Technical
Trustworthy & Explainable AI
Multi-task Deep Learning
EEG/ERP Signal Processing
Medical Image & 3D MRI Analysis
Computer Vision & Human Segmentation
Building intelligent, explainable, and trustworthy AI systems for mental health, neurological diagnostics, and the future of clinical practice.
Founded and directed by Dr. Faezeh Rohani at Üsküdar University, İstanbul, the MIRAI Laboratory is a multidisciplinary group dedicated to advancing explainable and trustworthy AI within medicine, neurotechnology, signal processing, and robotics.
We develop AI-driven methods to support the full spectrum of healthcare. Our core research interests include:
MIRAI integrates machine learning, deep learning, and biomedical signal processing to analyze diverse datasets, including clinical, behavioral, motor, and medical imaging data. We believe medical AI must be ethically responsible, interpretable, and reproducible. We actively welcome interdisciplinary collaborations with hospitals, research centers, and global digital health teams to create real-world impact.
Our work organizes around three interconnected pillars — the technical foundation, the clinical applications it serves, and the educational mission that carries it forward.
Trustworthy & Explainable AI
Multi-task Deep Learning
EEG/ERP Signal Processing
Medical Image & 3D MRI Analysis
Computer Vision & Human Segmentation
Pediatric Mental Health & ADHD
Neurofeedback & Motor Behavior
Cognitive-Motor Interventions
Knee Osteoarthritis & MSK Imaging
Clinical Validation of AI Models
Graduate & Undergraduate Mentorship
AI in Medicine Webinar Series
Open-Source Research & Datasets
International Collaborative Projects
Interdisciplinary Student Training
Healthcare AI is evolving faster than the frameworks meant to govern it. At MIRAI we believe a model is only as good as the trust it earns — from the patient whose life it touches, the clinician who must defend it in practice, and the regulator who must approve it. That belief drives every step of how we work.
We start with real clinical questions, not benchmark scores. Every project at MIRAI begins with a clinical or educational need — an unmet diagnostic gap, a workflow barrier, a population that's been overlooked — and ends only when the resulting AI tool can be openly evaluated by the people who will actually use it. Along the way we invest heavily in explainability, uncertainty estimation, robustness under data shift, and transparent reporting, so that domain experts can understand why a model is doing what it does, not just what it predicts.
We also believe that trustworthy AI must be reproducible AI. Whenever possible we release annotated datasets, source code, trained model weights, and lightweight software applications alongside our publications — lowering the barrier for low-resource hospitals, rural clinics, and student researchers to build on what we publish. Open science is not an afterthought at MIRAI; it is part of the methodology.
MIRAI is, at its heart, a teaching laboratory. We host an active community of PhD and M.Sc students, research assistants, and undergraduate contributors from across Üsküdar University and partner institutions worldwide. Students lead their own projects, co-author publications, present at international venues, and grow into independent researchers under structured mentorship.
Through the AI in Medicine Webinar Series — co-hosted with the Üsküdar University Department of Software Engineering — we bring leading voices in clinical AI to a global audience, free and open to anyone. Past and upcoming sessions explore topics from responsible AI in healthcare governance to explainable models in musculoskeletal care, with speakers drawn from major academic and clinical institutions.
Our international partnerships extend the reach of every project: collaborations with the HExAI Research Laboratory at the University of Pittsburgh on explainable AI for musculoskeletal imaging, with the Brain and Trauma Foundation in Switzerland on EEG/ERP and neurofeedback research, and with peer institutions across Europe, North America, and Asia. These partnerships are how good ideas travel — and how MIRAI students become global researchers.
MIRAI's vision is a future where artificial intelligence quietly augments every layer of clinical care — making earlier diagnoses possible, supporting clinicians instead of replacing them, and reaching patients in places that the traditional healthcare system has historically left behind. We are building toward that future one careful, trustworthy model at a time, and we welcome students, clinicians, and collaborators who share that ambition to join us.