AIMI @ MICCAI 2024 in Marrakesh

At this year’s “International Conference on Medical Image Computing and Computer Assisted Intervention” (MICCAI), AIMI researcher Jingna Qiu presented her latest work on prototype sampling, a novel method that uses image captions to guide selective annotation of whole slide images. The method identifies class prototypes for each task-specific class from large-scale histopathology image-caption datasets and then detecting unlabeled regions that resemble these prototypes. By applying this method, we achieved full annotation performance with less than 20% of data labeled in a single active learning iteration for both breast tumor segmentation and nuclei detection tasks.
To dive deeper into our research, take a look at our recent publications or get in touch!
It was truly inspiring to engage with researchers from across the globe and exchange ideas. A big thank you to the organizers for putting together such a great event!