analiza aiqbench
This study challenges the default assumption that 3D models are superior for volumetric medical imaging, demonstrating that 2.5D CNNs often provide a more stable and computationally efficient trade-off than full 3D representations or Vision Transformers in lung CT tasks. The authors present a resource-performance frontier highlighting specific failure modes, such as threshold instability in 3D CNNs and degenerate predictions in transformers, which are critical considerations for deploying models in class-imbalanced screening scenarios. Rather than claiming architectural superiority, the work offers a pragmatic framework for selecting input dimensionality based on the specific balance between discrimination accuracy, stability, and computational cost required for a given application.
When is 3D Worth It? A Resource-Performance Frontier for CNNs and Transformers in Lung CT
arXiv:2606.06950v1 Announce Type: cross Abstract: Three-dimensional models are widely assumed preferable for volumetric medical imaging, yet their practical value depends on whether performance gains justify added computational cost and complexity. Rather than proposing a new architecture, we study how input dimensionality (2D, 2.5D, 3D) affects model behavior across convolutional neural networks (CNNs) and Vision Transformers (ViTs) under a fixed training protocol. Using a leakage-free NLST cohort (n = 1,977) with supporting LIDC-IDRI data, we find that the 2.5D CNN offers the most favorable discrimination-stability trade-off in our comparison (ROC-AUC 0.682, 95% CI [0.546, 0.799]) with a stable operating point. In contrast, 3D CNNs show threshold instability, and transformers exhibit degenerate predictions, such as all-positive predictions. Confidence intervals are wide and overlapping, so we present these results as a controlled resource-performance frontier and a failure-mode taxonomy rather than as definitive superiority claims. For class-imbalanced lung cancer screening classification, 2D and 2.5D inputs provide a more reliable trade-off between performance, stability, and computational efficiency than full 3D representations.
Title:When is 3D Worth It? A Resource-Performance Frontier for CNNs and Transformers in Lung CT
Abstract:Three-dimensional models are widely assumed preferable for volumetric medical imaging, yet their practical value depends on whether performance gains justify added computational cost and complexity. Rather than proposing a new architecture, we study how input dimensionality (2D, 2.5D, 3D) affects model behavior across convolutional neural networks (CNNs) and Vision Transformers (ViTs) under a fixed training protocol. Using a leakage-free NLST cohort (n = 1,977) with supporting LIDC-IDRI data, we find that the 2.5D CNN offers the most favorable discrimination-stability trade-off in our comparison (ROC-AUC 0.682, 95% CI [0.546, 0.799]) with a stable operating point. In contrast, 3D CNNs show threshold instability, and transformers exhibit degenerate predictions, such as all-positive predictions. Confidence intervals are wide and overlapping, so we present these results as a controlled resource-performance frontier and a failure-mode taxonomy rather than as definitive superiority claims. For class-imbalanced lung cancer screening classification, 2D and 2.5D inputs provide a more reliable trade-off between performance, stability, and computational efficiency than full 3D representations.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2606.06950 [cs.CV]
(or arXiv:2606.06950v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2606.06950 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 276 cuvinte
Subiecte conexe
Microsoft’s AI chief says superintelligence is near, but won’t take your job
The Verge · 8 iun. 2026
Amazon Research Awards recipients announced
Amazon Science · 27 mai 2026
Humanity’s Last Invention — Richard Socher of Recursive
Latent Space · 14 sept. 2026
Links #3: 2026/06 Part 1
LessWrong · 15 iun. 2026