Search papers, labs, and topics across Lattice.
This paper introduces a diffusion-based framework with LoRA adapters to synthesize realistic lung nodules in chest radiographs (CXRs) with fine-grained control over radiological characteristics. The method addresses size and shape control through mask-conditioned training and characteristic control using separate LoRA modules for specific features. To enable balanced integration of multiple features, the authors propose an orthogonality loss term during LoRA composition training to mitigate overlapping attention regions and non-orthogonal parameter spaces. Experiments demonstrate improved downstream nodule detection and radiologist evaluations confirm the fine-grained controllability of the generated nodules, outperforming existing nodule generation approaches.
Synthesizing lung nodules with diffusion models just got a whole lot more controllable, thanks to a novel LoRA merging strategy that tackles overlapping attention and non-orthogonal parameter spaces.
Early detection of lung cancer in chest radiographs (CXRs) is crucial for improving patient outcomes, yet nodule detection remains challenging due to their subtle appearance and variability in radiological characteristics like size, texture, and boundary. For robust analysis, this diversity must be well represented in training datasets for deep learning based Computer-Assisted Diagnosis (CAD) systems. However, assembling such datasets is costly and often impractical, motivating the need for realistic synthetic data generation. Existing methods lack fine-grained control over synthetic nodule generation, limiting their utility in addressing data scarcity. This paper proposes a novel diffusion-based framework with low-rank adaptation (LoRA) adapters for characteristic controlled nodule synthesis on CXRs. We begin by addressing size and shape control through nodule mask conditioned training of the base diffusion model. To achieve individual characteristic control, we train separate LoRA modules, each dedicated to a specific radiological feature. However, since nodules rarely exhibit isolated characteristics, effective multi-characteristic control requires a balanced integration of features. We address this by leveraging the dynamic composability of LoRAs and revisiting existing merging strategies. Building on this, we identify two key issues, overlapping attention regions and non-orthogonal parameter spaces. To overcome these limitations, we introduce a novel orthogonality loss term during LoRA composition training. Extensive experiments on both in-house and public datasets demonstrate improved downstream nodule detection. Radiologist evaluations confirm the fine-grained controllability of our generated nodules, and across multiple quantitative metrics, our method surpasses existing nodule generation approaches for CXRs.
