Objective:To explore the construction and efficacy of cranial magnetic resonance imaging(MRI)classification model based on attention mechanism in type 2 diabetes patients with cognitive impairment.Methods:The case data of 100 patients with type 2 diabetes who were treated in the General Hospital of North China Petroleum Administration Bureau from June 2022 to January 2024 were retrospectively selected.A total of 100 MRI images of cranial FLAIR_LongTR sequence with cognitive impairment(32 cases)and those without cognitive impairment(68 cases)were respectively collected.The images of the above two kinds of samples were horizontally and vertically translated to expand to 1000 samples,respectively.The samples were randomly divided into training samples(n=700)and test samples(n=300)as the ratio of 7:3 according to affine transformation data augmentation method.Then,the attention mechanism model was established to test the images with full scan of the test samples.The ability of the attention mechanism system in screening cognitive impairment was analyzed according to the method of setting threshold value.The 100 MRI images of cranial FLAIR_LongTR sequence of patients in our hospital from January 2024.From January to May 2024 were used as a verification set to verify the diagnostic value of attention mechanism.Results:With the increasing of iteration times,the sample loss of training and verification of attention mechanism model gradually decreased and tended toward stability,and the accuracy of training set and verification set gradually increased and tended toward stability.In the attention mechanism model,the average loss rate of training samples was 10.024%,and that of test samples was 15.247%.In the attention mechanism model,the average accuracy of training samples was 99.078%,and the average accuracy of test samples was 99.753%.Receiver operating characteristic(ROC)curves showed that the area under curve(AUC)of attention mechanism model was 0.998,which can better diagnose cognitive impairment of patients with type 2 diabetes than the resNET model(AUC=0.656)(Z=3.437,P<0.001).Conclusion:The constructed cranial MRI classification model by using attention mechanism has favorable diagnostic value for cognitive impairment in patients with type 2 diabetes.

Objective:To explore the construction and efficacy of cranial magnetic resonance imaging(MRI)classification model based on attention mechanism in type 2 diabetes patients with cognitive impairment.Methods:The case data of 100 patients with type 2 diabetes who were treated in the General Hospital of North China Petroleum Administration Bureau from June 2022 to January 2024 were retrospectively selected.A total of 100 MRI images of cranial FLAIR_LongTR sequence with cognitive impairment(32 cases)and those without cognitive impairment(68 cases)were respectively collected.The images of the above two kinds of samples were horizontally and vertically translated to expand to 1000 samples,respectively.The samples were randomly divided into training samples(n=700)and test samples(n=300)as the ratio of 7:3 according to affine transformation data augmentation method.Then,the attention mechanism model was established to test the images with full scan of the test samples.The ability of the attention mechanism system in screening cognitive impairment was analyzed according to the method of setting threshold value.The 100 MRI images of cranial FLAIR_LongTR sequence of patients in our hospital from January 2024.From January to May 2024 were used as a verification set to verify the diagnostic value of attention mechanism.Results:With the increasing of iteration times,the sample loss of training and verification of attention mechanism model gradually decreased and tended toward stability,and the accuracy of training set and verification set gradually increased and tended toward stability.In the attention mechanism model,the average loss rate of training samples was 10.024%,and that of test samples was 15.247%.In the attention mechanism model,the average accuracy of training samples was 99.078%,and the average accuracy of test samples was 99.753%.Receiver operating characteristic(ROC)curves showed that the area under curve(AUC)of attention mechanism model was 0.998,which can better diagnose cognitive impairment of patients with type 2 diabetes than the resNET model(AUC=0.656)(Z=3.437,P<0.001).Conclusion:The constructed cranial MRI classification model by using attention mechanism has favorable diagnostic value for cognitive impairment in patients with type 2 diabetes.