Research On CT Image Processing Method Based On Improved Top-hat Algorithm

At present,pulmonary nodules are the most common type of lung disease,and lung cancer has the highest morbidity and mortality,which is a serious threat to human health.At present,the main means of clinical lung nodule screening is chest CT plain and enhanced detection,and the doctor diagnosed the lung CT image with the following problems: The correctness of doctor diagnosis depends on many subjective factors;Doctors have a large workload,and doctors of different ages have different diagnostic results;Clinical experts need to find information related to lung disease from a huge number of CT examinations,which is itself a highly complex and technically high work.From the perspective of assisting doctors and experts,this paper uses image processing methods to provide automatic lesion detection analysis,which can complete CT examination and analysis in a very short time,and reduced doctor's misdiagnosis rate and reading time.The paper studies the detection of lung nodules and the calculation of eigenvalues.First,the characteristics of CT images of the lungs were analyzed.Secondly,the edge detection algorithm of lung CT image,the regional growth algorithm commonly used in lung CT image,and the deep learning algorithm are studied,and summarized its algorithm principle and its advantages and disadvantages.Finally,the combination of Top-Hat algorithm and wavelet algorithm is adopted,and the Top-Hat algorithm is improved in multi-scale,multi-structure,and multi-direction.Aiming at the CT medical images of the lungs,the methods of image acquisition,image preprocessing,lung parenchymal segmentation,image contrast enhancement,suspected edge detection,and suspected pulmonary nodule diagnosis were studied.The paper processed a large number of pulmonary CT medical images,combined with the morphological features of pulmonary nodules and blood vessels in the lung parenchyma,using circular and linear structural elements to extract the suspected edge information.The wavelet decomposition and reconstruction algorithm is used to decompose the lung parenchyma into low frequency components and high frequency components.Combined with improved Top-Hat algorithm improves the contrast of low-frequency components and reduces the noise of high-frequency components.The improved the contrast of the lung parenchyma area after remodeling increased by 135%,Top-Hat algorithm is used to detect the edge information of the suspected region.Experiments show that the edge closure degree of the image extracted by the algorithm can reach 84.7%,and the key characteristic values such as perimeter,area and diameter of the suspected points are extracted,which provides a reference for doctors' diagnosis and improves the diagnostic efficiency and accuracy of the doctor.