transcriptomics. Nucleic Acids Research, 52(D1): D882 D890. |
| Dataset Splits | No | MEATRD is trained on eight human normal breast ST datasets (i.e., 10x-h NB-{v03-v10}) and tested on eight human breast cancer (i.e., 10x-h BC-{A1-H1}) ST datasets. Reported metrics and standard deviations are averaged over five independent runs. For a fair comparison, the F1 score is calculated with the threshold matching the actual proportion of true anomalies (Shenkar and Wolf 2021). The paper describes which datasets are used for training and testing (normal breast vs. breast cancer datasets), and states metrics are averaged over five independent runs, but it does not specify explicit train/validation/test splits (e.g., percentages or counts within a single dataset) or provide details on how those runs partition the data. |
| Hardware Specification | No | No specific hardware details such as GPU models, CPU types, or memory specifications are mentioned in the paper for the experimental setup. |
| Software Dependencies | No | No specific software dependencies with version numbers (e.g., Python, PyTorch, TensorFlow, CUDA) are explicitly mentioned in the paper. |
| Experiment Setup | Yes | Table 2: Sensitivity analysis of hyperparameter in MEATRD across eight human breast cancer datasets. Default settings are marked in gray. Metric Parameter α Parameter β Embedding dimension Bottleneck dimension 0.9 0.5 0.1 0.9 0.5 0.1 128 256 512 16 64 256 AUC 0.678 0.723 0.709 0.654 0.723 0.699 0.705 0.723 0.721 0.723 0.715 0.682 F1 0.696 0.741 0.725 0.668 0.741 0.718 0.726 0.741 0.735 0.741 0.728 0.711 Metric Detection dimension MGDAT Layers MGDAT Layers 64 128 256 2 3 4 1 2 4 AUC 0.606 0.720 0.723 0.694 0.723 0.533 0.718 0.723 0.721 F1 0.623 0.732 0.741 0.719 0.741 0.565 0.730 0.741 0.737. Here, we conduct sensitivity analyses on eight 10x-h BC datasets to examine the effects of MEATRD s key hyperparameters, including α and β, which control the relative weights between gene and image modalities in Stage II and III; the dimensions of visual and gene embedding from Stage I, bottleneck embedding in Stage II, and the inputs to the one-classification classifier in Stage III; the number of MGDAT layers and its attention heads. |