Locating Cephalometric Landmarks with Multi-Phase Deep Learning

Abstract

Cephalometric analysis has long been one of the most helpful approaches in evaluating cranio-maxillo-facial skeletal profile. To perform this, locating anatomical landmarks on an X-ray image is a crucial step, demanding time and expertise. An automated cephalogram analyzer, if developed, will be a great help for practitioners. Artificial intelligence, including machine learning is emerging these days. Deep learning is one of the most developing techniques in data science field. The authors attempted to enhance the accuracy of an automated landmark predicting system utilizing multi-phase deep learning and voting. To guarantee objectivity, an open-to-the-public dataset, cephalometric images accompanied with coordinate values of 19 landmarks, were used. A regressional system was developed, consisted with convolutional neural networks of three phases. First phase network was to determine approximate position of each landmark, inputting whole area of compressed original images. Five secondary networks were to narrow down the area, based on the first phase prediction. Third phase networks were trained by small areas around respective landmarks, with original resolution. Third phase prediction with voting was done inputting 81 shifted areas. Successful detection rates improved as the phase advances. Voting in third phase improved successful detection rate. In comparison with previously reported benchmarks, using the same dataset, proposed system marked better results. Within the physical limitation of memory and computation, multi-phase deep learning may be a solution to deal with large images.

Three-Dimensional Craniofacial Landmark Detection in Series of CT Slices Using Multi-Phased Regression Networks

“…One solution is to compress the images [51] and then input them into deep learning networks, but the compression process results in the loss of detailed information. In this study, we took the multi-phase deep learning method, which is used to predict landmarks in 2D cephalograms [27][28][29], and applied it to 3D craniofacial images. The conceptual premise was to emulate the way that one finds a place on a map when provided with the address.…”

“…Previous studies [27][28][29] of 2D cephalograms used the database that was published at ISBI 2015, along with previous benchmarks [24,26]. The authors were unable to obtain a database of feature-point three-dimensional coordinates for craniofacial CT.…”

“…Various kinds of learning-based methods [45][46][47][48][49][50][51][52][53][54][55] have been reported. In our experience with 2D cephalograms [27][28][29], a multi-phased deep learning system was able to predict coordinate values with high precision. It first made a rough prediction for the whole area of the image; then, it marked down smaller areas of interest in the following phases.…”

“…It outperformed the previous benchmarks with the same datasets used in the challenge. The use of multi-phased regression deep learning neural networks [27] with regression voting [28,29] enhanced the prediction accuracy. Kim et al utilized a two-stage method with their own larger datasets [30].…”

Skeletal facial asymmetry: reliability of manual and artificial intelligence-driven analysis

“…23–26 Unlike the time-consuming manual cephalometric analysis, AI can assess images within seconds, reducing the analysis time by up to 80 times. 27 The widespread adoption of AI and CBCT in dental diagnostics has led to the development of several AI-based programmes, such as WebCeph (Assemble Circle, Gyeonggi-do, Korea), WeDoCeph (Audax, Ljubljana, Slovenia), and CephX (ORCA Dental AI, Las Vegas, NV, United States). These programs automatically identify anatomical measurement points, evaluate landmarks, calculate angles and distances, and generate automated analysis reports with significant findings.…”