Exploring the value of relabelling for multi-organ tumor segmentation from CT scans

Dzvinka Yarish, Nikita Fordui, Donatas Vaiciukevičius, Yuliia Siur

This project explores value of using labels generated from existing model for training better model for 3D CT scans data.

During project two datasets were used: KITS-19 and kidney part of DeepLesion dataset.

Link to presentation.