In recent years, the exponential growth of data availability has profoundly influenced the application of deep learning architectures in medical domains. Alongside the surge in data quantity, its complexity has also escalated, driven by advancements in signal processing techniques. Medical data is now richer and more intricate: high-resolution images, large-scale omics datasets with vast dimensionality, videos, unstructured data types such as graphs, etc. are extensively present and collected in the medical domain.

From one side, the advent of methods such as convolutional neural networks, graph neural networks, and transformers has enabled state-of-the-art performance across diverse input types, whether structured or unstructured. From the other side, while research on the methodological and theoretical aspects is advancing rapidly, the real-world integration of these models into clinical practice remains comparatively slow.

The Workshop aims to explore cutting-edge deep learning solutions designed to manage and use these diverse bio-medical data types, enabling personalized medicine and digital prevention. In addition, are encouraged papers discuss the needs and all the procedures to facilitate its use in clinical practice such as reliability, explainability, ethics, etc.

Possible topics of interest include (not limited to):

• From Medical Images to Video: Transitioning from static images to temporal and dynamic data, improving insights from sequential information.

• Deep Learning Architectures for Medical imaging: where transformers meet graph representation learning, explainability and neurosymbolic

• Enhancing Imaging Features with Multimodal Data: Integrating imaging data with other modalities like genomic, clinical, and environmental data to achieve more comprehensive predictive models.

• Explainable AI in biomedical data: adding explainability to enhance applicability of DL methodologies.

• Large language models in medical Applications

• Trustworthy AI in Medicine: methods and works on which the trustworthiness of models in models analysing biomedical imaging

• Development of new Deep Learning Architectures (Graph neural networks, convolutional neural networks, transformers, etc.) for biomedical data analysis

Important Dates

• Article Submission Opens: 31 January 2025

• Workshop Paper Submission Deadline: extended to 27 March 2025 20 March 2025

• Paper Acceptance Notification: 31 March 2025

• Camera-Ready Paper Submission Deadline: 1 May 2025

• Early Registration Deadline: 1 May 2025

• Workshop Date: 30 June 2025, from 2pm to 5:30pm, Room 1

PROGRAM AVAILABLE

Author Instructions

• Papers on IEEE Proceedings: the workshop allows full paper submissions, up to 8 pages according to IJCNN 2025 paper template. These papers will undergo a regular review process. If accepted, full papers will be published in the IJCNN 2025 proceedings. Papers must be submitted through the IJCNN 2025 CMT System and will undergo the same rigorous evaluation process as regular papers. 

It is mandatory to anonymize your paper. Any clues that may reveal author identities will be considered a violation and may lead to rejection. 

• Attendance fee: Workshop attendance is free for participants registered for the main conference, while workshop-only attendees will be charged a reduced registration fee. 

• Submission Guidelines : use the link https://cmt3.research.microsoft.com/IJCNN2025/Track/3/Submission/Create 

1. Select 2025 International Joint Conference on Neural Networks (IJCNN 2025) conference.

2. Your role should be "Author".

3. Create New Submission --> Workshop Track. 

4. Fill in all the required fields, but make sure that in the Subject Areas field, you select "New Trends for Artificial Intelligence in Medicine (TAIM): providing clinicians the best Deep Learning Architecture".

All article preparation guidelines must comply with the rules established by the IJCNN main conference Author Instructions.