top of page
smart inject
Smart Inject
Smart Inject
A device that automatically performs administration in the
tail veins using a collaborative robot with AI camera.
Animal
Restraint Device
Injector
Edge Computer
Robot Arm
Method of Use
Fixation of anesthetized rat/mouse
Imaging with Artificial Intelligence camera
(Application of vision machine learning algorithm)
Align with the tail vein location
and move to the robot arm
Needle detection and fine tuning
Tail vein administration
Experiment
Green Line : Needle , Red Line : Tail vein
Injection of 50 μl of saline solution
Artificial Intelligence aligns tail vein and needle, and a collaborative robot
controlled by GUI in 0.01mm increments automatically injects drugs.
（A）
(B)
Tail vein recognition using vision machine learning
（A）Tail vein recognition. Rat tail(left), vein recognized as red line(right).
(B) Schematic of U-net image segmentation model.
Pre-injection
Start injection
Post-injection
（Success）
（Failure）
（A）
Tail intravenous (I.v) injection of 8% Evans blue by automated administrator.
（A）Injection process images of success injection and failure injection.
White arrows indicate the Evans blue flowing the vein in success injection.
Pre-injection
Post-injection
Pre-unjection
Post-injection
(B)
(B) Representative images of SD-
rats that received I.v injection, sho
wing blue coloration of their eyes,
ears, nose, and paws only followin
g success injection.
（Success）
（Failure）
（C）
（C) Pre- and post-injection images of success and failure case.
%EC%A0%9C%EC%9E%91%EC%9A%A9_%EB%8D%B0%EC%9D%B4%ED%84%B0%EC%8B%9C%ED%8A%B8%20%EC%88%98%EC%A0%95-14.jpg)
Figure 1. Accurate entry into the tail vein can be confirmed through chandes in the pressure value inside the needle.
While drug injection is in progess by AIIS, the pressure sensor value inside the needle changes in a certain pattern, and the success of insertion into the tail vein can be judged through the change in pressure value. If needle insertion into the tail vein was successful, a pressure value of 7mmHg or more was observed, and a pattern was observed where the pressure rose during injection and then fell again when drug injection was completed.
%EC%A0%9C%EC%9E%91%EC%9A%A9_%EB%8D%B0%EC%9D%B4%ED%84%B0%EC%8B%9C%ED%8A%B8%20%EC%88%98%EC%A0%95-13.jpg)
Figure 2. Tail vein recognition using vision machine learning.
(A) Schematic of Deeplabv3+ image segmentation model. (B) Tail vein recognition within the GUI. Vein recognized as green line, needle recognized as red line. Accurate injection positioning with automatic adjustment of tail vein and needle position by AI.
Product Competitiveness
Administration accuracy achieved more than 90%
Enhancement in the accuracy
of tail vein positioning
Machine learning-based
image processing
Application of algorithm
Improved administration accuracy
Vein identification for AI
3D-scan library of
experiment animal
In-house development of 3D animal immobilizing apparatus
by establishing an animal library
Diversification of
administration routes
多角Implementation of multi-angle
multi-route adaministration solution
using a 6-axis robotic arm
bottom of page