Monitoring of Lung Ultrasound Acquisition using Infrared Sensors and Artificial Intelligence

Lung diseases contribute significantly to global mortality rates. The conventional diagnostic techniques based on medical imaging used for lung disease diagnosis normally require specialized personnel and complex infrastructure, posing challenges in rural areas. In response to these challenges, Volume Sweep Imaging in the lung (VSI-L) is an ultrasound-based acquisition protocol designed to empower non-specialized healthcare providers by following the movement of the transducer when performing a set of ultrasound scans along the chest. Later, the acquired data is sent to the radiologist for future analysis through a telecommunication system. VSI-L has been tested in clinical trials showing its capacity for tele-ultrasound. Even though it is a standardized protocol, human error remains a concern, which is reflected in not maintaining the correct position and speed of the transducer or producing videos that are difficult for the radiologist to interpret. In response to these challenges, a training system based on an infrared sensor is proposed that allows following the trajectory of the ultrasound transducer, through which the movement coordinates are acquired, which through a Machine Learning program are classified to evaluate whether The ultrasound procedure was performed correctly. This approach showed positive results when classifying the traces made, obtaining an accuracy of approximately 95%.