Prediction Of Ki-67 Index Of Pituitary Adenoma Based On Deep Learning

Objective:The artificial intelligence and clinical data of patients with pituitary adenomas were used to build a neural network to predict the Ki-67 index of pituitary adenomas before operation.And explore the feasibility and effectiveness of predicting the Ki-67 index of pituitary adenomas by MRI data,pituitary hormone degree and clinical data of patients before operation.Methods:The data of patients who meet the inclusion criteria were collected,including preoperative MRI imaging data,preoperative clinical data of pituitary hormones and basic information of patients.We use Google open-source artificial intelligence framework ‘Tensorflow’ to build a deep learning model.All the data were randomly divided into the training set and the test set according to the proportion of 75% and 25%.All the data is processed.The imaging data of the patients were formatted,and the images were identical to 512*512.Other clinical data were normalized,and the basic information of patients was also normalized.The training set was used to build the artificial intelligence model.After fully training,the test set was used to test the performance of the model.Results:After 200 iterative trainings,the loss function of the model gradually tends to be stable and no longer decreases,which means that the model is fully trained.After fully training,the test set was used to test the performance of the model,with an mean absolute deviation(MAD): 19.20%;the coefficient of determination(R~2): 0.89.The mean absolute deviation is the average of the absolute deviation of all individual predicted values from the arithmetic mean.The maximum value of R~2 is 1,so the closer to 1,the better of the model is.Conclusion:The deep learning model built in this study has good predictive performance.On the test set,the prediction of Ki-67 index by the model is highly consistent with the Ki-67 index obtained by pathology.Combined with some evaluation indicators,it can be considered that in this study the artificial intelligence is feasible to predict the Ki-67 index of pituitary adenomas before operation,it plays a role in preoperative diagnosis to a certain extent.