Abstract:

Objective To develop a portable handhold 3D Ultrasound imaging and diagnostic system, and to validate its reliability and accuracy in the clinical diagnosis of carotid atherosclerotic plaque.
Materials and methods First acquisition enrolled 55 carotid arteries images by a portable probe which was an important part of the 3D Ultrasound imaging and diagnostic system. 25 arteries were the training set and 30 for validation set. Second acquisition enrolled 84 arteries as an independent validation set. The diagnostic result of our System in carotid plaque including AUC, specificity, sensitivity, accuracy was calculated.
Results A set of learning-based method for automatic segmentation, reconstruction and diagnosis was established. In the validation set and independent validation set, the AUC, sensitivity, specificity and accuracy achieved 0.708 (95%CI: 0.509-0.908)/0.860 (95%CI: 0.773-0.953), 83.3%/83.9%, 58.3%/88.7%, 73.3%/86.9%, respectively.
Conclusion The proposed portable handhold 3D Ultrasound imaging and diagnostic system achieved a satisfactory sensitivity, specificity and accuracy in the diagnosis of carotid plaque. Although dependence on operators was an inevitable problem in routine ultrasound examination, the diagnosis results of our system were not highly dependent on examiner experience. In the future, the system could be applied to carotid atherosclerosis screening in communities or remote settings to help address the shortage of ultrasound examiners.

Key words: Portable; 3D Ultrasound imaging and diagnostic system; Carotid atherosclerotic plaque; Automatic