Fully Automatic Segmentation of Lumbar Vertebrae from CT Images using Cascaded 3D Fully Convolutional Networks
This addresses the challenge of accurate medical image segmentation for clinical applications, but it is incremental as it builds on existing deep learning methods.
The paper tackled the problem of segmenting lumbar vertebrae from CT images with varying fields of view using cascaded 3D fully convolutional networks, achieving an average Dice coefficient of 95.77% and an average symmetric surface distance of 0.37 mm.
We present a method to address the challenging problem of segmentation of lumbar vertebrae from CT images acquired with varying fields of view. Our method is based on cascaded 3D Fully Convolutional Networks (FCNs) consisting of a localization FCN and a segmentation FCN. More specifically, in the first step we train a regression 3D FCN (we call it "LocalizationNet") to find the bounding box of the lumbar region. After that, a 3D U-net like FCN (we call it "SegmentationNet") is then developed, which after training, can perform a pixel-wise multi-class segmentation to map a cropped lumber region volumetric data to its volume-wise labels. Evaluated on publicly available datasets, our method achieved an average Dice coefficient of 95.77 $\pm$ 0.81% and an average symmetric surface distance of 0.37 $\pm$ 0.06 mm.