A Novel Approach of Controlling Stoppage of Drip Infusion Using Image Processing on Raspberry PI Platform

Abstract

Intravenous drip diffusion is a common practice to treat patients in hospitals. During treatment, nurses must check the condition of the infusion bag frequently before running out of fluid. This research proposes a novel method of checking the infusion bag using an image processing technique on a compact Raspberry PI platform. The infusion monitoring system proposed here is based solely on capturing the image of the infusion bag and the accompanying bag/ tube. When the infusion fluid enters the patient, the surface of the liquid will decrease, and at the end will reach the bottom of the infusion bag. When the image of the fluid surface touches the bottom of the infusion bag, a mechanism will trigger a relay, and then activate a pinch valve to stop the flow of the infusion fluid before it runs out. The entire system incorporates a digital camera and Raspberry as the image processor. The surface of the liquid is determined using the Canny Edge Detection algorithm, and its relative position in the tube is determined using the Hough Line Transform. The raw picture of the infusion bag and the processed image are then sent via a wireless network to become part of a larger system and can be monitored via a simple smartphone equipped with the proper application, thus becoming an Internet of Things (IoT). With this approach, nurses can carry on other tasks in caring for the patients while this system substitutes some work on checking the infusion fluid.

Deep learning-based monitoring technique for real-time intravenous medication bag status

“…For example, Song et al [ 16 ] set an additional camera holder aside the IV pole to photograph a pole-hanged IV bag and monitored the liquid level variations in the drip chamber using conventional image-processing techniques based on Canny edge detection. Pranjoto et al [ 17 ] added an additional camera arm to the IV pole to photograph two pole-hanged IV bags (180° spacing) and monitored the liquid level in the IV bags based on Canny edge detection. Both studies used camera images for liquid residue monitoring instead of contact sensors; however, they did not apply DL techniques.…”

“…Previously, most studies focused on contact-sensing methodologies, for example, attaching various sensing devices, such as photo-sensors, capacitive sensors and RFID tags, on the surface of a drip chamber or an IV bag [ 6 – 13 ], embedding load cells into an IV pole to measure the weight variations of IV bags hung on the pole [ 14 ], and integrating miniaturized flow sensors into an IV catheter [ 15 ]. Recently, some researchers have focused on contactless methodologies using conventional image-processing technologies [ 16 , 17 ]. Artificial intelligence (AI) techniques are being developed rapidly; therefore, AI techniques have been applied in several IV-treatment monitoring purposes using cameras [ 18 , 19 ].…”

A preliminary study of ecosmart intra venous system on patients’ comfort

“…However, the system developed has not been equipped with a feature to program the number of droplets infusion fluids. In this system, the nurse manually adjusts the tension of the roller that clamps the hose on the infusion set so that there are (Pranjoto et al, 2021). Likewise, the device is not equipped with a battery, so it can still be carried out when the patient changes therapy with this device.…”

The impact assessment of automated drip infusion control using weighing scale and pinch method on subjects

“…There have also been many types of research on drip infusion automation devices via wireless connections [11][12][13][14][15][16][17][18], but the development carried out in this study has a special feature i.e. an indication of reduced volume of infusion fluid from the measured infusion weight [9][10], [13], [19][20]. A Fluid intravenous alert monitoring device is even potentially unable to detect when the fluid runs dry [21].…”