A few of the ideas, topics, and commonplaces that have been gaining steam on arXiv during the last few months (yes, I do a lot of filtering to keep away most of the genAI stuff; explainer).
1. Performance Estimation Problem: Quoting Interpolation Constraints for Computing Worst-Case Bounds in Performance Estimation Problems: The Performance Estimation Problem (PEP) approach consists in computing worst-case performance bounds on optimization algorithms by solving an optimization problem: one maximizes an error criterion over all initial conditions allowed and all functions in a given class of interest. The maximal value is then a worst-case bound, and the maximizer provides an example reaching that worst case. Interesting idea!
Some recent articles:
- Computer-Assisted Design of Accelerated Composite Optimization Methods: OptISTA
- Analysis of Schedule-Free Nonconvex Optimization
- Numerical Design of Optimized First-Order Algorithms
- Learning Acceleration Algorithms for Fast Parametric Convex Optimization with Certified Robustness
- Linear Convergence of the Proximal Gradient Method for Composite Optimization Under the Polyak-Łojasiewicz Inequality and Its Variant
2. Surgical decision-making: I'm not at all comfortable with the type of AI being used in these papers, and I'd like to have numbers on what's actually the cognitive bottleneck (I wouldn't be surprised if it's way before or way after surgery), but it's an important problem nonetheless.
Some recent articles:
- Towards a better understanding of abdominal wall biomechanics: in vivo relationship between dynamic intra-abdominal pressure and magnetic resonance imaging measurements
- HealthiVert-GAN: A Novel Framework of Pseudo-Healthy Vertebral Image Synthesis for Interpretable Compression Fracture Grading
- Spatial-ORMLLM: Improve Spatial Relation Understanding in the Operating Room with Multimodal Large Language Model
- Now and Future of Artificial Intelligence-based Signet Ring Cell Diagnosis: A Survey
- Uncovering Neuroimaging Biomarkers of Brain Tumor Surgery with AI-Driven Methods
3. Preoperative planning: See above.
Some recent articles:
- Automated surgical planning with nnU-Net: delineation of the anatomy in hepatobiliary phase MRI
- Real-Time, Population-Based Reconstruction of 3D Bone Models via Very-Low-Dose Protocols
- Data-Efficient Learning for Generalizable Surgical Video Understanding
- Edge2Prompt: Modality-Agnostic Model for Out-of-Distribution Liver Segmentation
- Skip priors and add graph-based anatomical information, for point-based Couinaud segmentation