“Can we see the full circle of how AI enhances dental practice?”
This is one of the many questions that prompted the design and conduct of this workshop. Dr Huang Yu-Ping is a dentist and researcher from Taiwan and is passionate about AI and radiology. His recent research was on cephalometric analysis where he developed codes towards deep learning of machines to mark 19 anatomical landmarks in a lateral Ceph.
Dr Huang used the codes he had developed as part of his research project. The research involved machine learning on the marking of 19 anatomical points used for cephalometric tracing. The 19 points were marked and labelled by experts on 400 radiographs. 300 radiographs were used for training while 100 was used to evaluate the accuracy of the machine in detecting the landmarks. Dr Huang shared his codes with the students for use in this workshop only.
Following a brief introduction on machine learning and its applications in dentistry, the students were taken through the successful training of the machine in detecting the landmarks using the 300 labelled radiographs.
The codes were uploaded to google drive and read by google colab®. We were excited when we found that the machine could detect the landmarks to an accuracy of 1-3 mm. We wondered how the accuracy will improve if there were 400000 labelled images instead of the 300 used for training.
|Few Take-home Messages from the Workshop|
|Knowing how an AI product is developed helps us understand the enormous possibility of technology. This is akin to students learning the lab aspects in the fabrication of prostheses. Understanding the process helps us to provide constructive feedback for improvements|
|Some students provided feedback that they might not need the background information on how AI is trained. The lesson plan had been provided to the students and students could make a choice to attend. The activity was recorded and was made available for all students.|
|Some students expressed interest in learning coding to write their ideas. Some students can be offered coding classes that can supplement their skills in AI.|