A Study Of Computer Aided Measurement Of Colorectal Polyp Size

【Introduction】With the rapid improvements of people’s living standards,the incidence of colorectal cancer is increasing.At present,the adenoma-cancer evolution theory is widely recognized and majority of colorectal cancer is evolved from colorectal adenomas,which have a high correlation with the carcinogenesis of adenomas.The establishment of accurate and objective measurement tools for colorectal polyps is of great significance for the selection of treatment and postoperative follow-up monitoring.Computer-aided diagnosis(CAD)is now widely used in the medical field.Early studies have already shown the potential of AI-based CAD systems in the diagnosis of several luminal gastrointestinal diseases,including Barrett’s esophagus,esophageal squamous cell cancer,gastric cancer,and Helicobacter pylori gastritis,and colorectal polyps.In this study,CAD is proposed to establish an accurate and objective measurement tool for colorectal polyps as well as to help determine the imaging size of colorectal polyps.【Materials and Methods】The method in this investigation is to accurately predict the size of colorectal polyps on the basis of ascertaining colorectal polyps.As the distance between the colonoscopy and the lesion increases,the magnification has a decreasing trend.According to this trend,the colorectal polyps’ real size can be calculated according to the captured image and the shooting distance.First,real lumen was used in vitro to simulate the intestinal tract,one small ball with a diameter of 15.8mm were used to simulate colorectal polyps,and then the small ball was photographed from different distances by electronic colonoscopy(EC-590 WM was used)to record the shooting distance and the diameter of the small balls in the image that taken.The curve fitting was created by the correlation between the diameter(cm)of the small ball and the shooting distance(cm)of the small ball in the image.We fit the resulting data with different functions,and then choose the most suitable function as the fitting curve.According to the fitting curve and the true diameter of the small ball,the magnification factor of the electronic colonoscopy is obtained.The study used the CAD system to train the overall shape of colorectal polyps,and image segmentation technique was employed to accurately identify the range of colorectal polyps.Finally,based on the magnification factor,the real value of polyps was predicted according to the shooting distance and the segmentation image size.【Results】In this study,the distance was artificially controlled in the endoscopic operation,and the image was taken at a preset distance.The study fit the obtained data with common functions such as polynomial function,exponential function,logarithmic function,etc.Results show that polynomial fits better the correlation between shooting distance and ball size,and the higher the degree of polynomial,the better the fitting effect.However,in order to avoid over-fitting,we choose cubic polynomial as the fitting function.After determining the fitting curve,the CAD system can calculate the true size of the colorectal polyps according to the magnification of the corresponding distance.【Conclusions】There is still a lack of practical technical methods to objectively evaluate the size of colorectal polyps in the clinical practice of colonoscopy.The CAD system in this study was able to identify colorectal polyps and predict their size,providing data for surgical resection and postoperative follow-up monitoring.There are also some defects in the study,including the need for manual distance measurement,relatively strict image inclusion criteria and Static Image.Based upon this approach,more devices would be involved and expanded for practical testing and applications,which is expected to obtain clinically practical values in other areas.But its clinical practicality needs further prospective study.