University of Catania
Department of Mathematics and Computer Science
MED-IPLAB develops computational methods for medical image analysis with a strong focus on brain MRI, lesion segmentation, disease progression modelling, brain-age estimation, and data-driven clinical support systems.
News
We are thrilled to announce the Special Issue "Open the Brain: Horizons and Challenges in Brain Image Analysis" on Pattern Recognition Letters.
The issue focuses on critical aspects and innovative solutions in brain-focused medical imaging, showing how AI is overcoming challenges and opening new possibilities for diagnosis and treatment of neurological conditions.
We are excited to announce that our paper, "ICPR 2024 Competition on Multiple Sclerosis Lesion Segmentation - Methods and Results", has been accepted at the International Conference on Pattern Recognition (ICPR 2024).
The report summarizes the outcomes of the MSLesSeg Competition, which challenged participants to develop automatic and robust methods for multiple sclerosis lesion segmentation in MRI scans across longitudinal timepoints and previously unseen patient cohorts.
We are thrilled to announce that our paper "HypDeformNet: Edge-Deployable Deep Architecture with Jacobian-Stable Hyperbolic Deformation and Lipschitz Distillation for Immunotherapy Response Prediction" has been accepted at the 28th International Conference on Pattern Recognition (ICPR 2026).
This contribution introduces an advanced deep learning framework for CT-based prediction of response to PD-1/PD-L1 immunotherapy in metastatic cancer, combining hyperbolic deformations, Jacobian-stability constraints, and Lipschitz-controlled knowledge distillation.
Today we inaugurated the Medical Imaging course in the Master's Degree in Computer Science at the University of Catania.
The 6-CFU course is designed for students who want to understand biomedical imaging from clinical diagnosis to AI and machine learning applied to medical images, bringing into the classroom tools, datasets, and methodologies developed in IPLab research activities.
Imagine being able to predict how a person's brain will look 5 years from now, especially in the presence of neurodegenerative diseases. This means early diagnosis, better prognosis, and more targeted therapies.
This is exactly what our new work TADM-3D enables. TADM-3D is a 3D Temporally-Aware Diffusion Model that generates realistic future brain MRI volumes, capturing longitudinal brain progression at the patient level.
The MSLesSeg dataset for multiple sclerosis lesion segmentation is now publicly available, together with its official publication in Nature Scientific Data.
MSLesSeg marks a major milestone in MS research as the largest expert-annotated dataset currently available for MS lesion segmentation, combining multimodal MRI scans, expert-validated lesion masks, clinical data, and benchmark comparisons against state-of-the-art methods.
We are proud to announce that our paper titled "Dopamine-Inspired Neuro-Modulated Radiomic Pipeline for Computed-Tomography-Driven Prediction of Immunotherapy Response in Metastatic Cancer Treatment" has been accepted at the PHAROS: Adaptation, Fairness, Explainability in AI Medical Imaging Workshop in conjunction with ICCV 2025.
This work introduces a biologically-inspired computational framework that integrates dopamine-inspired neuro-modulation mechanisms into a hybrid spatio-temporal augmented pipeline for improved prediction of PD-L1 inhibitors immunotherapy response from CT scans.
Imagine being able to predict how a person's brain will look 5 years from now, especially in the presence of neurodegenerative diseases. This means earlier diagnosis, better prognosis, and more targeted therapies.
This is exactly what our new work TADM-3D enables.
TADM-3D is a 3D Temporally-Aware Diffusion Model that generates realistic future brain MRI volumes, capturing longitudinal brain progression at the patient level.
Novel Metric for Detecting Memorization in Generative Models for Brain MRI Synthesis accepted at IEEE/CVF WACV (Winter Conference on Application of Computer Vision) 2026.
Co-authored by Antonio Scardace, Lemuel Puglisi, Francesco Guarnera, Sebastiano Battiato, Daniele Ravi.
We are thrilled to announce that our paper titled "DeepSeg-MS: A 3D network based on hierarchical multi-scale learning for MRI multiple sclerosis lesion segmentation" has been accepted at the PHAROS Workshop, in conjunction with ICCV 2025 in Honolulu, Hawaii.
This work introduces DeepSeg-MS, a novel 3D deep learning architecture tailored for accurate and robust segmentation of multiple sclerosis lesions from multimodal MRI.
Research Areas
Lesion segmentation, MRI benchmarking, diffusion models, attention mechanisms, and evaluation on heterogeneous longitudinal datasets.
Brain progression prediction, atrophy mapping, protocol-robust brain-age estimation, and quantitative structural MRI analysis.
Deep learning workflows for COVID-19 imaging, radiogenomics, immunotherapy outcome prediction, and orthopedic CT analysis.
Public datasets, competition reports, research software, and collaborations connecting academic models with clinical environments.
