Home |

Multimodal data integration and fusion in medical imaging

Mon, 05 Oct 2026 00:00:00 +0000

The problem

Development and validation of a strategy to combine complementary modalities and/or feature representations, improving robustness, interpretability, and transferability across clinical tasks with heterogeneous or partially available data.

What you will do

As in every MSc thesis in the BEAR group, you will:

develop a novel module, neural architecture or learning strategy that addresses an open challenge in medical imaging (not a plain replication of existing methods);
validate it in one or more clinically relevant application domains, on established medical imaging benchmarks.

Possible application domains

Radiology (miscellaneous) · Neuro-radiology · Neuro-oncology · Ophthalmology

Example to follow

Modality-Aware Token Interaction (MATI): a module that structures multimodal interaction within a single token stream by partitioning the embedding space into modality-aligned subspaces, integrated into a neural backbone (e.g. UNETR) and validated on an established benchmark (e.g. BraTS for multimodal brain tumour segmentation), with optional generalisation to more backbones and benchmarks.

Missing-modality reconstruction in multimodal medical imaging

Mon, 05 Oct 2026 00:00:00 +0000

The problem

Development and validation of a model to reconstruct missing or corrupted imaging modalities from available ones, while preserving anatomically relevant information, supporting incomplete longitudinal or multimodal datasets without discarding patients with missing acquisitions.

What you will do

As in every MSc thesis in the BEAR group, you will:

develop a novel module, neural architecture or learning strategy that addresses an open challenge in medical imaging (not a plain replication of existing methods);
validate it in one or more clinically relevant application domains, on established medical imaging benchmarks.

Possible application domains

Radiology (miscellaneous) · Neuro-radiology · Neuro-oncology · Ophthalmology

Example to follow

Clinically conditioned medical image generation

Mon, 05 Oct 2026 00:00:00 +0000

The problem

Development and validation of a model to produce anatomically plausible medical images consistent with clinically relevant disease and treatment conditions.

What you will do

As in every MSc thesis in the BEAR group, you will:

develop a novel module, neural architecture or learning strategy that addresses an open challenge in medical imaging (not a plain replication of existing methods);
validate it in one or more clinically relevant application domains, on established medical imaging benchmarks.

Possible application domains

Radiology (miscellaneous) · Neuro-radiology · Neuro-oncology · Ophthalmology

Example to follow

Long-term spatial forecasting of disease progression

Mon, 05 Oct 2026 00:00:00 +0000

The problem

Development and validation of a model to forecast disease progression from longitudinal medical imaging, predicting the future spatial disease state and the corresponding image.

What you will do

As in every MSc thesis in the BEAR group, you will:

develop a novel module, neural architecture or learning strategy that addresses an open challenge in medical imaging (not a plain replication of existing methods);
validate it in one or more clinically relevant application domains, on established medical imaging benchmarks.

Possible application domains

Radiology (miscellaneous) · Neuro-radiology · Neuro-oncology · Ophthalmology

Example to follow

Medical image-to-text generation with visual grounding

Mon, 05 Oct 2026 00:00:00 +0000

The problem

Development and validation of a model that generates structured reports from medical images while grounding statements in image regions or visual evidence.

What you will do

As in every MSc thesis in the BEAR group, you will:

develop a novel module, neural architecture or learning strategy that addresses an open challenge in medical imaging (not a plain replication of existing methods);
validate it in one or more clinically relevant application domains, on established medical imaging benchmarks.

Possible application domains

Radiology (miscellaneous) · Neuro-radiology · Neuro-oncology · Ophthalmology

Example to follow

Quality awareness in medical imaging

Mon, 05 Oct 2026 00:00:00 +0000

The problem

Development and validation of a quality-aware pipeline that detects artifacts, corrupted inputs, or low-confidence cases and adapts preprocessing, inference, or review accordingly.

What you will do

As in every MSc thesis in the BEAR group, you will:

develop a novel module, neural architecture or learning strategy that addresses an open challenge in medical imaging (not a plain replication of existing methods);
validate it in one or more clinically relevant application domains, on established medical imaging benchmarks.

Possible application domains

Radiology (miscellaneous) · Neuro-radiology · Neuro-oncology · Ophthalmology

Example to follow

Multi-task learning in medical imaging

Mon, 05 Oct 2026 00:00:00 +0000

The problem

Development and validation of a framework for diagnostic detection, segmentation, and classification together with prognostic prediction, supporting interpretable decision-making.

What you will do

As in every MSc thesis in the BEAR group, you will:

develop a novel module, neural architecture or learning strategy that addresses an open challenge in medical imaging (not a plain replication of existing methods);
validate it in one or more clinically relevant application domains, on established medical imaging benchmarks.

Possible application domains

Radiology (miscellaneous) · Neuro-radiology · Neuro-oncology · Ophthalmology

Example to follow

FAIR-TT: Foundational AI for brain tumor segmentation

Mon, 01 Jan 0001 00:00:00 +0000

Work unit “Foundational AI for brain tumor segmentation: From prompt-based inference to clinical validation”, part of the FAIR project for the funding of technology transfer activities “FAIRMODE – Foundational AI for Reliable MODEls: Foundational methodologies for validation and automatic explanation in clinical and multimodal generative environments”.

Principal Investigator: Paolo Giorgini
Period: 1 June 2025 – 30 October 2025
Programme: FAIR – Future Artificial Intelligence Research, technology transfer activities
Project code: PE00000013 – CUP B533D22000980006

FAIR – Spoke 2

Mon, 01 Jan 0001 00:00:00 +0000

Research carried out by BEAR members within Spoke 2 of the PNRR project FAIR – Future Artificial Intelligence Research (PE00000013).

Selene Tomassini wins the L'Oréal-UNESCO For Women in Science Young Talents Italia 2026

Mon, 15 Jun 2026 00:00:00 +0000

On 15 June 2026 in Milan, Selene Tomassini received the L’Oréal-UNESCO “For Women in Science” Young Talents Italia 2026 award, granted by L’Oréal Italia and the Italian National Commission for UNESCO, for her project “mAInds: modality-aware AI for neuro-oncological decision support”.

Letizia Girardi's Master's thesis wins the Giulia Cecchettin Award and an Honorable Mention at the Bernardo Nobile Award

Thu, 26 Feb 2026 00:00:00 +0000

Letizia Girardi’s Master’s thesis has received two recognitions:

the Giulia Cecchettin Award for the Information Engineering macro-sector, in the competition promoted by the Collegio Ingegneri Venezia (26 February 2026). On 18 April 2026 she presented the awarded thesis at the 7th Excellence Engineers Conference;
an Honorable Mention at the 20th Bernardo Nobile Award, Area Science Park, Trieste.

A Concern-Oriented Governance Approach for Securing Agentic AI