Page 51 - GTM-1-2

Global Translational Medicine A Taxonomy of AI Assisted Medical Robots

1. Introduction current challenges to lay a foundation for the three major
trending research in Section 3, as shown in Figure 2.
The past decade has witnessed the advancements
achieved by artificial intelligence-assisted robotics 2.1. Surgery automation
for therapeutic tasks (AI-ART). In consideration of
commercial translational requirement and better artificial 2.1.1. Gripper contact force sensing
intelligence application to therapeutic tasks, AI-ART can Due to the limitations in the development of physical sensors,
be categorized into three groups based on their applications the contact force sensing techniques that are used in current
and underlying principles. The nascent application of commercial or academical research tend to “isolate” the
artificial intelligence-assisted robotics for therapeutic hands of surgeons from the tissues or skin of the patients [45,47] .
tasks, which includes the exploration application of robotic The tactile sensation isolation from manual instruments
manipulators to fulfill standard surgical procedures and or artificial intelligence-assisted medical robots could be
the incorporation of early artificial neural network or disastrous, especially in surgical tasks like tissue retraction
statistics-based algorithms, has had its ups and downs. surgery, during which deformable connective tissues would
In this review, we regard medical artificial intelligent be manipulated recurrently . To sense and control the
[48]
robots, such as laparoscopic robots, medical wearable interaction force while using artificial intelligence-assisted
equipment that has benefited from artificial intelligent robotic techniques for therapeutic tasks, researchers are
algorithms, and intelligent soft medical robots, as AI-ART. exploring the recreation of tissue palpation, temperature,
The chronological development of AI-ART based on and even corrosive sensing with improved gripper design .
[49]
methodology and milestones is shown in Figure 1.
To address the technical sensing problem, advancements
[50]
2. Taxonomy of artificial intelligence- have been made. Luca et al. presented a simulation of
assisted robotics for therapeutic tasks Ruffini receptors with deep neural networks and optical
gratings, which could be applied to manufacture tactile-
By setting the search criteria and selecting influential sensitive skin. This bio-inspired polymetric matrix skin
journals, such as Web of Nature, Cell, and other journals, might be a novel research direction to implementing tactile
involving AI-ART in various aspects over a long time sensation while owning a different principle with that of
span, we identified the taxonomy range. The topics of human. The researchers employed the convolutional neural
the taxonomy are related to specific clinical therapeutic network (CNN) to decode the fiber Bragg grating sensor
tasks (e.g., laparoscopy surgical robots in Section 2.1) or signals, achieving median errors of 35 mN and 3.2 mm, and
enabling technologies that help to remove urea for dialysate demonstrating the advantages of CNN algorithm. Tae et
regeneration for wearable artificial kidney (e.g., medical al. proposed the use of leech-inspired dry electrodes for
[51]
wearable robots in Section 2.3). A detailed exposition of auxiliary blood pressure sensing through surgical robots.
each clinical therapeutic task itself warrants a survey, but in Although their work was not directly meant to sense the
this work, we focus on the problems during the integration interaction force between the contact point of the robotic
process and the latest trending methods proposed in gripper and the patient’s tissue or skin, it provides an optional
AI-ART for each clinical therapeutic task. We also survey method to monitor the fluctuations in blood pressure during
the clinical artificial intelligence improvement in robots, the whole surgical procedure, as shown in Figure 3.
which empower the effective monitoring and update of the
applied AI-ART in specific clinical therapeutic task. In this 2.1.2. Automated surgery
section, we focus on summarizing the development of each Prolonged operation is often indicated in research or review
subtopic of laparoscopy surgical robots and discuss the works as the major risk for complications after surgery.

Figure 1. Chronological development of AI-ART in the past decade from the early 2010s’ magnetic helical microrobots to squamous lung cancer detection
and therapeutic tasks. The illustrations in each year are recreated and referenced as follows: 2012 , 2013 [2,3] , 2014 [4,5] , 2015 [6-9] , 2016 [10-13] , 2017 [14-17] ,
[1]
2018 [18-21] , 2019 [22-27] , 2020 [28-32] , and 2021 [33-38] .
Volume 1 Issue 2 (2022) 2 https://doi.org/10.36922/gtm.v1i2.176