Microcontroller-Based Automated Microscope for Image Recognition of Four Urine Constituents

In this study, an automated microscope was developed that mechanically focuses and captures images of the urine sediment for image recognition of four urine constituents in the urinalysis examination: (1) Red Blood Cells (RBC), (2) White Blood Cells (WBC), (3) Bacteria, and (4) Calcium Oxalates (CaC₂O₄) using PIC microcontroller. Automatic image-focusing and stage movement were successfully achieved using stepper motors interfaced with PIC18F25J50 microcontroller. The microcontroller circuit board was mounted and attached to the fine-adjustment and stage-control knob of the microscope. The Edge Detection Method using Prewitt operators and a proprietary algorithm for moving lens was efficiently used for image auto-focusing. Image acquisition and mechanical actuators of the microscope were integrated into a software, enabling automation of microscope operations. A Capture tab in the GUI was designed to gather 70 images of the urine sediment before the system starts to analyze the images. Digital image processing and Artificial Neural Network created in Matlab were used to recognize and count the constituents which were incorporated in the GUI. The result of urinalysis was generated in PDF. Implementation of the system was done using 100 urine samples by examining same specimen with the automated and manual microscopy at the same time. Overall, automated capturing and recognizing of the four most common urine constituents in the microscopic examination generate results with an average speed of 3.66 minutes and an accuracy of 89.5%.