The Medical Image Computing and Computer Assisted Interventions (MICCAI) international conference was held for the first time in the African continent, for its 27th edition in Marrakech, Morocco. This year, the presence of our lab was one of the largest in recent years with 8 members of ViCOROB (5 PhD students, 1 postdoc and 2 professors) attending in person at the Palmeraie Conference centre. In a nutshell, ViCOROB participated in 7 challenges (FeTA, BraTS, TopCow, MBH-Seg, TriALS, ACOUSLIC, EPVS), sent 5 workshop papers and held 2 oral presentations at a breast imaging workshop. The outcomes of this participation were positive: several top-performing places in 5 out of 7 challenges, as well as winning an Educational Challenge. Moreover, MICCAI manifested the fruits of the MAIA joint master’s degree, a Medical Imaging Erasmus Mundus program impulsed by the UdG and particularly the ViCOROB community, with 17 MAIA alumni.
The ViCOROB participation
Our students started to collect some important results from the beginning of the conference activities during the first Satellite events day, where several workshops and challenges results were announced. From the brain imaging research group, Valeria Abramova, 3rd year PhD student and 3 years in-a-row MICCAI attendee, participated in the International Brain Tumor Segmentation (BraTS) Challenge 2024 obtaining an outstanding 1st place in the meningioma radiotherapy segmentation task.
Clara Lisazo and Rachika E. Hamadache, 1st year PhD students, participated in the Topology-Aware Anatomical Segmentation of the Circle of Willis (TopCoW) challenge, for the second time, obtaining the 1st place in 3 tasks, bringing home once again 3 adorable Swiss toy cows directly from Zurich as awards. Rachika E. Hamadache also participated in the Fetal Tissue Annotation Challenge (FeTA) where she ranked 2nd place.
Ricardo Montoya Del Angel and Rachika E. Hamadache participated in Triphasic-aided Liver Lesion Segmentation challenge (TriALS) and they ranked 2nd in the Non-Contrast Lesion Segmentation task, winning a money award of 200$.
Cansu Yalçın, a second year PhD student obtained the 4th place in the MBH-Seg: Multi-class Brain Hemorrhage Segmentation in Non-contrast CT challenge winning a money award of 200$.
Ricardo Montoya Del Angel, a 2nd-year PhD student, had his paper accepted to the Deep-Breath: Deep Breast Workshop on AI and Imaging for Diagnostic and Treatment Challenges in Breast Care. His article, ELK: Enhanced Learning through Cross-Modal Knowledge Transfer for Lesion Detection in Limited-Sample Contrast-Enhanced Mammography Datasets, was among the featured works. Similarly, Melika Pooyan, with her paper MRI Breast Tissue Segmentation Using nnUNet for Biomechanical Modelling, and Hadeel Awwad, a 1st-year PhD student, with her work Graph Neural Networks for Modelling Breast Biomechanical Compression, also had their papers accepted under the supervision of Robert Martí and Eloy García. Both Melike Pooyan’s and Hadeel Awwad’s studies were selected for oral presentations, which were delivered by Robert Martí and Eloy García during the workshop event.
Kaouther Mouheb, Ricardo Montoya Del Angel and Nohemi Sofia Leon Contreras, MAIA graduates, won the MICCAI Educational Challenge with their submission named Introduction to Diffusion Model, winning the money award of 300$.
The conference highlights: Fair and accessible AI
The MICCAI conferences have long been a global gathering for researchers to showcase the latest advancements in medical imaging, shaping the future of AI applications in healthcare and setting new trends. This year’s theme, “Fair and Accessible AI,” was especially prominent. For the first time in MICCAI history, the conference took place in Africa, underscoring the significance of ensuring AI benefits all populations worldwide. A special session, “From AFRICAI to MICCAI,” highlighted the growing importance of education and awareness in AI, particularly across the African continent. Reflecting this focus, many studies presented during the conference tackled challenges related to limited datasets, compact models, and fairness, aiming to create AI systems that better serve underrepresented populations.
