Related Experiment Video
Updated: Jun 26, 2026
Three-Dimensional Reconstruction for the Whole Lung with Early Multiple Pulmonary Nodules
Published on: October 13, 2023
Hierarchical Multiresolution Approaches for Dense Point-Level Breast Cancer Treatment Data
Shengde Liang1, Sudipto Banerjee, Sally Bushhouse
1Shengde Liang is Graduate Assistant, Sudipto Banerjee is Assistant Professor, and Bradley P. Carlin is Mayo Professor in Public Health, all in the Division of Biostatistics, School of Public Health, University of Minnesota, Minneapolis, MN 55455. Sally Bushhouse is Director, Minnesota Cancer Surveillance System, Minnesota Department of Health, 85 E. 7th Place, P.O. Box 64882, St. Paul, MN 55164-0882. Andrew Finley is Assistant Professor, Department of Forestry and Department of Geography, Michigan State University, East Lansing, MI 48824.
This study introduces a new computational method for analyzing spatial data, making complex statistical modeling feasible. It reveals insights into breast cancer treatment choices based on proximity to radiation facilities.
Area of Science:
- Statistical modeling
- Geostatistics
- Spatial analysis
Background:
- Geostatistical analysis faces computational challenges due to large covariance matrices.
- Hierarchical Bayesian models exacerbate these issues within Markov chain Monte Carlo (MCMC) algorithms.
Purpose of the Study:
- To develop a computationally feasible approach for spatial correlation modeling in complex settings.
- To enable full posterior inference for model parameters and spatial response surfaces.
- To investigate the relationship between distance to radiation facilities and breast cancer surgery choices.
Main Methods:
- A hierarchical multiresolution approach for modeling spatial correlation at two scales.
- Reducing high-dimensional matrix inversions to lower-dimensional, feasible computations.
- Application to dense point-referenced breast cancer data from Minnesota.
Main Results:
- The proposed method successfully integrates into the MCMC framework, overcoming computational hurdles.
- Full posterior inference was achieved for parameters and the spatial response surface.
- The study provides evidence regarding breast cancer treatment decisions in relation to facility proximity.
Conclusions:
- The hierarchical multiresolution method offers a practical solution for complex spatial data analysis.
- This approach is applicable to real-world health outcome research, such as cancer treatment disparities.
- Accurate spatial modeling is crucial for understanding and addressing geographic variations in healthcare access and choices.