Technical Report Draft 1
Background
According
to the da Vinci Surgery website, the da Vinci surgical system is a
robotic-assisted surgical system that enables surgeons to operate “minimally
invasive surgery” using sophisticated surgical tools. The da Vinci System
consists of 3 separate components. Firstly, the surgeon console presents the
surgeon with the ability to control the surgical instruments while viewing the
surgery in high-definition 3D. Secondly, the patient cart is placed beside the
operating bed, embedded with the camera and tools for the surgeons to control.
Thirdly, the vision cart is the bridge that communicates between the components
and supports the high-definition vision system.
Even with
such sophisticated design and technology, the da Vinci surgical system is still
being limited as a master and slave system. This means that the da Vinci
surgical system relies on human inputs to process and function. Based on the
article “Accidents Happen” (2019), it states that KK Women’s and Children’s
Hospital’s (KKH) sees an average of 60 to 70 cases that require stitches each
week. With the absence of automated robotic technology, doctors are required to
attend to these patients. With such reliance on human intervention even for
cases that require small surgeries, it takes a toll on the workload of doctors.
This thus decreases their working efficiency which possibly extends the working
hours of doctors. According to The Washington Post, the article “Back to
extremely long shifts for new surgeons? Study finds few negatives.” (2016)
states that the average working hours of doctors fall between 16 to 28 hours
per week. Such increased working hours would then affect a doctor's ability to
make a sound judgment.
Thus,
we introduce the concept of incorporating artificial intelligence (AI)into the
system. Machine learning is a type of AI that allows computers to self learn
through the analysis of patterns without any explicit coding needed. Unlike
humans, robots are excellent at seeing patterns of big data and then producing
an accurate list of predictions within a very short time span. With the
incorporation of AI into the current technology, the system that previously
required human intervention can now function on its own. This greatly cuts down
the manpower needed for trivial cases ie. stitching.
Problem
Statement
Medical
robots have always assisted surgeons but they are limited to a master-slave
system. This over-reliance of the surgeon’s attention even for minor injuries
including small cuts and wounds decreases their working efficiency.
Purpose
Statement
This report proposes an improvement to the current Da Vinci
System to turn it into an ideal medical robot for performing auto-suturing on
minor wounds. Through the implementation of specifically designed algorithms
and medical data into the Da Vinci System, the robot will be able to perform
suturing without the guidance of a doctor.
Proposed
Solution
Introducing
AI into an already well-established surgical robot like the da Vinci surgical
robot, which has operated over 6 million successful surgeries, will help
minimize the resources needed to develop an entirely new robot. The da Vinci
system is equipped with a set of needle drivers, which are surgical instruments
needed for performing suturing at different angles. By introducing machine
learning into the da Vinci system, it can be thought of as an improvement to
the system as the robot can perform suturing with well-designed suturing tools
in the absence of a surgeon, while still having the master-slave option
available for major surgeries.
Suturing autonomously by
medical robots is made possible through the implementation of a trajectory
planning algorithm. The Raven II surgical robot, similar to the da Vinci
system, performed the auto-suturing with the integrated algorithm. Before
suturing begins, the kinematics of the needle held by the end effector of the
medical robot is analyzed thoroughly. This is to translate the current and end
pose of the end effector into data. The trajectory planning algorithm is broken
down into two components, “GoToPoint” and “PathGeneration”. “GoToPoint” brings
the robot to the target point, while “PathGeneration” constructs a trajectory
path for the robot to insert the needle into the tissue. Once the trajectory
path is generated, the medical robot will begin from its initial position and
follow the trajectory. The algorithm allows surgeons to request for
auto-suturing, however, for safety concerns, surgeons can interrupt and halt
the process.
The implementation of large
amounts of data by artificial intelligence can be processed by a system
structure similar to that of Artificial Intelligence in Medical Epidemiology
(AIMe). A data set is preloaded into the system and used for comparison against
every action taken by da Vinci.
Benefits
The
implementation of artificial intelligence (AI) allows the robot to apply
machine learning. The robotic hands will be able to learn the procedures
sequentially and therefore enables it to perform the procedures semi or fully
automatically.
The
implementation of AI also indirectly affects the efficiency of hospitals,
especially the Accident and Emergency (A&E) departments. The robotic hands
will be able to replace doctors or surgeons with tasks like stitching, freeing
them up for other patients that require more attention. This will boost the
efficiency of the A&E department of the hospitals which means that more
patients can be treated within the same amount of time compared to having
doctors or surgeons being there physically to stitch the patients up.
With
the implementation of machine learning through AI, the robotic hands will learn
and perform procedures in a standardized and sequential manner. The robotic
hand will pull the algorithm of the task from a database that it is assigned to
perform. This ensures that the procedures performed by the robotic hands are
sequential and standardized. Hence, it minimizes the possibilities of errors on
tasks made by humans through the implementation of AI. Furthermore, machine
learning enables robotic hands to analyze uncertainties such as the dimensions
of the wound, for accurate error propagation which further enhances its
capabilities.
Evaluation
Though
trajectory planning algorithm allows medical robots to perform auto-suturing,
certain limitations and settings need to be in place for the operation to work.
The suturing performed by the Raven II robot was not tested on a tissue model
due to the algorithm not being able to detect tissue movement. In a realistic
clinical setting, patients move as they breathe, causing tissue movement. The
angle of entry point for the insertion of the needle is crucial as it
determines the exit point of the needle. The trajectory planning algorithm is
not advanced enough to set the optimal angle of entry by itself, hence, a
desired entry point is adjusted by the surgeon before suturing begins.
The
prediction of the AI model depends on the analysis of the data input. Such
analysis is dependent on the availability of the data set that is present in
the AI model for analysis. Thus, the prediction is not perfect at a 100%
success rate as there may be a presence of a new situation where data on
dealing with it is absent.
At
the end of the day, the accuracy of the system in terms of diagnosis and
performing is still not guaranteed. The current algorithm is not sophisticated
enough to take tissue movement into consideration. The trajectory planning
algorithm is still in its early development stage; hence, more work needs to be
accomplished for auto-suturing to come into fruition. Although there is a
possibility of failure by the predictive model for analysis, the implementation
of the system is guaranteed to be free from human errors when performing
suturing.
Methodology
Research
articles and websites were used as references for information and data to
complete this report.
Secondary
research
In
building a strong design proposal, the team did extensive research using the official
product website on the Da Vinci Surgical System to identify their strengths so
as to integrate them into MediHand and also their weaknesses for possible
modifications to make MediHand a comprehensive product in its field. Research
articles were used such as the auto-suturing algorithm performed by the Raven
II robot, which supports the proposal report as the Raven II is like the da
Vinci system. The secondary sources were also used to strengthen the
credibility of the design proposal through examples and events such as having
over six million successful surgeries done by the Da Vinci Surgical system and
the suturing autonomously with the Raven II Surgical Robot through artificial
intelligence.
Conclusion
In
conclusion, for MediHand to be an autonomous surgical hand capable of
performing suturing, two modifications: implementation of machine learning and
algorithms for suturing should be implemented. These modifications allow
Medi-Hand to be capable of performing suturing autonomously without doctors’
intervention. This creates a much ideal situation in which doctors’ can be more
focused on much urgent matters.
References
AIME
Healthcare Sdn Bhd (n.d.). AI4Good. Retrieved from https://betterproposals.io/proposal/index.php?ProposalID=sE6L0xI0UtPfZjBBINS3JXP2vpVXlh2M18n36aOFVuc&ContactID=-4khI1mEueUpsUws_jFpjaxL0_ACn7uMzUgG_Gl55UE
Channel
NewsAsia. (2019, November 26). Accidents happen - but when might a child need
plastic surgery for scar therapy? Retrieved from https://cnalifestyle.channelnewsasia.com/wellness/plastic-surgery-children-accident-12123156
Dehghani,
H., Farritor, S., Oleynikov, D., Terry, B. (2018). Automation of Suturing Path
Generation for da Vinci-Like Surgical Robotic Systems. Retrieved from
Intuitive.
(n.d). Retrieved from https://www.davincisurgery.com/
Kang, R.,
Branson, D. T., Gulielmino, E., & Caldwell, D. G. (2012). Dynamic modelling
and control of an octopus inspired multiple continuum arm robot . Computer
& Mathematics with Applications, 64(5), 1004–1016. Retrieved from https://www.sciencedirect.com/science/article/pii/S0898122112002234
The
Washington Post. (2016, February 03). Back to extremely long shifts for new
surgeons? Study finds few negatives. Retrieved from https://www.washingtonpost.com/news/to-your-health/wp/2016/02/02/back-to-extremely-long-shifts-for-new-surgeons-study-finds-few-negatives/
Comments
Post a Comment