Automatic muscle tendon junction tracker using deep learning.
The idea for this AI tool flashed through my mind mid 2019 not wanting to spend hours and hours evaluating image data by hand. “State-of-the-Art can do better!”, I thought.
But first this needed scientific prove. It was not until the evening of a stormy farewell party at the Institute of Physics (University of Graz) in November 2019 when Robert Jarolim and I decided to give it a try. Well lets say, we did have some dense Christmas holidays.
We developed and published this opensource version 1.0 of #deepMTJ, an automatic muscle tendon junction tracker using deep learning. Our trained network shows robust detection of the muscle tendon junction on a diverse data set of varying quality with a mean absolute error of 2.55 ± 1 mm. We show that our approach can be applied for various subjects and can be operated in real-time (up to 7x faster then previous methods employing computer vision algorithms).