DWI and texture analysis of head and neck tumours

Malignant Head and Neck tumours count for approximately 3% of all malignancies. Although they are not very common, it is important to evaluate their exact extent. This applies especially to the neck where numerous structures are located in a relatively confined area; the great vessels, muscles, spine and nerves as well as pharynx, larynx and oesophagus. To surgically remove the tumour it must not have spread into adjacent structures, because, if there is trans-compartmental extent, of tumour the following therapy most often changes. Therefore lymph node dissection or concordant radio- or chemotherapy may be required. Magnetic resonance imaging (MRI) is the best method to detect these tumours and to define their extent because of the good soft tissue contrast. Diffusion weighted imaging (DWI) is a new MR method to examine tissue on the microscopic level by the movement of water protons. Experiments have shown that pattern recognition of tissue is possible with certain computer evaluated texture parameters from MR images which the human eye cannot distinguish. The aim of this cross- sectional study is to determine the feasibility of texture analysis for the discrimination between benign and malignant head and neck tumours and to further investigate the application of diffusion weighted MR imaging in head and neck tumours. This project is designed for a period of one year. In the first phase the MR protocol will be defined. In the second phase 50 patients will be examine with the defined protocol on a 3 Tesla MR scanner with a dedicated head and neck coil. After the first examinations data analysis, representing the third phase, will be started. The MR (DICOM) images will be transferred to the texture program MaZda. For each lesion and MRI sequence, three manually drawn region-of-interests (ROIs) will be defined. The first ROI, covering the entire lesion, will be manually defined on the image that depicts the lesion in its greatest diameter. The other two ROI will be defined on adjacent slices, depending on the size of the lesion. For the analysis of the diffusion-weighted images, ROIs will be placed on the diffusion-weighted images in a similar manner. The first ROI, covering the entire lesion, will be again defined at the lesion`s greatest diameter and the apparent diffusion coefficient will be calculated. If the discrimination between benign and malignant tumours of the head and neck by texture analysis or DWI is feasible, these methods could be transferred to daily clinical practice. Furthermore we could determine the exact tumour extent and infiltration of adjacent structures by texture analysis.