Design And Implementation Of Colonic Lesion Detection System

In recent years,along with the emergence of new technologies in machine learning and deep learning,computer science is no longer a separate discipline.Computer science is combined with more fields and industries to develop and progress together.Among them,the medical field and computer science combined with the development of one of the earliest disciplines,and currently have achieved a certain amount of scientific research results.In the medical field,doctors use imaging equipment to make accurate diagnoses of patients.Traditional medical imaging is diagnosed by the doctor himself.Physicians' interpretation of images relies on their own knowledge and long-term work experience,which makes each physician have their own specific judgment on the judgment of the disease,that is,there are subjective problems.And there is a certain difference in the physical structure between people,and this subjectivity may trigger the risk of misinterpretation.Therefore,a medical image judgment system combining techniques such as machine learning and deep learning has appeared.The colon is commonly called the large intestine is an organ located in the human body.Common colon diseases include colon polyps,early colon cancer,and advanced colon cancer.Colonic lesions are one of the most frequent human diseases.Assisting doctors in diagnosis through colonic lesions can help doctors to perform colonoscopy-assisted examinations,reduce the doctor's work intensity,improve the accuracy of colonic lesion diagnosis,and thus reduce recurrence rate and cure rate..Therefore,this paper developed a colonic lesion detection system,and the development of this system is also an important part of hospital information construction.The development of this system combines deep learning techniques and is used for real-time diagnosis and scientific research by doctors.Real-time diagnosis requires the system to have stability and real-time performance,while the scientific research on real-time requirements is not strict,but the data volume requirements are relatively high,which is different from the real-time diagnostic system requirements,so separate design is adopted.Therefore,the research and development system of this paper consists of two parts:the lesion detection system and the scientific research system.Among them,the lesion detection system is mainly composed of case information,three modules of real-time colon detection and disease diagnosis.Among them,the function of the case information filling module is to record the basic information of the patient.The colon real-time detection system module is a colonic lesion AI(artificial intelligence)aided discovery,which is used by doctors to automatically detect lesions during colonoscopy.The specific implementation process of lesion detection is to train the model first,and complete the function of lesion detection by calling the model.Among them,the multi-target recognition algorithm SSD(Single Shot MultiBox Detector)of deep convolutional neural network is used to train the model.It mainly includes training set collection,label making,model training,verification and lesion recognition testing.The SSD algorithm uses multi-scale feature maps for detection,and convolves the different feature maps to obtain the final test results.The condition diagnosis module is for a doctor to fill out a diagnostic report.The scientific research system mainly consists of four modules:case information management,case information collection,case information follow-up,and case information statistics.The case information management stores the doctor's case information for the doctor to do research data.Case information statistics can be used for scientific research by comparing the proportion of different populations in the case of education,basic information,and clinical symptoms.The development of the system has improved the efficiency of the colonoscopy for doctors,reduced the workload of doctors,and provided convenience for doctors'scientific research.In addition,from the current medical digitalization,doctors can quickly operate the system.The development of this system provides auxiliary assistance for the examination of doctors,and also has practical significance.At present,the system has been running online,and this paper has certain commercial value.