Automatic lung segmentation in CT scans using guided filtering
Gábor Révy, Dániel Hadházi, Gábor Hullám · 2022
The segmentation of the lungs in chest CT scans is a crucial step in computer-aided diagnosis. Current algorithms designed to solve this problem usually utilize a model of some form. To build a sufficiently robust model, a very large amount of diverse data is required, which is not always available. In this work, we propose a novel model-free algorithm for lung segmentation. Our segmentation pipeline consists of expert algorithms, some of which are improved versions of previously known methods, and a novel application of the guided filter method. Our system achieves an IoU (intersection over union) value of 0.9236 ± 0.0290 (mean±std) and a DSC (Dice similarity coefficient) of 0.9601 ± 0.0158 on the LCTSC dataset. These results indicate, that our segmentation pipeline can be a viable solution in certain applications.