Prediction Models for Malignant Pulmonary Nodules Based-on Texture Features of CT Image

Lung cancer is one of the most harmful forms of cancer, which is the leading cause of cancer death in many regions of the world (Ahmedin Jl et al.,2005). The overall 5-year survival rate of lung cancer patients is only 14%, and remained at this level for the past two decades. However, when lung cancer is found at the early stage I or II, 5-year survival rates can be as high as 60-70% ( Beadsmoore CJ et al.,2003). Early diagnosis of lung cancer was only 15%( Li YR et al., 2007). Although histology diagnosis is the most accurate detection method in the medical environment, it is an aggressive invasive procedure that involves risks, discomfort and trauma, which restrict it to be used in the clinical practice. Digital CT (Computed Tomography), overcoming the shortages of histology diagnosis, has gradually become the best imaging diagnosis method of lung cancer. CT enables us to visualize lung anatomy in great detail and has been used to accurately diagnose lung diseases since the 1980s (Ye X et al.,2006).Detecting and diagnosing solitary pulmonary nodules (SPNs, referring to the lesion of lung field ≤ 3 cm in diameter), the most common manifestation of lung cancer, are critical since early identification of malignant nodules is crucial to the chance for successful treatment. But pulmonary nodules of lung cancer in CT images share similarity with benign cases to some extent, such as tuberculosis, inflammatory pseudotumor, hamartoma, and aspergillosis(Jee WC et al.,2008), which makes it difficult to distinguish, especially for the doctors who are not rich in clinical experience. With technique of computer rising, the computer-aided diagnosis (CAD) has become an auxiliary diagnosis tool (Jiang J et al.,2007), especially in diseases that can not be diagnosed efficiently. To improve the accuracy and efficiency of CT screening programs for the detection of early-stage lung cancer, a number of