Presenter of 1 Presentation
CEREBRAL MICROBLEED（CMB）AUTOMATIC DETECTION SYSTEM BASED ON THE “DEEP LEARNING”
Abstract
Background and Aims
To validate the reliability and efficiency of clinical diagnoses in practice based on a well-established system for automatic segmentation of Cerebral Microbleed (CMB).
Methods
This is a retrospective study based on MRI-SWI data sets from 1615 patients. All patients had been diagnosed with cerebral small vessel disease (CSVD) with clear cerebral microbleeds (CMBs) on MRI-SWI. The patients were divided into training, and validation cohorts of 1,285 and 330 patients respectively. The model training data were labeled layer by layer according to the consensus from two neuroradiologists with 15 years of work experience. After that, a three-dimensional convolutional neural network (CNN) was applied to the MRI data from the training and validation cohorts to construct a deep learning system (DLS) tested with the 72 patients, independent above the MRI cohort. The DLS tool was used as a segmentation program for these 72 patients. These results were evaluated and revised by 5 neuroradiologists and given an output analysis divided into Miss labeling, and Wrong labeling. The inter-rater agreement kappas test was used for quality analysis.
Results
The DLS achieved a Dice coefficient of 0.72. In the independent data's clinical evaluations, the neuroradiologists reported that more than 95% of the lesion could be directly detected, less than 5% of lesions were wrong labeled or miss labeled by our DLS. The inter-neuroradiologist-DLS agreement kappa value reaches 0.79 on average.
Conclusions
Automatic detection and segmentation of CMBs are feasible. The proposed well-trained DLS system could be a trusted tool for the segmentation and detection of CMB lesions.