DeepIGeoS: A Deep Interactive Geodesic Framework for Medical Image Segmentation

TLDR

Accurate medical image segmentation is essential for diagnosis and surgical planning, yet fully automatic CNNs often require refinement to achieve clinical robustness. The authors aim to create a deep interactive segmentation framework that merges CNNs with user inputs via geodesic distance transforms and a resolution‑preserving network to reduce refinement effort and boost accuracy. Their method first generates an automatic segmentation with a CNN, then refines it using a second CNN that incorporates user‑indicated mis‑segmentations and geodesic transforms, while enforcing hard constraints through a back‑propagatable Conditional Random Field, and is validated on 2D placenta and 3D brain tumor MRI datasets. Experiments show the approach markedly improves upon baseline CNNs and matches or exceeds traditional interactive methods, all while requiring fewer user interventions and less time.

Abstract

Accurate medical image segmentation is essential for diagnosis, surgical planning and many other applications. Convolutional Neural Networks (CNNs) have become the state-of-the-art automatic segmentation methods. However, fully automatic results may still need to be refined to become accurate and robust enough for clinical use. We propose a deep learning-based interactive segmentation method to improve the results obtained by an automatic CNN and to reduce user interactions during refinement for higher accuracy. We use one CNN to obtain an initial automatic segmentation, on which user interactions are added to indicate mis-segmentations. Another CNN takes as input the user interactions with the initial segmentation and gives a refined result. We propose to combine user interactions with CNNs through geodesic distance transforms, and propose a resolution-preserving network that gives a better dense prediction. In addition, we integrate user interactions as hard constraints into a back-propagatable Conditional Random Field. We validated the proposed framework in the context of 2D placenta segmentation from fetal MRI and 3D brain tumor segmentation from FLAIR images. Experimental results show our method achieves a large improvement from automatic CNNs, and obtains comparable and even higher accuracy with fewer user interventions and less time compared with traditional interactive methods.