HyperSORT: Self-Organising Robust Training with hyper-networks
This work addresses biases in medical imaging segmentation for improved robustness, but it is incremental as it builds on existing hyper-network and latent space methods for bias handling.
The paper tackles the problem of heterogeneous biases in medical imaging datasets, such as erroneous labels and inconsistent labeling styles, which degrade deep segmentation network performance, and introduces HyperSORT, a framework that uses a hyper-network to predict UNet parameters from latent vectors representing image and annotation variability, resulting in a structured mapping that identifies systematic biases and erroneous samples, validated on two 3D abdominal CT datasets including a synthetically perturbed AMOS dataset and the large-scale TotalSegmentator dataset with real unknown biases.
Medical imaging datasets often contain heterogeneous biases ranging from erroneous labels to inconsistent labeling styles. Such biases can negatively impact deep segmentation networks performance. Yet, the identification and characterization of such biases is a particularly tedious and challenging task. In this paper, we introduce HyperSORT, a framework using a hyper-network predicting UNets' parameters from latent vectors representing both the image and annotation variability. The hyper-network parameters and the latent vector collection corresponding to each data sample from the training set are jointly learned. Hence, instead of optimizing a single neural network to fit a dataset, HyperSORT learns a complex distribution of UNet parameters where low density areas can capture noise-specific patterns while larger modes robustly segment organs in differentiated but meaningful manners. We validate our method on two 3D abdominal CT public datasets: first a synthetically perturbed version of the AMOS dataset, and TotalSegmentator, a large scale dataset containing real unknown biases and errors. Our experiments show that HyperSORT creates a structured mapping of the dataset allowing the identification of relevant systematic biases and erroneous samples. Latent space clusters yield UNet parameters performing the segmentation task in accordance with the underlying learned systematic bias. The code and our analysis of the TotalSegmentator dataset are made available: https://github.com/ImFusionGmbH/HyperSORT