AI in medical image analysis
Contacts:
Stefano Giagu: stefano.giagu [at] uniroma1.it
Federico Giove: federico.giove [at] uniroma1.it
Alessandro Lonardo: alessandro.lonardo [at] roma1.infn.it
Alessandra Retico: alessandra.retico [at] pi.infn.it
Cecilia Voena: cecilia.voena [at] uniroma1.it
mailing list of the course: phdai-aimed [at] lists.infn.it
Course Topics
Introduction to medical imaging. X-ray CT
PET and MRI: principles and contrast generation
Images reconstruction. Noise in PET and MRI
Data representation and formats. Example pipeline for features extraction: from raw data to freesurfer cortical segmentation
Radiomic pipeline: radiomic features computation, feature selection and reduction, radiomic signatures
Hands-on on image feature exploration and representation
Machine learning techniques in medical image analysis
Hands-on on machine learning classification (neuro imaging MRI)
Deep learning-based classification and segmentation algorithms in medical image analysis
Data augmentation, and trasnfer learning techniques
Hands-on on deep learning methods for tissue classification
Hands-on on image lesion segmentation
Introduction to computers:hardware organization, firmware and software, performance definition and measurement.
Computer arithmetic: fundamenta data types, arithmetic and logical operations
Transformers and visual Transformers
Processor Architecture: functional units, registers, control unit, microprogramming; processing unit; pipelining, exceptions handling. Memory hierarchy: cache memory, virtual memory. Storage and I/O.
AI explainability and interpretability techniques and medical image applications
hands-on 1 on Visual Transformers
Overview of multicore systems, multiprocessors and clusters: parallel processing, classification, examples of many-core computing architectures (GPU), dedicated architectures for AI.
hands-on 2 on xAI
Final Exam
The final exam is composed of two parts:
a topic chosen by the candidate
a brief discussion of one of the other topics discussed during the course, chosen by the committee
News & Communications
Lectures will start on 7/3/23. See schedule for more details