📝 Classifying the skill level of laparoscopic surgeons on resource-constrained systems 📝

Great news today! Our paper "Novel low memory footprint DNN models for edge classification of surgeons’ postures" is being published on IEEE Embedded Systems Letters thanks to the joint efforts of our wonderful team: Alex, Terry, Marco, Maria Celvisia, Esther, and Luciano from Loughborough University, STMicroelectronics, University of Leicester, and University of Cambridge!

This scientific study looks at ways to improve training for laparoscopic surgery and decrease injuries related to these procedures. The researchers used advanced computer programs called deep neural networks to help with this. However, these programs often require a lot of computer power, which can be a problem. The study introduces two new computer models that use less power and can be used on smaller devices, like those found in laparoscopic surgery. These models were tested on small processors and the results showed that they were effective in classifying surgical skill levels while also being efficient in terms of memory and accuracy.

This study found that analyzing just the upper arms can be a good way to evaluate a surgeon's skill level. This is especially useful in situations where it's important to maintain sterility and when a small device is needed for processing and classification. These findings are just the start and more research is needed to confirm these results.