Automatic Near Real-Time Evaluation of 3D Ultrasound Scan Adequacy for Developmental Dysplasia of the Hip

Accurate detection and diagnosis of developmental dysplasia of the hip (DDH), a common hip instability condition among infants, relies heavily on acquiring adequate ultrasound (US) image data. Although 2D US is the standard modality used for DDH screening, 3D US has recently been considered as well. Presently there is no automatic method (or even a standardized manual method) capable of analyzing the US volume to determine whether that volume is adequate for extracting DDH metrics required for diagnosis. Scan adequacy in 2D has seen only one work on automation and there has been no work done on scan adequacy in 3D. We propose an automatic, near real-time method of assessing 3D ultrasound scans in developmental dysplasia screening and diagnostic applications using a convolutional neural network (CNN). Our classifier labels volumes as adequate or inadequate for subsequent interpretation based on the presence of hip anatomy needed for DDH diagnosis. We validate our approach on 40 datasets from 15 pediatric patients and demonstrate a classification rate of 100% with average processing time of just above 2 s per US volume. We expect automatic US scan adequacy assessment to have significant clinical impact with the potential to help in imaging standardization, improving efficiency of measuring DDH metrics, and improving accuracy of clinical decision making.