Image Processing for Monitoring Diagnostic Devices with Automated Pipetting

Abstract

This paper presents an image processing technique for monitoring diagnostic devices with an automated pipetting function. A video camera is employed to inspect whether the device operation, including the tip holder position, the absence of the tip, and the alignment of the holder and the tip, is appropriate. The processing results can be utilized for indicating the overall malfunction of the entire device or for warning that the user has forgotten to load the required tip. The image of the holder and the tip area is converted into a binary image with the fixed threshold obtained during device calibration. A horizontal projection analysis is utilized for robust identification of the tip and the holder area, and a vertical projection analysis is carried out for each area to detect the absence of the tip and the alignment of the tip and holder positions. The camera was added to a commercial diagnostic device, and the processing algorithm was implemented in the main controller of the device. Experimental results show that device monitoring with the proposed image processing technique worked well without failure.