Pulmonary Tumor Volume Detection from Positron Emission Tomography Images

Lung carcinoma is one of the most lethal of cancers worldwide. Positron emission tomography (PET) data has greater sensitivity and specificity in the staging of lung cancer than computer tomography (CT) or magnetic resonance imaging (MRI). Inaccurate detection of the tumor volume by the physicians is the largest source of error in the efficient planning of radiation treatment. In this paper we present an automated process of tumor delineation and volume detection from each frame of PET lung images. We have represented the data using spatial features (geometric moments) and frequency domain features (discrete cosine transform, wavelets). The performance of these features were analysed using k-nearest neighbor and support vector machines (SVM) classifiers. Wavelet features with SVM classifier gave a consistent accuracy of 97% with an average sensitivity and specificity of 0.81 and 0.99 respectively. The calculated volume from the delineated tumor by the proposed method matched the manually segmented volume by the physicians. This research will facilitate the physicians in accurate staging and radiotherapy treatment planning for lung tumors. It also eliminates the need for manual tumor segmentation thus reducing the physician fatigue to a great extent.