Volume Estimation of Pulmonary Lesion Using Chest CT Sequence

Abstract

Many of the imaging modalities like X-ray, Computed Tomography (CT), Magnetic Resonance Imaging (MRI), fMRI have emerged to capture high quality images of anatomical structures of the human body. Radiologist can also have a better visualization if the regions of interest in the images are extracted and visualized 3D. To extract region of interest, sometimes preprocessing steps are performed on the input data. Pulmonary lesion is a small round or oval-shaped growth in the lung. It consists of solid and non-solid portion. The estimation of solid and non-solid portion of the pulmonary nodules will help the clinicians in the diagnosis and to suggest the appropriate treatment methodology. Lesion volume estimation gives a brief idea about the area occupied by the lesion tissues, which in turn can help the radiologist to plan treatment accordingly. In proposed work, lesion is segmented using K-means algorithm and then volume of the lesion is estimated. The slices which have segmented lesion with solid and non-solid regions is used for 3D visualization. The results obtained using the proposed methodology is validated with the Slicer 3D software. Error in the estimated volume of the solid and non-solid portion of the lesion was found to be in the range of 1.11% - 3.30% and 0.1% to 4.55% respectively. Results from the proposed methodology, lesion extraction with solid and non-solid, 3D visualization of the same and volume estimation respectively are validated by taking feedback from the radiologists and segmented lesion slices are used to estimate the volume and 3D visualization in Slicer 3D software for validation.