Lobar Fissure Extraction in Isotropic CT Lung Images - An Application to Cancer Identification

The essential organ for respiration and inspiration of human beings are Lungs. It consists of five distinct lobes which are separated by three fissures (the boundaries of lung lobes are the areas containing fissures and having absence of bronchial trees). They are two oblique (left and right) fissures and one horizontal fissure. The left lung consist of left oblique fissure which separates the superior and middle lobes. The right lung consist of right oblique fissure which separates superior and middle lobes and right horizontal fissure which separates middle and inferior lobes. The identification of the lobar fissures in isotropic Computed Tomography (CT) images are very difficult even for the experienced surgeons because of its variable shape and appearance along with low contrast and high noise association with it. Further the fissure thickness is observed to be around 2 pixels (approximately 1.2mm) complicates the fissure identification. The identification of lobar fissure in CT images will be helpful for the surgeon to identify the cancer location before they plan for surgery. The surgical removal of the diseased lung is the final stage for treating the lung cancer. Therefore it is necessary to find the cancer location at the early stage to treat it. This paper presents an automated method to extract the left and right oblique fissures from the CT lung images. The proposed method is implemented in two phases. In the first phase, the fissure region is located. In the second phase, the found lobar fissures are extracted. The obtained results show that the proposed work can help the surgeon to identify the cancer location. Keywords-Computed Tomography (CT), Dual Tree Complex Wavelet Transform (DTCWT), Filter Bank and Discrete Wavelet Transform (DWT). 1. INTRODUTION The lungs are the very important organ for the human beings for respiration. The cells in human body are normally divided and grow in a control manner. When the tissues start expanding and the control process is lost then the situation is called cancer. Lung cancer is the deadliest cancer in the world for both men and women. Lung cancer has surpassed breast cancer as the leading cause of cancer deaths in women [1]. The cells form a mass or tumour that differs from the surrounding tissues from which it arises. The tumours take oxygen, nutrients, and space from healthy cells and because destroy the ability of normal tissues to function hence it is very dangerous. The diagnosis process of tumour is based on whether a pulmonary nodule is benign (normal) or malignant (cancerous). The most common way to differentiate the benign and malignant nodule is by examining the growth rate of nodule. The cancerous nodules can double in size on average every four months (some as slowly as 15 months, some as quickly as 25 days). Benign nodules, on the other hand, do not grow much if at all. Growth can be evaluated through the serious of CT or X-ray scans over the period of time. Another most common way to differentiate the malignant nodule from the benign is development based on its shape and surface. Cancerous nodules more likely to have irregular shapes and rougher surfaces and color variations. Benign nodules tend to be smoother and more regularly shaped, with more even color throughout. In most of the cases, CT scan or X-rays provide adequate information to make a reliable diagnosis. The doctors might choose to retrieve cells from the suspected nodules for a biopsy. The anatomy of human lungs are shown in Fig. 1. It consists of five distinct lobes that are separated by three fissures namely, left oblique fissure, right oblique fissure and right horizontal fissure respectively. The CT slice which shows the Right Oblique Fissure (ROF), Right Horizontal Fissure (RHF) and Left Oblique Fissure (LOF) are shown in Fig. 2. The surgical removal of the diseased lung is the final stage of treating the lung cancer. It is very important to find it at the early stage to limit the danger. The lung cancer identification is done by extracting the lobar fissure from the CT images. However the extraction of fissure region in isotropic CT images is very challenging even for the experienced surgeons. Fig. 1 Anatomy of human lungs In clinical settings the surgeons examine the stack of 2D clinical CT images for surgical planning to identify the diseased lung lobes but it takes long time to start the surgical procedure. To reduce the surgical planning time we proposed the automatic lobar fissure (boundaries of lobes) extraction by Dual Tree Complex Wavelet Transform (DTCWT). The proposed method first identifies the fissure region then extracts the identified lobar fissures.