Skip to content
View fruffini's full-sized avatar
9 followers · 71 following

Achievements

Achievement: Pull Shark

Achievements

Achievement: Pull Shark

Highlights

  • Pro

Block or report fruffini

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
fruffini/README.md

Hi there 👋, I'm Filippo

🧑🏻‍🎓 I'm a PhD student enrolled in the Italian National Program in Artificial Intelligence, jointly affiliated with Università Campus Bio-Medico di Roma (UCBM) and Umeå University (Department of Diagnostics and Intervention), supervised by Prof. Paolo Soda. My work sits within the DIGILUNG project — an AI-boosted digital twin for personalised lung cancer care.

🔬 My research focuses on trustworthy multimodal medical AI: survival prediction, radiology report generation, medical image synthesis, cross-modal retrieval, and the evaluation of vision–language models (shortcut learning, hallucination, robustness).

🌏 I spent research visiting periods at Shenzhen University and Umeå University (2025), and I presented our paper MDMT at MICCAI 2024.

📂 Selected Projects

  • Intermediate Multimodal Fusion — Systematic Review: A systematic review of intermediate fusion methods in multimodal deep learning for biomedical applications — taxonomy, notation, and a structured analysis of fusion strategies across 50+ works. Published in Image and Vision Computing (2025). 📄 Paper
  • MDMT: Multi-task / multi-modal framework for COVID-19 chest X-ray prognosis (AIforCOVID / BRIXIA). Presented at MICCAI 2024. 📄 Paper
  • NSCLC Multimodal Survival: Intermediate-fusion survival prediction for NSCLC combining tabular (NAIM + ODST) and foundation-model features. 📄 Paper - ArXiv
  • Prognosis Benchmark: A benchmark for clinical-outcome / prognosis prediction from chest X-rays, with standardised datasets, splits, and baselines for fair model comparison.
  • ShoViR: A benchmark for evaluating shortcut learning in radiology vision–language models, with dedicated resilience–sensitivity metrics (Submitted ad ECCV 2026). [📄 Paper](Under review for ECCV-26)
  • JoSS-DiT: Joint Synthesis and Segmentation via a Diffusion Transformer for medical image generation.

🛠️ Tools & Stack

Python · PyTorch · MONAI · Diffusers · Hugging Face · distributed training on HPC (SLURM, multi-GPU)

📝 Academic Service (Reviewer)

  • MICCAI 2026 — reviewer
  • Artificial Intelligence in Medicine (AIIM) — journal reviewer
  • ECCV — reviewer

🔥 News

  • [2025] Research visiting period at Umeå University.
  • [2025] Our systematic review on intermediate multimodal fusion appeared in Image and Vision Computing.
  • **[2024/2025]**Research visiting period at Shenzhen University.
  • [2024] Presented MDMT at MICCAI 2024.

📜 Find me online

Pinned Loading

  1. Handling-Missing-Modalities-in-Multimodal-Survival-Prediction-for-Non-Small-Cell-Lung-Cancer Handling-Missing-Modalities-in-Multimodal-Survival-Prediction-for-Non-Small-Cell-Lung-Cancer Public

    A missing-aware multimodal framework that fuses CT, Whole-Slide Histopathology, and clinical tabular data for survival prediction in unresectable stage II–III Non-Small Cell Lung Cancer (NSCLC) — w…

    Python 1 1

  2. PEFT_Prognosis PEFT_Prognosis Public

    The first systematic benchmark of fine-tuning strategies applied to CNNs and Foundation Models for COVID-19 prognosis prediction from chest X-rays, under realistic clinical constraints of data scar…

    Python 1

  3. Multi-Dataset-Multi-Task-Learning-for-COVID-19-Prognosis Multi-Dataset-Multi-Task-Learning-for-COVID-19-Prognosis Public

    A novel Multi-Dataset Multi-Task (MDMT) learning framework that predicts COVID-19 prognostic outcomes from chest X-rays by jointly training on two publicly available datasets with distinct but corr…

    Python 1

  4. CheXInstruct-MedGemma-Foundation-Finetuning-Framework CheXInstruct-MedGemma-Foundation-Finetuning-Framework Public

    Forked from cosbidev/CheXInstruct-MedGemma-Foundation-Finetuning-Framework

    A sick project for sick people

    Python

  5. GPU-efficiency-course-for-Deep-Learning-frameworks GPU-efficiency-course-for-Deep-Learning-frameworks Public template

    Shell 6

  6. Intermediate-Multimodal-Fusion-Bio Intermediate-Multimodal-Fusion-Bio Public

    Forked from cosbidev/Intermediate-Multimodal-Fusion-Bio

    A Systematic Review of Intermediate Fusion in Multimodal Deep Learning for Biomedical Applications.