News
LinkedIn-sourced updates and announcements from Arco Lab.
Updates
News
Read the latest news from ArCo Lab, including research milestones, academic achievements, project updates, and activities across the group.
ArCo Lab presents PET-CT tumor segmentation research at ISBI
ArCo Lab shares a new ISBI contribution presented by Elena Mulero and developed with Paolo Soda, Matteo Tortora, and collaborators. The work introduces a lightweight context-gated framework for PET-CT lung tumor segmentation, integrating complementary multimodal information for medical image analysis.
Open on LinkedIn
Congratulations to Dr. Fatih Aksu on earning his PhD!
ArCo Lab celebrates Dr. Fatih Aksu for completing his PhD. The update highlights his work on deep learning-driven classification of NSCLC histological subtypes using PET and CT imaging, spanning small-data triplet networks, federated learning, PET-CT fusion, and 3D medical foundation models.
Open on LinkedIn
Ludovica Pompilio presents medical AI security research
ArCo Lab highlights Ludovica Pompilio's presentation at an ACN event, focused on trustworthy AI in healthcare. The project addresses medical image security and privacy-preserving synthetic data generation for clinical AI support.
Open on LinkedIn
XGeM for multimodal medical data generation
ArCo Lab shares a new Computerized Medical Imaging and Graphics article introducing XGeM, a multi-prompt foundation model for multimodal medical data generation. The work supports flexible conditioning, clinically consistent outputs, and synthetic medical data generation.
Open on LinkedInGAN ensembles for diverse medical data generation
ArCo Lab presents a new study on GAN ensembles for medical data generation. The work combines multiple generative models selected through multi-objective optimisation to improve synthetic image quality, diversity, and targeted data augmentation.
Open on LinkedInMask-aware transformers for incomplete volumetric imaging
ArCo Lab shares a new conference paper introducing MAViT, a mask-aware transformer for learning from incomplete volumetric medical data without interpolation or synthetic reconstruction, tested on brain MRI for Alzheimer's disease classification.
Open on LinkedInShow more
New paper on multimodal virtual biopsy
ArCo Lab presents a new study on multimodal virtual biopsy in breast cancer, combining FFDM and CESM imaging with generative AI to support lesion classification when contrast-enhanced data are unavailable.
Open on LinkedIn
Happy holidays from ArCo Lab
ArCo Lab shares a holiday message to its community, wishing a joyful season and a new year focused on ideas, collaboration, and innovation.
Open on LinkedIn
ArCo Lab at FAIR General Conference 2025
ArCo Lab took part in the FAIR General Conference 2025, with Paolo Soda joining a panel broadcast by ANSA and ArCo researchers presenting recent work on AI and medical imaging.
Open on LinkedInBenchmarking foundation models for COVID-19 prognosis
ArCo Lab shares a study comparing CNNs, foundation models, and parameter-efficient fine-tuning for COVID-19 prognosis prediction from chest X-rays under limited, imbalanced, and clinically complex data conditions.
Open on LinkedIn
Congratulations to Dr. Camillo Maria Caruso
ArCo Lab celebrates Camillo Maria Caruso for completing his PhD. His research addresses resilient medical AI models designed to remain reliable when healthcare data are incomplete or partially missing.
Open on LinkedIn
Irene Iele receives Ingenio al Femminile award
ArCo Lab celebrates Irene Iele for receiving the Giulia Cecchettin Award in Biomedical Engineering. Her thesis focuses on robust AI for diagnostic imaging, medical image translation, and distribution shift mitigation.
Open on LinkedInvMambaX for PET-CT lung tumor segmentation
ArCo Lab presents vMambaX, a lightweight Visual Mamba-based framework for PET-CT lung tumor segmentation, using context-gated cross-modal perception to fuse anatomical and functional imaging information.
Open on LinkedInTGCom24 features the launch of XGeM
ArCo Lab shares TGCom24 coverage of XGeM, a generative AI model for multimodal healthcare data. The work supports missing-view reconstruction, report generation, and synthetic medical data creation for clinical AI support.
Open on LinkedIn
ArCo Lab joins European Researchers' Night
ArCo Lab took part in the European Researchers' Night at Università Campus Bio-Medico di Roma, presenting XGeM and PyTrack to visitors as examples of AI for healthcare and sustainable mobility.
Open on LinkedInNew GAN-based method for low-dose CT denoising
ArCo Lab presents a new study on multi-scale texture loss for CT denoising with GANs. The work aims to improve low-dose CT image quality by preserving subtle and complex textural details.
Open on LinkedInArCo Lab awarded at MICCAI challenge
ArCo Lab reports a second-place result at a MICCAI challenge in Daejeon. The team presented a framework for generating 3D CT volumes from radiology reports using latent diffusion and vision-language pretraining.
Open on LinkedIn
Camillo Caruso presents incomplete 3D imaging research
ArCo Lab shares a presentation by Camillo Maria Caruso at an ICIAP workshop on AI and radiomics. The work introduces a transformer-based approach for learning directly from incomplete 3D medical scans.
Open on LinkedInMultimodal AI for atherosclerosis progression assessment
ArCo Lab presents a new study on multimodal ensemble learning for atherosclerotic disease progression. The work integrates imaging and clinical information to support vascular aging assessment and sub-clinical plaque detection.
Open on LinkedInMARIA for incomplete healthcare data
ArCo Lab presents MARIA, a multimodal transformer model for incomplete healthcare data. The method handles missing features and missing modalities, showing robust performance across diagnostic and prognostic tasks.
Open on LinkedInDeep learning for exoskeleton locomotion analysis
ArCo Lab shares a Frontiers study on deep learning for human locomotion analysis in lower-limb exoskeletons. The work uses wearable sensor data for terrain classification, locomotion parameter estimation, and sensor selection.
Open on LinkedInDoctor-in-the-loop AI for NSCLC response prediction
ArCo Lab presents a new explainable multi-view deep learning framework for predicting pathological response in non-small cell lung cancer. The work introduces a doctor-in-the-loop training paradigm where clinical experts guide model attention toward meaningful anatomical and lesion-level regions.
Open on LinkedIn
ArCo Lab presents recent research at IJCNN 2025
ArCo Lab took part in IJCNN 2025, presenting several contributions at the intersection of AI, medical imaging, generative models, and clinical applications, alongside keynote talks and workshop activities.
Open on LinkedIn
ArCo Lab at CBMS 2025 in Madrid
ArCo Lab joined CBMS 2025 in Madrid, presenting recent work on generative AI, digital twins, foundation models, and medical imaging. The post highlights multiple contributions and collaborations across healthcare AI research.
Open on LinkedIn
Welcome Ludovica Pompilio and Irene Iele
ArCo Lab welcomes Ludovica Pompilio and Irene Iele as new PhD students. Their research will address generative AI for privacy-preserving medical data sharing, imaging synthesis, time series prediction, and personalized healthcare.
Open on LinkedInArCo Lab announces its new identity
ArCo Lab announces its new identity at Università Campus Bio-Medico di Roma. The acronym reflects the lab's mission to bridge disciplines, connect data and decisions, and support AI innovation across research, healthcare, and industry.
Open on LinkedIn