The field of medical image analysis and vision-language understanding is rapidly evolving, with a focus on developing innovative models and techniques to improve disease diagnosis, treatment, and patient care. Recent research has explored the use of spatial transcriptomics, vision-language models, and multimodal learning to analyze medical images and extract relevant information. These approaches have shown promising results in improving the accuracy and efficiency of medical image analysis, enabling clinicians to make more informed decisions. Notably, the development of models that can adapt to different contexts and modalities has the potential to revolutionize medical image analysis, allowing for more precise and personalized treatment. Some particularly noteworthy papers in this area include the Scalable Generation of Spatial Transcriptomics from Histology Images via Whole-Slide Flow Matching, which proposes a flow matching generative model to predict spatial transcriptomics from whole-slide histology images, and the MedMoE framework, which incorporates a Mixture-of-Experts module to dynamically adapt visual representation based on the diagnostic context.
Advancements in Medical Image Analysis and Vision-Language Understanding
Sources
Scalable Generation of Spatial Transcriptomics from Histology Images via Whole-Slide Flow Matching
Query Nearby: Offset-Adjusted Mask2Former enhances small-organ segmentation
Full Conformal Adaptation of Medical Vision-Language Models
WoundAIssist: A Patient-Centered Mobile App for AI-Assisted Wound Care With Physicians in the Loop
MedMoE: Modality-Specialized Mixture of Experts for Medical Vision-Language Understanding
Spatial Transcriptomics Expression Prediction from Histopathology Based on Cross-Modal Mask Reconstruction and Contrastive Learning
Efficient Medical Vision-Language Alignment Through Adapting Masked Vision Models
Do MIL Models Transfer?
HER2 Expression Prediction with Flexible Multi-Modal Inputs via Dynamic Bidirectional Reconstruction
One Patient, Many Contexts: Scaling Medical AI Through Contextual Intelligence
Towards Scalable SOAP Note Generation: A Weakly Supervised Multimodal Framework
Improving Medical Visual Representation Learning with Pathological-level Cross-Modal Alignment and Correlation Exploration
Text to Image for Multi-Label Image Recognition with Joint Prompt-Adapter Learning
Anatomy-Grounded Weakly Supervised Prompt Tuning for Chest X-ray Latent Diffusion Models
IQE-CLIP: Instance-aware Query Embedding for Zero-/Few-shot Anomaly Detection in Medical Domain