A Semi-Supervised Retinal Vessel Segmentation Method via Adaptive Uncertainty Estimation.

Area of Science:

• Medical Imaging
• Computer Vision
• Machine Learning

Background:

• Supervised learning for medical image segmentation demands extensive expert annotation.
• Existing methods struggle to efficiently leverage large unlabeled datasets.
• Accurate vessel segmentation is crucial for diagnosing various medical conditions.

Purpose of the Study:

• To develop a semi-supervised learning technique for vessel segmentation.
• To reduce the reliance on labor-intensive, expert-level data labeling.
• To enhance segmentation accuracy by effectively utilizing unlabeled data.

Main Methods:

• Introduced an adaptive uncertainty estimation (AUE) method for semi-supervised vessel segmentation.
• Employed a teacher-student network architecture to preserve high-confidence pixels.
• Utilized adaptive thresholding for pixel-level uncertainty estimation.

Main Results:

• The AUE method demonstrated superior accuracy compared to supervised and other semi-supervised approaches.
• Achieved state-of-the-art performance on the STARE public retinal dataset.
• Effectively acquired new features from unlabeled data, improving predictive accuracy.

Conclusions:

• Semi-supervised learning with AUE offers a promising solution for efficient medical image segmentation.
• The proposed method significantly reduces the need for manual data annotation.
• This technique advances the field of automated medical image analysis for vessel segmentation.