SGRNet: Spatially Guided Radiology Network for Structured Radiological Reporting of Head and Neck Cancer
为解决头颈癌CT图像报告自动生成中的幻觉风险和数据稀缺问题，提出SGRNet，利用空间引导和结构化报告方法提高准确性和安全性。
0 citationsRead paper
Selected work
Representative Papers
MK-ResRecon: Multi-Kernel Residual Framework for Texture-Aware 3D MRI Refinement from Sparse 2D Slices
This study addresses the challenges of prolonged MRI scan times, susceptibility to motion artifacts, and the need for repeated acquisitions by proposing a two-stage framework—MK-ResRecon and IdentityRefineNet3D—that enables high-quality 3D brain MRI reconstruction from only 12.5% sparse 2D axial slices. The method uniquely integrates a multi-kernel texture-aware loss with an end-to-end 3D joint optimization strategy: the former employs a multi-kernel convolutional residual network to predict missing slices while preserving fine anatomical details, and the latter fuses original and predicted slices to produce smooth, coherent 3D volumes. Validated on a large-scale, heterogeneous dataset of T1-weighted contrast-enhanced brain MRIs, the approach achieves hallucination-free, high-fidelity reconstructions that generalize clinically under extremely sparse input conditions, substantially reducing scan duration and enhancing patient comfort.
0 citationsRead paper
A Proposed Biomedical Data Policy Framework to Reduce Fragmentation, Improve Quality, and Incentivize Sharing in Indian Healthcare in the era of Artificial Intelligence and Digital Health
This work addresses the critical barriers hindering AI and digital health advancement in India—namely, fragmented biomedical data, inconsistent data quality, and insufficient incentives for data sharing. To overcome these challenges, the study proposes a multi-layered incentive policy framework that systematically integrates Shapley value–based benefit allocation, data paper recognition, and open data metrics into the national research evaluation system. The framework further incorporates federated learning, rigorous data quality assessment, structured peer review, and compliance mechanisms aligned with Indian regulations such as the Digital Personal Data Protection Act (DPDPA), National Data Sharing and Accessibility Policy (NDSAP), and Biotech-PRIDE guidelines. By harmonizing technical and institutional approaches, this initiative fosters a high-quality, interoperable, and sustainable biomedical data ecosystem, laying the foundational infrastructure for AI-driven healthcare research in India.
0 citationsRead paper
Multi-RADS Synthetic Radiology Report Dataset and Head-to-Head Benchmarking of 41 Open-Weight and Proprietary Language Models
Automatically mapping standardized RADS classifications from narrative radiology reports faces challenges including complex guidelines, constrained outputs, and a lack of systematic evaluation. This work introduces RXL-RADSet, the first synthetic multimodal radiology report benchmark covering ten RADS standards, comprising 1,600 radiologist-validated samples. The authors conduct a head-to-head evaluation of 41 open-source small language models (0.135–32B parameters) alongside GPT-5.2 under a unified prompting strategy. Results show that GPT-5.2 achieves 99.8% validity and 81.1% accuracy with guided prompting, while open-source models of 20–32B parameters reach approximately 99% validity and over 70% accuracy, demonstrating significant performance gains with scale. Guided prompting consistently outperforms zero-shot settings across all evaluated models.
0 citationsRead paper
Predicting Genetic Mutations from Single-Cell Bone Marrow Images in Acute Myeloid Leukemia Using Noise-Robust Deep Learning Models
This study addresses the dual challenges of inaccurate clinical labels and image noise in single-cell bone marrow image-based mutation prediction for acute myeloid leukemia (AML). We propose a two-stage noise-robust learning framework: (1) a deep learning binary classifier achieves 90% accuracy in identifying myeloid blasts; (2) a weakly supervised four-class mutation classifier—trained on coarse, patient-level mutation labels—demonstrates robustness to 20% label noise, attaining 85% mutation prediction accuracy. The method integrates pathologist-in-the-loop validation, noise-robust optimization, and interpretability design. To our knowledge, it is the first approach to directly infer AML driver mutations from blast cell images without per-cell molecular annotations, thereby balancing diagnostic reliability and clinical interpretability. This work establishes a novel paradigm for imaging genomics.
0 citationsRead paper
Recent publications
Latest Papers
SGRNet: Spatially Guided Radiology Network for Structured Radiological Reporting of Head and Neck Cancer
为解决头颈癌CT图像报告自动生成中的幻觉风险和数据稀缺问题，提出SGRNet，利用空间引导和结构化报告方法提高准确性和安全性。
0 citationsRead paper
MK-ResRecon: Multi-Kernel Residual Framework for Texture-Aware 3D MRI Refinement from Sparse 2D Slices
This study addresses the challenges of prolonged MRI scan times, susceptibility to motion artifacts, and the need for repeated acquisitions by proposing a two-stage framework—MK-ResRecon and IdentityRefineNet3D—that enables high-quality 3D brain MRI reconstruction from only 12.5% sparse 2D axial slices. The method uniquely integrates a multi-kernel texture-aware loss with an end-to-end 3D joint optimization strategy: the former employs a multi-kernel convolutional residual network to predict missing slices while preserving fine anatomical details, and the latter fuses original and predicted slices to produce smooth, coherent 3D volumes. Validated on a large-scale, heterogeneous dataset of T1-weighted contrast-enhanced brain MRIs, the approach achieves hallucination-free, high-fidelity reconstructions that generalize clinically under extremely sparse input conditions, substantially reducing scan duration and enhancing patient comfort.
0 citationsRead paper
A Proposed Biomedical Data Policy Framework to Reduce Fragmentation, Improve Quality, and Incentivize Sharing in Indian Healthcare in the era of Artificial Intelligence and Digital Health
This work addresses the critical barriers hindering AI and digital health advancement in India—namely, fragmented biomedical data, inconsistent data quality, and insufficient incentives for data sharing. To overcome these challenges, the study proposes a multi-layered incentive policy framework that systematically integrates Shapley value–based benefit allocation, data paper recognition, and open data metrics into the national research evaluation system. The framework further incorporates federated learning, rigorous data quality assessment, structured peer review, and compliance mechanisms aligned with Indian regulations such as the Digital Personal Data Protection Act (DPDPA), National Data Sharing and Accessibility Policy (NDSAP), and Biotech-PRIDE guidelines. By harmonizing technical and institutional approaches, this initiative fosters a high-quality, interoperable, and sustainable biomedical data ecosystem, laying the foundational infrastructure for AI-driven healthcare research in India.
0 citationsRead paper
Multi-RADS Synthetic Radiology Report Dataset and Head-to-Head Benchmarking of 41 Open-Weight and Proprietary Language Models
Automatically mapping standardized RADS classifications from narrative radiology reports faces challenges including complex guidelines, constrained outputs, and a lack of systematic evaluation. This work introduces RXL-RADSet, the first synthetic multimodal radiology report benchmark covering ten RADS standards, comprising 1,600 radiologist-validated samples. The authors conduct a head-to-head evaluation of 41 open-source small language models (0.135–32B parameters) alongside GPT-5.2 under a unified prompting strategy. Results show that GPT-5.2 achieves 99.8% validity and 81.1% accuracy with guided prompting, while open-source models of 20–32B parameters reach approximately 99% validity and over 70% accuracy, demonstrating significant performance gains with scale. Guided prompting consistently outperforms zero-shot settings across all evaluated models.
0 citationsRead paper
Predicting Genetic Mutations from Single-Cell Bone Marrow Images in Acute Myeloid Leukemia Using Noise-Robust Deep Learning Models
This study addresses the dual challenges of inaccurate clinical labels and image noise in single-cell bone marrow image-based mutation prediction for acute myeloid leukemia (AML). We propose a two-stage noise-robust learning framework: (1) a deep learning binary classifier achieves 90% accuracy in identifying myeloid blasts; (2) a weakly supervised four-class mutation classifier—trained on coarse, patient-level mutation labels—demonstrates robustness to 20% label noise, attaining 85% mutation prediction accuracy. The method integrates pathologist-in-the-loop validation, noise-robust optimization, and interpretability design. To our knowledge, it is the first approach to directly infer AML driver mutations from blast cell images without per-cell molecular annotations, thereby balancing diagnostic reliability and clinical interpretability. This work establishes a novel paradigm for imaging genomics.
0 citationsRead paper