Dear participants,

Please read below for some clarifications based on questions we received from fellow participants of this challenge.

Dataset, optimal/suboptimal images: optimal and suboptimal frames are defined based on how closely the images approximate to the definition of the standard plane for abdominal circumference measurement. Because of this, it is true that not every case will have an optimal frame. It is also true that not every case will have a suboptimal frame.

AI task: the task is to find the best possible frame in each case. This may correspond to an optimal frame, if such frame is available; or to a suboptimal frame, if no optimal frame is available. This is evaluated by the Weighted Frame Selection Score (WFSS). This metric will assign full marks if you choose a suboptimal frame where no optimal frame existed. It will penalise your algorithm if it selects a suboptimal frame where an optimal one was available. Your algorithm should only output -1 when it does not find any suitable (optimal or suboptimal) frame for AC measurement. Otherwise, it should provide the frame number representing the image that was identified as either optimal or suboptimal and selected for measurement. Remember that the provided mask should be a binary mask.