Selected work

Representative Papers

LoRCA: LoRA Cycle Adaptation for Histology to HiP-CT Translation with DINOv3

Aug 07, 2026

This work addresses the challenge of nonlinear registration between histological sections and HiP-CT volumetric data, which arises from significant modality discrepancies. To tackle this without requiring paired training data, the authors propose a structure-preserving cross-modal image translation method. Leveraging a frozen DINOv2 backbone as a semantic structural anchor and modality-specific LoRA adapters, the approach enables efficient and generalizable cross-modal representation learning within a cycle-consistent adversarial training framework. This design effectively mitigates content drift while enhancing structural consistency. Experimental results demonstrate that the proposed method outperforms CycleGAN in terms of Fréchet Inception Distance (FID), mutual information, and edge preservation metrics, and substantially improves feature correspondence in downstream registration tasks.

0 citationsRead paper

Equal Marginal Power for Co-Primary Endpoints

Feb 20, 2026

This study addresses the common issue in randomized controlled trials where sample size determination for co-primary endpoints often results in imbalanced marginal statistical power across endpoints. To resolve this long-standing challenge, the authors propose a novel and systematic approach to sample size calculation grounded in multiple hypothesis testing theory. By strategically leveraging the degrees of freedom inherent in multiplicity adjustments, the method ensures equal marginal power for all co-primary endpoints. The proposed strategy demonstrates superior design properties and operating characteristics compared to existing ad hoc rules, offering a practical and efficient solution for balanced power allocation in trials with co-primary endpoints.

0 citationsRead paper

Recent publications

Latest Papers

LoRCA: LoRA Cycle Adaptation for Histology to HiP-CT Translation with DINOv3

Aug 07, 2026

This work addresses the challenge of nonlinear registration between histological sections and HiP-CT volumetric data, which arises from significant modality discrepancies. To tackle this without requiring paired training data, the authors propose a structure-preserving cross-modal image translation method. Leveraging a frozen DINOv2 backbone as a semantic structural anchor and modality-specific LoRA adapters, the approach enables efficient and generalizable cross-modal representation learning within a cycle-consistent adversarial training framework. This design effectively mitigates content drift while enhancing structural consistency. Experimental results demonstrate that the proposed method outperforms CycleGAN in terms of Fréchet Inception Distance (FID), mutual information, and edge preservation metrics, and substantially improves feature correspondence in downstream registration tasks.

0 citationsRead paper

Equal Marginal Power for Co-Primary Endpoints

Feb 20, 2026

This study addresses the common issue in randomized controlled trials where sample size determination for co-primary endpoints often results in imbalanced marginal statistical power across endpoints. To resolve this long-standing challenge, the authors propose a novel and systematic approach to sample size calculation grounded in multiple hypothesis testing theory. By strategically leveraging the degrees of freedom inherent in multiplicity adjustments, the method ensures equal marginal power for all co-primary endpoints. The proposed strategy demonstrates superior design properties and operating characteristics compared to existing ad hoc rules, offering a practical and efficient solution for balanced power allocation in trials with co-primary endpoints.

0 citationsRead paper