A review of image processing leading to artificial intelligence methods to detect instruments in ultrasound guided minimally invasive surgical procedures

Ultrasound is a medical diagnostic imaging modality with the ability to visualize the underlying tissues and organs in the body in realtime. A handheld probe with piezo electric crystals radiates a beam of ultrasonic sound waves in a thin plane, acoustic microphones detect the bounced sound wave and images of the underlying tissue structures are depicted on screen. The realtime advantage has led to ultrasound's alternative use as a guidance for medical treatments such as minimally invasive procedures. Minimally Invasive Surgical procedures require skillful interpretation of the ultrasound image to distinguish the location of the instrument, it's motion and maneuver it via different tissue regions to reach a specified target. A software solution which outlines the location of the instrument would greatly assist medical doctors. This review presents the problem statement and various methods utilized to detect instruments. The methods using pure image processing algorithms are presented with various ranges of success. Recently, the use of Artificial Intelligence (AI) methods have been explored and these show improved detection. Such applications of AI in medicine could result in a strong impact and should be advocated further. The challenge of clear identification of instruments continues to persists in order to develop a system which can provide medical doctors real-time and highly accurate ultrasound guided minimally invasive instrument detection.