Automated segmentation of organs and tumors from partially labeled 3D CT in MICCAI FLARE 2023 Challenge.

Andriy Myronenko, Dong Yang, Yufan He, Daguang Xu

Keywords: FLARE23, 3D CT, Segmentation, Auto3DSeg, MONAI

TL;DR: We use an automated segmentation method Auto3DSeg from MONAI for FLARE23 challenge submission.

Abstract: Automated organ and tumor segmentation from 3D CT is one of the key tasks in medical image analysis. In this work, we describe our solution to the FLARE 2023 challenge (team NVAUTO). We use an automated segmentation method Auto3DSeg available in MONAI. Our method achieves a 93% average organs class Dice score, and a 43% tumor class Dice score based on the 5-fold cross validation.

Submission Number: 27