Tubular Shape Aware Data Generation for Semantic Segmentation in Medical Imaging
This addresses the resource-intensive annotation process for medical imaging segmentation, offering a practical solution for interventional radiology, though it is incremental as it builds on existing GAN and shape-prior techniques.
The paper tackled the problem of limited annotated data for segmenting tube-like objects in chest X-ray images by developing a synthetic data generation method using a GAN with shape constraints, achieving accuracy comparable to fully-supervised models with only 10–20 weakly-labeled images.
Chest X-ray is one of the most widespread examinations of the human body. In interventional radiology, its use is frequently associated with the need to visualize various tube-like objects, such as puncture needles, guiding sheaths, wires, and catheters. Detection and precise localization of these tube-like objects in the X-ray images is, therefore, of utmost value, catalyzing the development of accurate target-specific segmentation algorithms. Similar to the other medical imaging tasks, the manual pixel-wise annotation of the tubes is a resource-consuming process. In this work, we aim to alleviate the lack of the annotated images by using artificial data. Specifically, we present an approach for synthetic data generation of the tube-shaped objects, with a generative adversarial network being regularized with a prior-shape constraint. Our method eliminates the need for paired image--mask data and requires only a weakly-labeled dataset (10--20 images) to reach the accuracy of the fully-supervised models. We report the applicability of the approach for the task of segmenting tubes and catheters in the X-ray images, whereas the results should also hold for the other imaging modalities.