Project summary

Recent advancements in hybrid classical-quantum information technologies are expected to have a huge impact on our daily life, from healthcare and renewable energy to information processing and secure communication. Their major opportunities arise in the very young but rapidly rising field of Quantum Machine Learning (QML), being at the intersection of Artificial Intelligence (AI) and quantum computing.

ThAI-MIA finds itself in this fermenting and exciting research field putting its application focus on Medical Image Analysis (MIA). Driven by a multidisciplinary spirit, our goal is to design and develop new hybrid AI algorithms relying on QML and cutting-edge Deep-Learning (DL) methodologies for MIA. During their whole life cycle, from design to use in the actual clinical practice, the AI algorithms will follow an ethics-by-design approach to be validated as trustworthy, i.e. legal, ethical, and robust. 

As the main use case, we will address the challenge of Autism Spectrum Disorder (ASD) early screening in preterm infants directly from images acquired in Neonatal Intensive Care Units (NICUs). 

Through ThAI-MIA we will also lay the scientific foundations of Quantum AI Ethics as a novel sub-field of AI Ethics that is poorly explored to date but is essential to meet the European Digital Strategy’s priority to develop trustworthy technology. ThAI-MIA will promote quality research that addresses the current open challenges in both technical (QML and MIA), clinical (preterm infants’ with ASD) and ethical (trustworthiness assessment) research fields, hence with a promising scientific, technological, social, medical, and economic impact in highly strategic areas for Italy and Europe.

The project has delivered on all three of its core objectives. On the technical side, novel hybrid quantum-classical architectures have been designed, theoretically analysed, with contributions validated both in simulation and on real quantum hardware, demonstrating competitive performance on clinically relevant tasks. On the clinical side, dedicated data acquisition campaigns in a real NICU setting have produced a substantially expanded dataset of preterm infants, enabling the development and rigorous evaluation of automated movement analysis tools for early ASD screening. On the ethical side, the project has produced a systematic Quantum AI Ethics framework, extending established AI ethics principles — explainability, fairness, and sustainability — to the specific challenges posed by quantum computing and hybrid quantum-classical architectures, contributing to the consolidation of this emerging research field.