Research
Papers I've written, survived the review process for, and am reasonably proud of.
Tuan Truong, Melanie Dohmen, Sara Lorio, Matthias Lenga
An end-to-end multimodal framework for DICOM series classification that jointly models image content and acquisition metadata using bi-directional cross-modal attention and a sparse, missingness-aware metadata encoder based on learnable feature dictionaries.
Tuan Truong, Guillermo Jiménez-Pérez, Pedro Osório, Matthias Lenga
Exploring LLMs and vision-language models for zero-shot and few-shot detection of burned-in Protected Health Information in radiology images.
Tuan Truong, Ivo M. Baltruschat, Mark Klemens, Grit Werner, Matthias Lenga
A benchmark of AI-based pipelines — including YOLOv11, EasyOCR, and GPT-4o — for pixel-level PHI detection across multiple imaging modalities.
Tuan Truong, Matthias Lenga, Antoine Serrurier, Sadegh Mohammadi
Extending the FAIR4Cov framework to a broader set of respiratory conditions by fusing instance-level and representation-level audio features from multiple pretrained models.
Tuan Truong, Farnaz Khun Jush, Steffen Vogler, Matthias Lenga
A systematic benchmark of pretrained vision embeddings — supervised, self-supervised, and domain-specific — for near- and duplicate detection in radiology across multiple modalities.
Tuan Truong, Sadegh Mohammadi, Matthias Lenga
A study on the transferability of self-supervised vision features (SimCLR, SwAV, DINO) to medical image classification tasks across diverse imaging datasets.
Melanie Dohmen, Mark Klemens, Ivo M. Baltruschat, Tuan Truong, Matthias Lenga
A study evaluating similarity and quality metrics for assessing MR image-to-image translation, with implications for generative model benchmarking in medical imaging.
Melanie Dohmen, Tuan Truong, Ivo M. Baltruschat, Matthias Lenga
Identifying five common pitfalls in using reference-based metrics to evaluate synthetic medical images, with practical recommendations for researchers.