2026
Intermediate Layers Encode Optimal Biological Representations in Single-Cell Foundation Models
ICLR 2026 Workshop on Learning Meaningful Representations of Life (LMRL) · Poster
Biomedical engineering × machine learning
Vincenzo Yuto Civale
Machine Learning Researcher
PhD researcher building efficient and interpretable machine learning systems for biomedical data, with a focus on bioimaging, genomics, single-cell data, and multimodal learning.
01. Selected Work
Machine Learning
Single Cell Annotation Tool
An automated single-cell annotation tool designed to identify and classify cell populations from high-dimensional datasets. By decoding biological heterogeneity, the system simplifies data interpretation, allowing researchers to swiftly identify latent patterns and distinct cellular phenotypes.
- Pytorch
- Scanpy
- Transformers
Medical XR
AR Medical Imaging Plugin
A custom Unity plugin designed for real-time medical interoperability. It uses the OpenIGTLink protocol to establish a seamless bridge between surgical planning software and AR environments, enabling the low-latency streaming and visualisation of complex 3D volumes.
- Unity
- Hololens2
- IGTLink
Research Project
HistoFocus
An optimised Vision Transformer (ViT) framework for histopathological tumour diagnosis, which uses Upper Confidence Bound (UCB) scores to dynamically prune redundant tokens. This approach accelerates cancer detection workflows by 5x while maintaining state-of-the-art accuracy
- Pytorch
- Transformers
- Reinforcement Learning
02. Publications
2024
Realistic Texture Mapping of 3D Medical Models Using RGBD Camera for Mixed Reality Applications
Applied Sciences · 14(10), 4133
03. About
I am a PhD researcher working on machine learning for biomedical data. My research focuses on representation learning and multimodal methods across bioimaging and omics, with an emphasis on models that are efficient, interpretable, and reliable.
I started coding during the pandemic, which led to a two-year research fellowship at the Bioimaging Laboratory. There I worked on medical image analysis and mixed-reality applications for surgical planning and training, while also supervising thesis students.
I enjoy working at the boundary between research and engineering: understanding why a model works, building the infrastructure needed to evaluate it properly, and turning promising ideas into reproducible systems.
- Bioimaging
- Single-cell
- Multimodal ML
- Computer vision
- Genomics
- Efficient ML
04. Experience
PhD Student @ University of Siena
09/2025 - Present
- Researching machine learning methods for biomedical data, with a focus on multimodal representation learning across imaging and omics.
- Developing and evaluating models for genomics, single-cell data, and computational pathology.
- Working on efficient and reproducible pipelines for biomedical representation learning and model evaluation.
Research Fellow @ University of Florence
09/2023 - 06/2025
- Developed deep learning pipelines for medical image segmentation, with performance evaluated in collaboration with radiologists.
- Conducted research on neurological behavioral differences between digital natives and immigrants as part of a PRIN project.
- Contributed to medical image volume visualization and interoperability tools for HoloLens 2.
05. Contact
I am interested in research collaborations and engineering opportunities at the intersection of machine learning and biomedicine. The best way to reach me is by email.