CEPHA29: AUTOMATIC CEPHALOMETRIC LANDMARK DETECTION CHALLENGE

In modern orthodontics, quantitative cephalometric analysis is the most common clinical and research tool that plays an essential role in orthodontic diagnosis and treatment planning. The accurate and precise localization of cephalometric landmarks enables the quantification and classification of anatomical abnormalities, however, the traditional manual way of marking these landmarks is a very tedious job. Endeavours have constantly been made to develop automated cephalometric landmark detection systems but they are inadequate for orthodontic applications. The main reason for this is that the number of publicly available datasets as well as the images provided for training in these datasets are insufficient for an AI model to perform well. To facilitate the development of robust AI solutions for quantitative morphometric analysis, we propose the CEPHA29 Automatic Cephalometric Landmark Detection Challenge. In this regard, we present a benchmark dataset consisting of 1000 cephalometric X-ray images acquired from 7 different radiographic imaging devices. The clinical experts of our team labelled each cephalogram in the dataset with 29 cephalometric landmarks and the CVM stage of the patient. The challenge requires researchers to develop algorithms that (1) automatically localize 29 anatomical landmarks on the cephalogram, and (2) classify the patient’s cephalogram in one of the six CVM stages. The “Train” set along with the ground truth annotations will be released to the participants so that they can design and train their landmark detection methods, whereas the “Validation” set will be provided to the participants without ground truth annotations so that they can evaluate the performance of their proposed method to qualify for the final round. Finally, a “Test” set will be used by the organizers to finalize the best-performing models/methods. Our challenge will not only help derive forward the research and innovation for automatic cephalometric landmark identification but also prove to be the beginning of a new era in the field. For additional information about the challenge, have a look at our preprint.