Utilizing Uncertainty Estimation in Deep Learning Segmentation of Fluorescence Microscopy Images with Missing Markers
This work addresses a domain-specific challenge in biomedical imaging by providing quality assurance for segmentation models when encountering unseen marker combinations, which is incremental but practical for researchers in microscopy.
The paper tackles the problem of estimating segmentation quality for fluorescence microscopy images with missing markers, where deep learning models struggle due to varying channel combinations. It proposes a method using uncertainty estimation and a Random Forest model to predict segmentation metrics, showing an improvement in segmentation performance.
Fluorescence microscopy images contain several channels, each indicating a marker staining the sample. Since many different marker combinations are utilized in practice, it has been challenging to apply deep learning based segmentation models, which expect a predefined channel combination for all training samples as well as at inference for future application. Recent work circumvents this problem using a modality attention approach to be effective across any possible marker combination. However, for combinations that do not exist in a labeled training dataset, one cannot have any estimation of potential segmentation quality if that combination is encountered during inference. Without this, not only one lacks quality assurance but one also does not know where to put any additional imaging and labeling effort. We herein propose a method to estimate segmentation quality on unlabeled images by (i) estimating both aleatoric and epistemic uncertainties of convolutional neural networks for image segmentation, and (ii) training a Random Forest model for the interpretation of uncertainty features via regression to their corresponding segmentation metrics. Additionally, we demonstrate that including these uncertainty measures during training can provide an improvement on segmentation performance.