Research on the Application of 3.0T Magnetic Resonance 32-Channel Based on 3D Convolutional Neural Network in the Diagnosis and Treatment of Acute Cerebral Infarction
Objective To explore the application value of 3.OT Magnetic Resonance 32-Channel based on 3D Convolutional Neural Network(3D-CNN)in the diagnosis and treatment of Acute Ischemic Stroke(AIS).Methods Patients diagnosed with AIS according to the early treatment guidelines between November 2022 and June 2023 were selected as research subjects,with a total of 280 study samples included.Each patient underwent a 3.OT MRI scan in the acute phase of the stroke,with sequences including T1,T1c,T2,CBF,CBV,DWI,Tmax,and TTP.The image resolution was 2x2x2 mm3.MRI data from 180 patients were used as the training set,and 100 patients who had undergone MRI examinations constituted the test set.The 3D-CNN model was used to train the training set and predict the test set,outputting the segmentation results of AIS lesions.AIS lesions consist of the ischemic core and the surrounding penumbra,with the brain tissue in the ischemic core being completely necrotic and irrecoverable.This algorithm can identify both the core and the penumbra,not just the penumbra.The performance of this algorithm was evaluated based on the DICE coefficient,accuracy,sensitivity,Average Symmetric Surface Distance(ASSD),and Hoffman distance.Results The algorithm demonstrated high segmentation performance on the test set,with a DICE coefficient of 0.87±0.05,accuracy of 0.91±0.04,sensitivity of 0.85±0.06,ASSD of 1.23±0.32 mm,and Hausdorff distance of 1.56±0.41 mm.Compared with manual marking results by imaging experts,there was no significant statistical difference(P＞0.05).Conclusion The 3.0T Magnetic Resonance 32-Channel based on 3D-CNN has high application value in the diagnosis and treatment of AIS,effectively segmenting AIS lesions,and providing more accurate and faster diagnostic basis for clinical doctors.
Acute Cerebral InfarctionPenumbra3D Convolutional Neural Network3.0T Magnetic Resonance32-Channel
万方数据
维普
