Adaptive rank allocation improves 3D medical image prediction accuracy
Residual-Guided Adaptive Tucker Rank Allocation for Parameter-Efficient MCI Conversion Prediction
Machine Learning
Summary
Predicting whether people with mild memory problems will develop Alzheimer's is important but challenging. The authors developed a smart method to efficiently train 3D brain image models by adjusting how much detail the models keep in different parts. Their method uses leftover errors to decide where to focus training effort, leading to better prediction accuracy compared to standard methods. This approach uses fewer training parameters, making it more efficient for medical imaging tasks.
What this means in practice
- •For medical imaging engineers: Improve Alzheimer's disease progression prediction models using adaptive parameter allocation in 3D neural networks.
- •For healthcare ai developers: Develop more parameter-efficient AI tools to help process and analyze 3D medical imaging data under strict computational budgets.
Tested on one dataset.
Authors
Guanghua He, Hancan Zhu, Gaohang Yu, An Zhang
Abstract
Allocating limited trainable capacity across convolutional layers is a central challenge in parameter-efficient fine-tuning (PEFT) of pretrained three-dimensional (3D) networks. We propose Residual-Guided Adaptive Tucker Rank Allocation, which jointly selects input- and output-channel ranks under a fixed adapter-parameter budget while preserving the spatial kernel modes. To guide rank selection, we learn task-specific proxy updates from downstream training data and approximate them using candidate Tucker rank pairs. A perturbation analysis bounds the output discrepancy between the fixed proxy network and its Tucker approximation in terms of layerwise reconstruction residuals, motivating residual-based allocation. We normalize each squared residual by the corresponding proxy-update energy to compare relative approximation errors across layers. Dynamic programming minimizes their sum under a global budget constraint that accounts for both channel factors and the Tucker core. This yields a globally optimal allocation over the specified candidate rank pairs. We evaluate the method on the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort for predicting conversion from mild cognitive impairment to Alzheimer's disease within 36 months, using hippocampal segmentation pretraining. Across five random seeds, the proposed method achieves a mean area under the receiver operating characteristic curve (AUC) of 0.7706, compared with 0.7239 for fixed-ratio Tucker adaptation and 0.7430 for full fine-tuning. The total number of trainable parameters is limited to approximately 0.618 million, accounting for only 1.45% of the full model. Adaptive Tucker rank allocation thus provides an effective approach to parameter-efficient transfer learning in 3D medical imaging.