SU‐C‐BRA‐02: A Lung Tumor Autocontouring Algorithm Based On Particle Filter for Dynamic Magnetic Resonance

Abstract

Purpose:This study proposes a novel autocontouring algorithm for lung tumors based on particle filter. It is developed in the context of MR‐linac treatments and is validated on in‐vivo dynamic magnetic resonance images.Methods:A sequential Monte Carlo method called particle filter is combined with Otsu's thresholding technique to contour lung tumors on dynamic MR images. Four non‐small cell lung cancer (NSCLC) patients were imaged with a 1.5 T MR (MAGNETOM Aera, Siemens Healthcare, Erlangen, Germany) for one minute at a rate of 4 images per second. A pretreatment image‐processing step includes a manual contour of the tumor, the tumor's displacement, and its descriptive statistics. During treatment, the contours are automatically generated by thresholding the image in a window surrounding the center of mass of the particles. The Dice similarity coefficient (DSC), the precision, the recall, the Hausdorff distance (HD) and the difference in centroid positions (Δd) are calculated to compare with the expert's contours.Results:This autocontouring algorithm presents continuous adaptability and is independent of pre‐treatment training. The number of particles is proportional to the area of the tumor and increases the computational time at a rate of 2 ms for every 500 particles, whereas the contouring step adds a constant 14 ms. The contours’ comparison is obtained with a mean DSC of 0.89 to 0.91, mean precision of 0.88 to 0.92, mean recall of 0.88 to 0.95 and mean Δd of 0.6 to 2.3 mm.Conclusion:This work presents a new autocontouring algorithm for NSCLC patients on dynamic MR images. Combining the autocontours with a motion prediction algorithm presents a promising method for lung tumor online tracking treatments.