PulseTrendingJournal ClubResearchersJournalsExplore
Instagram
HomeTrendingJournal ClubExplore
Synapse
⌘+K
Synapse
July 3, 2026TomographyOpen Access

Efficient Uncertainty Quantification in Medical Imaging via Mamba State Space Models

View Full Paper
Ask AI
Bookmark
Share

Authors

AGAli Güneş

Discussion

Loading...

Member takes

Overview

Randomized trial demonstrates improved uncertainty quantification in medical imaging, indicating promise for clinical settings.

Key Points

  • The aim is to develop a reliable and efficient method for uncertainty quantification in medical imaging workflows.
  • Developed UQ-Mamba for embedding uncertainty quantification in a Mamba state space model.
  • Generated uncertainty estimates during a single deterministic forward pass with minimal additional computational overhead.
  • Utilized error propagation principles and log-variance parameter propagation through the state transition matrix.
  • Achieved 89.71% accuracy with ECE = 0.0217 on OrganMNIST using only 466K parameters.
  • Improved Mamba baseline on PathMNIST by 2.42 percentage points, ECE = 0.1188 after temperature scaling.
  • Reached mAUC = 0.8196 on CheXpert chest radiographs.

Cite This Study

Ali Güneş (2026) studied this question.

synapsesocial.com/papers/6a47520e5c29257aa2578e5fhttps://doi.org/10.3390/tomography12070096
View Full Paper
Ask AI
Bookmark
Share