Fig 1.
Robotized system representation.
Robotized system representation on a Leica DM750 optical microscope from Microbiology Laboratory of Drassanes-Vall d’Hebron International Health Unit. Red discontinuous arrows represent space positions. Blue discontinuous lines represent connections. (1) Storage and focus 3D pieces. The Arduino controller is stored in the grid box. The servo (Z) motor is directly connected to the Arduino controller and subjected by a holder arm to change the Z position. (2) Microscope stage pieces. Three individual stage pieces were placed on the microscope stage. Two servo motors are attached to the stage 3D pieces to move the sample through the X-Y axis, and connected to the Arduino controller. (3) Mobile phone adapter pieces. The smartphone adapter is positioned on the ocular lens of the microscope. The smartphone is connected via Bluetooth (BLE) to the Arduino controller to guide the entire robotized procedure.
Fig 2.
Representation of the electronic circuit and configuration.
(S) Servo motor; (V) Volts; (R) Resistance; (Ω) Ohms; Ground (GND).
Table 1.
Optical microscope measurements for the design of 3D universal adaptable pieces.
Table 2.
Description of all 3D pieces employed for automated system building and assembly.
Fig 3.
All pieces were designed with Ultimaker Cura 5.1 slicing and Tinkercad softwares. (A) Microscope stage pieces (1) Biological sample gripper to hold the slide. A toothed track allows for horizontal movement with the horizontal servo motor. (2) Stage holder. A toothed track allows for vertical movement with the vertical servo motor. (3) Main microscope stage piece to support the system. (B) Microscope auto-focus and storage pieces (1) Servo motor holder. This piece had an adjustable height to be attached on the fine adjustment wheel of the microscope. (2) Wheel holder with adjustable diameter. (3) Storage box for Arduino and board controllers. (C) Mobile phone adapter pieces for microscope ocular lens. An adjustable clamp allows the adapter to be attached to the eyepiece lens. The roughened PLA mount holds the smartphone device in order to capture images without moving.
Fig 4.
Pseudo-code of the android application settings (algorithm 1) and arduino controller (algorithm 2).
Fig 5.
Sample scanning X-Y movement representation.
Automated snake-like movement of the system to capture and scan the whole blood smear samples. Automated scanning strategies using imaging techniques for object detection through X-Y axes. Snake-like movement for image acquisition of a rectangular shape sample. Images from Microbiology Laboratory of Drassanes-Vall d’Hebron International Health Unit.
Fig 6.
Variance of Laplacian values for auto-focus system analysis.
A total number of 30 microscopic fields (images) of 6 different Giemsa stained thick blood smear samples (5 FoV/sample) for malaria diagnosis were analyzed. Each FoV has a range of Variance of Laplacian values depending on the position/image [60 total positions (30° on each side)] of the fine adjustment microscope wheel. Focused images were represented as the higher Variance of Laplacian values (peaks correspond to in-focus positions/images).
Table 3.
Autofocus analysis of thick blood smear Giemsa stained.
Table 4.
Summary table comparison of the state-of-the-art automated microscope designs.
Fig 7.
Microscopic images from Microbiology Laboratory of Drassanes-Vall d’Hebron International Health Unit.
Giemsa stained thick blood smear sample with detection of leukocytes and malaria trophozites by YOLOv5x neural network performance. 1000x magnification.
Fig 8.
Microscopic images from Microbiology Laboratory of Drassanes-Vall d’Hebron International Health Unit.
Urine sediment sample with detection of Schistosoma haematobium eggs by YOLOv8x neural network performance. 100x magnification.