Multiclass Segmentation of Brain Intraoperative Ultrasound Images with Limited Data

During tumor resection surgery, intraoperative ultrasound images of the brain show anatomical structures like the sulci, falx cerebri and tentorium cerebelli, as well as the tumor. After resection started, the resection cavity is also visible. These elements help with the localization and tumor resection, and can be used to register the preoperative MRI to intraoperative images, to compensate for the tissue deformation occurring during surgery. In this work, we compare single- and multi-class segmentation models for the sulci, falx cerebri, tumor, resection cavity and ventricle. We present strategies to overcome the severe class imbalance in the training data, and train a model with limited data. We show that a multi-class model may leverage inter-class spatial relationships and produce more accurate results than single-class models.