Perceptual Analysis of Medical Images for Reliable Diagnosis

Organizers

Meriem Outtas (INSA Rennes, France)
Maria Martini (Kingston University, United Kingdom)
Aladine Chetouani (University of Orleans, France)

Abstract

The advent of AI in general and in the medical image field in particular raises a number of questions, the main one being reliability. Since the diagnosis heavily relies on medical imaging techniques and the ability of healthcare professionals or computer vision algorithms to visualize or analyze internal organs and tissues, the concept of good diagnostic quality takes on its full meaning. Another challenge of great relevance in the context of medical data and medical imaging is security: patient records and imaging are vulnerable to unauthorized access and alteration which could lead to incorrect diagnoses or treatment decisions. AI models trained on medical imaging require massive quantities of high-quality images to obtain accurate and reliable results in terms of diagnosis. Furthermore, it is important to guarantee data availability while preserving patient privacy. This special session considers the issue of the quality of medical images and its impact on early, accurate and confident diagnosis.

Topics of interest include, but are not limited to:

Task based quality assessment;
Perceptual evaluation of the image quality;
Visual attention mechanism in diagnosis tasks;
Image databases including different modalities (PET/, SPECT, CT, MRI, US, optical) of real or synthesized images generated by generative models;
Methods that ensure the privacy, confidentiality and integrity of medical images;
Any approach involving a possible change in the visual and diagnostic quality of medical imaging.

Artificial Intelligence in Dentistry (AID)

Organizers

Douglas Teodoro (University of Geneva, Switzerland)
Julian Leprince (University of Geneva, Switzerland)

Abstract

This special session aims to gather research at the intersection of deep learning and dentistry, particularly for multimodal analysis tasks (visual, textual, etc.). This encompasses analyses of various data types, including dental radiograph analysis, intraoral scans, cone beam CT scans, and photograph analysis, alongside electronic dental record analysis. In particular, we are interested in algorithms and methods that address the challenges of longitudinal dental image analyses for the development of intelligent diagnostic systems, treatment planning algorithms, patient management tools, and educational resources tailored to the specific needs of dental professionals and patients alike.

Topics of interest include, but are not limited to:

Detection and characterization of dental caries progression.
Early detection of periodontal disease.
Monitoring orthodontic treatment progress.
Fusion of clinical data (radiographs, electronic health records) with patient-specific information for personalized treatment planning.
Risk stratification and decision support.
Personalized oral hygiene recommendations.
Simulation for dental education and training programs.
Bias mitigation in dental diagnosis and treatment planning.
Improved transparency and explainability.
The development of frameworks for the responsible deployment of AI in clinical dentistry.