Venipuncture Training System


The blood collection is one of the most fundamental procedures in nursing. Trainings using simulators, which are a kind of manikins, become popular instead of trainings in clinical practices. However, conventional training simulators never provide any feedback to the trainees, which is the most important elements in skill trainings.

Therefore, the purpose of this study is development of novel training simulators for blood collections with functions to provide advices. A simulator for blood collections from the medial cubital vein was then developed. It consists of a manikin of the arm, three cameras, a RGB-D camera, and a laptop computer. It provides advices based on the results of image processing on the images obtained from the cameras.

An experiment was conducted to verify effectiveness of proposed method. The results suggest that advices provided from the simulator are useful to obtain correct skill of blood collection.

Fig.1 Conventional simulator

Fig.2 System overview


In the blood collection, the procedure of motions (setting armband, skin disinfection, etc.) is important. Therefore, we regard the blood collection as finite automaton (FA) with states of each action. (Fig.3) According to the image processing on the physician’s action, the transitions of FA are executed and evaluated.

Fig.3 Matching between input and registered image

The above three cameras have each role. These cameras recognize the location of the tools (injector by camera A, armband by camera B and cotton gauze by camera C) for the blood collection. Three image processing for each camera are integrated into the total recognition algorithm. In addition, RGB-D camera recognizes the physician’s hand motion to improve the recognition rate of injector.

Fig.4 Camera A: Injector recognition

Fig.5 Camera B: Armband recognition

Fig.6 Camera C: Cotton gauze recognition