Robot-Assisted Fracture Surgery: Surgical Requirements and System Design
Robot-Assisted Fracture Surgery: Surgical Requirements and System Design
IOANNIS GEORGILAS 1 2 3
GIULIO DAGNINO 0 2 3
PAYAM TARASSOLI 2 4
ROGER ATKINS 2 4
SANJA DOGRAMADZI 2 3
0 The Hamlyn Centre for Robotic Surgery, Imperial College London , London SW72HR , UK
1 Department of Mechanical Engineering, University of Bath , Building 4E3.45, Bath BA27AY , UK
2 Mechanical Engineering, University of Bath , Building 4E3.45, Bath BA27AY , UK. Electronic mail:
3 Bristol Robotics Laboratory, University of the West of England , Coldharbour Lane, Bristol BS161QY , UK
4 University Hospitals Bristol , Upper Maudlin Street, Bristol BS28HW , UK
-The design of medical devices is a complex and crucial process to ensure patient safety. It has been shown that improperly designed devices lead to errors and associated accidents and costs. A key element for a successful design is incorporating the views of the primary and secondary stakeholders early in the development process. They provide insights into current practice and point out specific issues with the current processes and equipment in use. This work presents how information from a user-study conducted in the early stages of the RAFS (Robot Assisted Fracture Surgery) project informed the subsequent development and testing of the system. The user needs were captured using qualitative methods and converted to operational, functional, and non-functional requirements based on the methods derived from product design and development. This work presents how the requirements inform a new workflow for intra-articular joint fracture reduction using a robotic system. It is also shown how the various elements of the system are developed to explicitly address one or more of the requirements identified, and how intermediate verification tests are conducted to ensure conformity. Finally, a validation test in the form of a cadaveric trial confirms the ability of the designed system to satisfy the aims set by the original research question and the needs of the users.
System design and development; Computer-as-
sisted surgery, Medical robotics, Percutaneous fracture
Medical devices must be well-designed to provide
high quality care for patients.28 To be considered
‘welldesigned’, a medical device must be clinically effective
and safe, while also able to fulfil the needs of the
users.30 This requires taking into consideration a
number of factors including the capabilities and
working pattern of the clinical users, the needs of the
patients, the environment in which the device will be
used, and the system(s) of which it will be part of Ref.
29. All these factors will inform the design of the
device. Poorly designed devices increase the risks for
human error,23 as well as for incidents and accidents in
To increase the adoption rate of a medical device,
developers must have a clear and thorough
understanding of the clinicians, patients and carers who will
use the device.30 Conducting an early user research is
necessary for developers to understand and specify the
context of use and the user and organizational
requirements.24 Failing to adequately study the
potential users at the beginning of development may
result in assumptions about their needs, capabilities
and characteristics. So, the device will be developed
and evaluated based on incorrect information. This has
serious implications not just for the safety of the new
device, but also for its commercial success.30 The
development of medical devices in both commercial
and research domains1 as well as the regulatory bodies,
i.e. Food and Drug Administration (FDA) in the US
and Medicines and Healthcare products Regulatory
Agency (MHRA) in the UK, strongly suggest that a
user-driven approach is necessary to ensure a
functional product for the clinical, safety–critical
Although some manufacturers of medical
equipment already integrate human factors principles in
their products, there is still a lack of commensurate
work on the practicalities of such engagement.6
Therefore, a user-centred approach should be
conducted at the early stage of a development project in
order to obtain a better and safer product7 by
including the needs and views of the users.
Based on the user needs, a set of requirements can
be developed to drive the design process.
Unfortunately, the needs are usually abstract and expressed in
natural language making it difficult to formulate
technical specifications. Capturing and organising
requirements is a crucial part of the design process.16
Technical requirements can be derived by using user
proxies in the form of expert evaluators.35 A
framework using ontological charts to capture the user needs
along with other constrains and assist with the design
process for medical devices has been proposed.21 A
common theme in incorporating user-views is using
information modelling techniques,22 for example the
V-model of design.18
The V-model is a waterfall approach that
encourages up-front planning for the development process. It
allows for a systematic testing and validation regime
for the entire development life-cycle,31 aiming to follow
a good design approach that incorporates validation as
a main development activity and not an afterthought.2
In this paper we describe the user-driven approach
used in designing and developing a system for
robotassisted fracture surgery (RAFS). The RAFS project
aims to develop a robotic system that assists surgeons
to perform reduction of intra-articular fractures in a
minimally invasive way. It provides the surgeon with
physical and virtual assistance to minimise operational
time and issues associated with open surgery (i.e.
infection), leading to shorter recovery times and
postoperation costs. The system has been developed in
close collaboration with clinicians and has been tested
in realistic conditions.13
User-driven design is widely implemented in robotic
applications for medical systems.25,26 The approach
proposed here is based on an early-stage user study, to
capture user needs, and the V-model of system
development. The user study consists of a series of
interviews with surgeons to understand the clinical practice,
instruments used, and procedural challenges. An
earlier prototype of RAFS (Fig. 1)33 was presented to
provide context for the interviews. Based on the
information gathered the requirements for the system
were elicitated and using the V-model the system was
developed by a suitable workflow, system architecture
and sub-systems along with their respective testing
criteria and metrics. The individual functionality has
been verified at the sub-system level and integrated and
tested to the complete system. The final testing and
validation was conducted on cadaveric specimens
demonstrating the ability of the re-designed system to
satisfy the originally set requirements. A final user
study was conducted after the system validation to
gather clinicians’ assessment of the test results and
potential utilization of the system in the clinical
practice. This was part of a broader health economics and
market research of robot-assisted fracture surgery.
This paper will present the methods used, namely
the qualitative method and the details of end-user
interviews and the V-model of design for system
development. In the results section we will first present
the requirements that have been derived from the
interviews and then how these have been met by the
architecture, workflow, and sub-systems of the RAFS
system. Finally, at the discussion we will summarise
lessons learned from the design process.
MATERIALS AND METHODS
The end-user part of the design process involved a
qualitative study conducted through interviews with
orthopaedic surgeons experienced in intra-articular
fracture reduction. This study consisted of two phases:
) define the objectives of the study; and (
interviews with potential clinical users of the device to
specify the requirements for the RAFS system.
The RAFS development team (DT) is composed of
three engineers and two orthopaedic surgeons. In the
first phase of this study, the DT discussed identification
of potential users and applications for the proposed
device in order to focus the study on the needs of the
users and to collect data that can be easily
implemented in the design and development process.
The DT recognised the following research
Identify the target clinical users.
Identify the potential clinical application for the
Identify barriers to safe and effective system
Collect user opinions on possible design features.
Refine and validate the concept for the new device.
The DT identified that the most suitable pilot
clinical application for RAFS is a distal femoral fracture
(DFF). This was due to the large fragments created in
this kind of trauma and the relatively simple soft tissue
structures in the region of the distal femur when
approaching from the anterior side. For this reason,
potential users for the RAFS device are orthopaedic
surgeons with expertise in knee fracture reduction.
Interviews with Clinical Users
In the initial user study a total of 13 individual
faceto-face, semi-structured interviews were conducted
with experienced (average experience 16 years)
orthopaedic surgeons from the UK, and the EU
(Table 1). Each interview lasted between 25 and
60 min. During the interviews we adopted an approach
for the surgeons to discuss as freely as possible joint
fracture reduction surgery and related issues and
limitations. Probing questions were used when necessary
to encourage the surgeons to provide more details.
Additional questions were used to clarify the themes of
The aims of the interviews were: (
) to investigate
the current state-of-the-art in joint fracture reduction
surgery with focus on DFFs; (
) to investigate
limitations and issues related to the current surgical
) to define users’ requirements for RAFS in
terms of its operational characteristics (e.g. size,
integration in Operating Theatre (OR), interaction, etc.);
4) to define expected medical functions for RAFS.
Familiarity with other robotic systems and
imagebased technologies was taken into account to
normalise the sample for personal experiences and
A broader market research was conducted by an
external company. As part of that 18 Orthopaedic
Surgeons and Heads of Orthopaedic Departments
from US, UK, and Germany were interviewed. The
aim of these interviews were to assess the potential of
) the system adoption from the financial viewpoint,
) the proposed clinical workflow, and (
usability of RAFS. The results related to (
) and (
will be further discussed.
The recordings were transcribed for the data
analysis to produce results strictly linked with the research
objectives defined by the DT in the first phase of this
The interviews were transcribed, categorised and
coded according to the grounded theory method.4
Categories and example codes are showed in Table 2.
The coded data revealed surgeons’ ideas and
opinions (common and conflicting) from which we
generated the system requirements.
Operational, Functional, and Non-Functional
There are different types of requirements. The
operational requirements define the major purpose of
the system. Functional requirements specify what the
system has to do in order to satisfy the operational
requirement. Non-functional requirements define
system constrains or modifying influences on the system.
Non-functional requirements can be split into the
performance requirements that define how a function
should be implemented and system requirements on
external parameters that are affecting the design of the
system. The non-functional requirements can lead to
errors and safety compromises15 and should be defined
using methods to ensure appropriate definition.20,27
In this work, an approach similar to the work from
Ulrich and Eppinger17 is followed. Namely, the user
defines the operational requirement, and in the case of
an expert user, provides insights into functional and
non-functional requirements. The requirements are
organised in a hierarchy, with functional requirements
being the top-level requirements and the
non-functional requirements being more detailed. Most
requirements will be defined from regulatory, safety,
and environment constrains that can be initiated by a
user but involve a degree of expanding based on the
technical literature and practice. In the specific study,
the regulatory and safety environment was dictated by
current FDA and MHRA guidance and
requirements.32 Based on this analysis, the coded data was
converted into functional and non-functional
requirements. The main approach was to convert any need or
desire expressed from the users into a technically
described description. For functional requirements the
system was required to ‘‘achieve’’ a goal, while for
nonfunctional requirements the system was required to
‘‘satisfy’’ a criterion.
V-model of the Development Process
The V-model is based on the principle that the
development process is moving from the generic to the
specific up to the lowest level of resolution usually the
component level and then the integration process
follows the reverse direction. It is important to note that
the downwards process is not only setting requirements
and technical specifications but also the criteria and
methods for testing integration on the upwards
The implementation of the V-model needs a
description of the overall system architecture in order
to satisfy the criteria, i.e. the fundamental components
required to achieve the functional requirements. Based
on these division of functionality, each sub-system is
described in detail to address the requirement. Finally,
the units of the sub-systems are defined to address
User Study Outcomes
One key point in the development of a new medical
device is to understand the application field of the
system. The results from the qualitative study
emphasised the current surgical procedure and the limitations
for using a minimally invasive approach in DFF
it/geh litya itccoanu itsssuem ing ltrconom tye iliitrzaone ittronag fftccoonepo ftrrrcsauee
izeSw pedeS itrobP edR ftoS Iagm tyseS IUG faS tS Ine roP thO
iiangm iiegngm daeagm rccycaua ittxy
itr-veopae it-rrvopeaa ittfssue itcon ladee
reP tIn oS eduR anuM
rcycau gaeam iltaoua
liitzaaonu ssce itccoanude itftsssdeuo itrsssopooe itcvoneude
isV )cA R S O R
ryge ivan rkflo ud ta
re fix ce seu itr
rspenu iillynam ilrcgauw igng trcue trcaue rskopa ifttsso tseopo tcuoem liitengm
O M S Iam raF rF W S O O H
angeam csobneo litvaone e
gical management. A summary of the state-of-the-art
in surgical treatment of DFFs, from the diagnosis of
the fracture to the post-operative evaluation of the
patient’s outcome is presented in Fig. 3a.
The investigation of the minimally invasive surgical
management of DFFs highlighted the limitations
related to the current procedure. Four key issues
emerged from the interviews with the orthopaedic
surgeons with a prominent level of consensus: (
surgical site imaging; (
) difficult access through small
) challenging and often inaccurate
reduction of bone fragments followed by disadvantages of
the external fixation; and 4)soft tissue damage due to
the lack of the site visualisation.
Based on the interview results, the requirements that
the RAFS system should address can be summarized
into operational, functional and non-functional. The
hierarchy of requirements is as follows: the operational
requirement at the top, functional requirements at the
component level and Non-Functional Requirements in
the third tier providing a context for the Functional
Requirements. The hierarchy of Functional and
NonFunctional requirements are summarized in Table 3.
Operational requirements A system that will enable
and assist surgeons in the performance of reduction of
intra-articulate joint fractures (IJF) in a minimally
invasive manner within existing clinical practice and
national health system protocols.
The functional requirements (FRx) have also been as
FR1 The system can access the IJF from different
orientations (i.e. different angles)
FR2 The system can attach to IJF fragments
FR3 The system manipulates IJF fragments (i.e.
rotation and translation)
FR4 The surgeon stays in control of the system
FR5 The system enables visualization of IJFs
From interviews some of the system non-functional
requirements have been defined but further ones were
created to comply with safety and certification
procedures for medical devices.
As an added safety criterion, a study to collect force
data in fracture reduction orthopaedic operations9,19
were conducted providing specific thresholds and force
requirements for the system. Specifically, FR2
extended to read:
FR2. The system can attach to IJF fragments under
manipulation forces of 150 N.
while NFR3 extended to read:
The system creates sufficient working space
inside the joint by applying forces of 300 N.
Workflow, Architecture and Sub-system Design
The first step was to revisit the proposed workflow.
From the various imaging and robot navigation
requirements, it was inferred that we need to develop
an image-guided control algorithm. For this type of
activity, it is standard to use optical tracking tools and
our task was to determine clinically acceptable and
technically feasible points of the tool attachments.
Based on the above workflow requirements and
according to the V-model of the design, the general
architecture of the system was defined. Regarding the
hardware architecture, based on the requirements
related with the physical aspects (FR1–FR3) and space
limitations (NFR12–NFR15), a modular approach
was selected over a large monolithic mechanism. The
testing of the entire system was performed on synthetic
bones in laboratory setting and the adopted precision
metric was the positional accuracy of the reduction,
e.g. the normal distance between the fracture lines. In
these verification tests, the entire architecture and
workflow was shown to operate. Specifically, the
system achieved virtual reduction of the fracture with a
maximum residual positioning error of 0.95 ± 0.3 mm
(translational) and 1.4 ± 0.5 (rotational) and
correspondent physical reductions with an accuracy of
1.03 ± 0.2 mm and 1.56 ± 0.1 .11
FR1 For the multi-orientation approach to fracture
fragments, a hybrid geometry for the system has been
designed, in the form of a serial robotic mechanism
called the carrier platform (CP) for gross positioning
and orientation in respect to the patient’s limb and a
hexapod parallel mechanism called the robot fracture
manipulator (RFM) for fine manipulation of the
fragments. The CP consists of two linear and two
Description FR1 FR2 NFR1
revolute joints in a configuration that allows move
around the limb of the patients and the approach from
various angles. The RFM is attached to the CP in this
FR2/NFR1/NFR2/NFR17 In order to allow the
secure attachment of the system to the fragments and to
cause a minimum possible damage to the surrounding
soft-tissue, a new percutaneous fragment manipulation
device (PFMD) that replaces traditional manipulation
pins has been developed to satisfy one of the essential
safety requirements. The PFMD provides the
attachment of the RFMs to the bone fragments via a single
incision less than 10 mm. The PFMD can be anchored
to the bone mono-cortically by using a unique
geometry pin (UGP), an anchoring system (AS), and a
gripping system (GS). The PFMD has been
characterised and its deformation is evaluated showing that
for forces of 150 ± 15 N, the maximum deformations
of the device is 5.8 mm.
FR3 The fragment manipulation is achieved by the
combined operation of the RFM and the Image-based
navigation system. Using the data from the optical tool
the system controls the motion of the RFM8 with
system’s positioning accuracy and repeatability
showing a maximum positioning RMSE of 1.18 ± 1.14 mm
(translations) and 1.85 ± 1.54 (rotations). More
details of the navigation control of the RFM can be
found in Supplement S3.
FR3/NFR3 DFF requires traction of the tibia to
restore the original length and rotation of the joint. In
the current clinical practice, this is performed by
pulling the patient’s foot manually or using a traction
table. This allows the surgeon to apply a constant and
adjustable traction force to facilitate the reduction
process.19 We introduced in the RAFS system a
computer-controlled version of the traction table, i.e. the
automated traction table (ATT).
FR4/NFR5 The system is semi-automated, so that
the surgeon first pre-plans the reduction of the fracture
in the workstation, and then the robotic
system—connected to the fracture—executes the physical
reduction accordingly. Moreover, the surgeon can
adjust the plan intra-operatively based on the progress
of the operation.13 For these to be achieved the host
PC runs the reduction software’s graphical user
interface (GUI) that creates the link between the surgeon
and the robotic system. The GUI allows the surgeon to
interact with CT-generated 3D models of the fracture.
Virtual paths of the 3D fragment models generate
corresponding motion of the robotic system.
FR4/NFR6/NFR7 The GUI provides the surgeon
with both 2D views of each anatomical plane (i.e.
sagittal, frontal, transverse) and a 3D view of
CTgenerated fracture models. The user controller chosen
for this application is the Leap Motion, which is able
to track and synthetize a 3D position and orientation
(6DoF) of the hands in its workspace. Also, three foot
pedals that provide on–off inputs to the system are
) to grab and release the fragment models,
) to select a specific anatomical plane for interaction,
) to merge two fragments together that are
further manipulated as one fragment.11
FR5/NFR8/NFR9 A pre-operative CT scan of the
fracture is taken, and the resulting dataset segmented
to generate 3D models (CAD model) of each bone
fragment. The models are imported and displayed in
the reduction software so that the surgeon can interact
with them using the GUI as described above.11
FR5/NFR10 The surgeon virtually reduces the
fracture using the GUI by manipulating and matching
the broken fragment to move them to the original
unbroken position. This generates the desired final
poses for each fragment. Pre-operative planning data
are stored in the system and used for intra-operative
robot motion calculations to achieve the physical
reduction of the fracture.13
FR5/NFR11 The system is equipped with an optical
tracking system (Polaris Spectra, NDI Inc., tracking
accuracy 0.25 mm) which provides intra-operative
real-time update of the 3D models through the optical
tools attached to the orthopaedic pins inserted into the
NFR4/NFR12–NFR15 These non-functional
requirements are related to the geometry of the system
and the way it integrates with the staff and the existing
equipment in operating theatres (OR) and current
practice in orthopedic surgery. To this end, the overall
geometry and physical footprint of the system were
considered which inspired the modular structure of the
system shown Fig. 4. The different components of the
system are rigidly connected, i.e. the CPs and the ATT
are secured on a portable rigid wheeled frame which
can be easily moved and replaced by a OR trolley.
NFR16 To ensure the safety of the system, the latest
regulations and certifications were followed in the
design and development of all sub-systems. Table 4
summarises the different standards used. Special
attention was given to activities that emulate a quality
management system (QMS) leading to conformity to
ISO13485. To this end, we focused on the design and
development inputs, verification and validation, and
used relevant standards as inputs to the process.
Moreover, the validation test (cadaveric study) was
documented according to ISO13485 regarding the
acceptance criteria and statistical techniques used.
Based on the operational and safety requirements,
the most suitable validation test was the use of human
cadavers (trials approved by the National Research
Ethics Committee, REC Reference: 15/WM/0038,
UK). The specimens used were right and left lower
limbs from male (n = 4) and female (n = 3) cadavers
with no bone defects on which the desired fractures
were created. Specimens were collected from the
proximal femur to the end of the foot. For the creation
of appropriate fracture shapes (T and Y, 33-C13) in a
tr s it
a n n
p it e
tn ic n
e e a
m p s
e s le
ir r p
u e ic
re lla r
c a l
ifi ic a
c s r s
e y e e
p h n s
s p e o
— d : r
k a t1 up
o d ra ry
tw ho p to
e t — la
na em lts ug
raen ssce ise s r
a c v d f
ilt a de an tsn
o ) l
p DC ica sd e
e A de t
m M m e u
n (C to tm qe
l n n e —
a io e m s
lo c m re m
te ge su tse
— e a a y
y d n e s
loo ino am fm tn
n is k o e
h l s ) m
c o ir n e
te c f io g
h o s a
n it i n
o n c a
ta sw ito re m
m ty r
lo fe fo
e r itco se
FRA iitrpo ftrohe tsenm :tr1po rcpoe ine nad
teh sce ts ire –4 le icd n
n u 4 cy e io
e qe tr c m s
predictable and reproducible manner, an accepted
technique of osteotomy was used. From the validation
testing it has been shown that the system performs
within the required operational requirements and
achieves reductions of 1 mm and 5 .13
Final Interview Study
The final interview study identified three key
findings related to the process described above. Firstly, the
clinical workflow presented received an average score
of 3.8 out of 6, where 0 indicates ‘‘Not at all
acceptable’’ and 6 ‘‘Highly acceptable’’. With the manual
actions of the surgeon, i.e. the pre-operative Virtual
Reduction and the intra-operative actions of Robot–
Patient connection and Insertion of orthopaedic pins,
scoring 2.5 out of 6, where 0 indicates ‘‘not at all
challenging’’ and 6 ‘‘Highly Challenging’’. Secondly,
regarding the optimal representation of the fracture, 17
out of 18 participants preferred a combination of 2D
and 3D views (the outlier preferred 3D views only).
Finally, regarding the physical dimensions of the
system, 8 out of 18 preferred the current size, 8 out of 18
preferred a smaller size, and 2 out of 18 a larger size.
The requirement elicitation study provided critical
insights into the difficulties and issues related with the
DFF reduction. One of the most notable problems was
the limited visualisation provided by the available
intra-operative imaging technologies for the adoption of
minimally invasive management of fractures.
Moreover, the typical radiological assessment of the
fracture, either with plain X-rays (pre- and
Intraoperatively) or with CT scanning (pre-operatively)
does not provide any information about the soft tissue
damage and location. Assessment of the reduction
accuracy is not in the regular practice and
misalignments are often detected when follow-up morbidities
occur, e.g. arthritis. Also the mind-to-hand
coordination of the surgeon, due to poor visualisation renders
minimally invasive procedures challenging. This
prompted the development of 3D real-time image
guidance for RAFS.
The second key issue that emerged from the
interviews was related to the congested nature of the
operation, i.e. multiple anatomical structures in a
cluttered environment. This limitation is contributed
primarily to the neurovascular structures and major
ligament structures, especially in posterior condyle
cases. The soft tissue poses a challenge in the fracture
reduction both intra-operatively and post-operatively.
In the first case, soft tissue can affect the quality of
fracture reduction, and disrupt the correct anatomical
position of the ligaments generating tension in the
fragments, or tissue swelling. At the same time the soft
tissue damage due to the operation, must be kept to a
minimum to avoid tissue scarring and fibrosis affecting
negatively the healing process. The space constraints
and soft tissue constraints make not only the reduction
process difficult but also keeping the fragments in place
before and during fixation. Moreover, the correct
fixation implant selection and positioning proves difficult
both due to space constrains and pre-operative
visualisation, affecting the correct anatomical restoration
of the articular surface leading to post-operative
Finally, the reduction can be impaired by the bone
quality, e.g. by osteoporotic bones; both in terms of
reduction and fixation. The ‘softer’ bones are prone to
breaking and are more difficult to grasp and
Based on these discussions, the first point that this
investigation had to tackle was the workflow of the
proposed intervention, specific with regards to
imageguidance. In current practice there are no provisions
for imaging and navigation and a new workflow was
proposed where pins would be placed prior to initial
CT imaging.11 On a second iteration of assessing the
system it was found that the proposed workflow could
potentially violate other requirements (e.g. NFR17)
and an alternative workflow was proposed along with
a technical requirement, i.e. the use of image
registration prior to operation and using CT-scan data and
fiducials in the theatre. The revised workflow can be
seen in Fig. 3b with details of its implementation
presented in Ref. 14. The workflow assessment in the final
interview study was judged as acceptable by the
The second point this investigation has achieved is
the architecture that is fit for purpose and adaptable to
the wide spectre of requirements and constrains. The
three physical sub-systems were identified to be the
Robot Fracture manipulator (RFM), the carrier
platform (CP), and the automated traction table (ATT).
For the software, sub-systems of the functional entities
were identified as graphical user interface (GUI),
imaging and registration (IR), navigation and high
level-control (NHLC), and low-level control. The first
two are implemented on a workstation and the latter
two on a dedicated embedded controller.
During the design and development of the system,
each requirement has been analysed and the final
subsystems were aligned to satisfy all of them. The main
focus was on functional and non-functional
requirements with each subsystem tackling a number of
The CP is tackling requirements related to the wide
work envelope of the system. For FR1, the two linear
joints allow motion along the axis and radially around
the limb, while the revolute joints allow for rotation in
the perimeter of the limb and at the angles oblique to
the axis of the limb.12 Also for NFR4/NFR12 the CPs
are of such a size that allow the approach from
different angles while at the same time will (
) allow space
for the placement of an image-intensifier while the
system is attached to the fragments Fig. 4, and (
leave most of the surgical field free for the surgeon to
manually fixate the fracture. The size of the system was
also addressed in the final interview study and the
clinicians were split between the current and a smaller
size indicating that further investigations are needed.
The detailed operation and axis of motion of the CPs
and its kinematic analysis are reported in Ref. 10 and
in Supplement S1. The operation of the CP is tested
and verified against the set criteria.
For dealing with the key manipulation requirement,
FR3, the RFM has been proposed. Also the compact
nature of the RFM can tackle NFR15 to allow access
to the surgical field. The RFM is an automated
computer-controlled parallel-robot8 with 6
degrees-offreedom (DOF), i.e. three translations and three
rotations along/around x, y, z axes. The use of a
parallelrobot is a preferred choice for orthopaedic applications
where high load carrying capacity and precise
positioning accuracy-repeatability are of paramount
importance. The parallel-robot has been designed and
manufactured in-house ad hoc with the desired
characteristics. The struts have been developed as linear
actuators based on a ball screws and a brushed DC
motor with integrated gearbox and rotational encoder
(RE10–MR–GP10K, Maxon Motor) providing
hightorque, precise positioning (0.485 lm resolution). The
6 linear actuators produce a resulting load capacity of
360 N (force) and 12 Nm (torque) in the testing and
verification process reported in Ref. 8.
For providing traction (FR3/NFR3) the ATT is
proposed. The ATT is a 4-DOF mechanism (two
prismatic and two revolute joints) Fig. 4, connected to
the tibia through an orthopedic boot and a leg holder.
The ATT allows for precise traction that will create
space for the performance of reduction maneuvers.
Details of the use, and testing and verification of the
ATT can be found in Ref. 13 and in Supplement S1.
Addressing the issues related with anchoring the
system on the bone a new PFMD was designed and
tested composed of three elements the UGP, the GS,
and the AS. The UGP is a custom-designed
non-cannulated 6 mm diameter orthopaedic manipulation pin
with 4 distinctive cross-sections. These sections allow
for the different functionalities, i.e. connection to
RFM via the GS, attachment of an optical tool for
navigation purposes,10,11 attachment to the AS, and a
threaded metric M6 section that is screwed into a single
cortical plane of the fragment exhibiting good pull-out
characteristics. The AS (Fig. 2b) is a custom designed
system that firmly embeds the UGP into the bone
fragment using a drilling template (DT) to hold four
stainless steel nails that the surgeon drills into the bone
fragment. More details about the testing and
verification of the PFMD can be found in Supplement S2.
The Navigation and Tracking System is based on the
Polaris optical tracker. The tracking device is using
optical tools that are being placed on crucial parts of
the system, namely the fragments, the RFM, and the
tibia in the case of DFF. To enable intra-operative
image guidance, the relative position of each pin with
respect to the bone fragment in which it is inserted is
calculated through intra-operative surgical
registration.14 Once the relative pose of each pin bone is
known, and assuming that it does not change over time
(i.e. the object constituted by the pin and the bone
fragment is considered rigid), the pose of each bone
fragment is updated in real-time using the optical
tracker by connecting an optical tool to the pin.13 This
depicts the actual pose of each fragment in the 3D
space during the surgery. Intra-operative imaging
allows surgeon to monitor progress of the physical
fracture reduction performed by the robotic system.
More information about the testing of the navigation
and tracking system can be found in Refs. 13,14 and
The tracking information is used for the Control of
the system and Fig. 5 shows the control architecture of
the system, with the surgeon in control of the robotic
device and planning the surgical procedure from a
workstation. The system employs a host–target
structure composed by a PC (host) and a real-time
controller with FPGA (target), and a low-level motor
controller. The target controller (compactRIO-9068,
National Instruments) processes the surgeon’s virtual
reduction, and generates motion commands which are
sent to the low-level motor controller (EPOS2 24/3,
Maxon Motor) that executes the movement of the
robotic system to achieve the physical reduction of the
fracture.11 More details about the testing of the control
scheme can be found in Supplement S3.
The interaction with the user is ensured via the
specially design GUI in the workstation of the system.
The 2D views (projections) of the fracture allow the
surgeon to perform the virtual reduction. The 3D view
allows the surgeon to move the camera around the
model in the virtual environment to assess the outcome
of the reduction. The use of 2D and 3D views was
favoured by the clinicians as indicated in the final
interview study. The surgeon intuitively interacts with
the 3D models using their hands through a user
controller to virtually reduce the fracture in the virtual
environment. This way the requirements for
pre-operative planning but also intra-operative control of the
process, i.e. under sterile conditions can be achieved.
This paper presented a user-centred approach for
the design and development of a novel medical device.
The interviews with the surgeon at an early stage of the
medical device development allowed the research team
to capture the needs and current issues of the clinical
practice. Following a design and development
approach based on established methods like the
Vmodel of design the final system has been built and
tested to perform within the requirements. The final
results demonstrated that appropriate design methods
allow the development of a complex system within time
frames and constrains to achieve its goals. Future
works include the formulation of a design and
development approach which can be applicable to other
healthcare systems requiring the input from the users.
ELECTRONIC SUPPLEMENTARY MATERIAL
The online version of this article (https://doi.org/
10.1007/s10439-018-2005-y) contains supplementary
material, which is available to authorized users.
This work is a summary of independent research
funded by the National Institute for Health Research
(NIHR)’s Invention for Innovation (i4i) Programme
(II-SB-0712-20002). The views expressed are those of
the authors and not necessarily those of the NHS, the
NIHR or the Department of Health.
CONFLICT OF INTEREST
The authors declare that they have no conflict of
interest. S. Dogramadzi, G. Dagnino, I. Georgilas
have a patent GB1513436.4 pending.
This article is distributed under the terms of the
Creative Commons Attribution 4.0 International
which permits unrestricted use, distribution, and
reproduction in any medium, provided you give
appropriate credit to the original author(s) and the source,
provide a link to the Creative Commons license, and
indicate if changes were made.
1Alexander, K. , and P. J. Clarkson . Good design practice for medical devices and equipment, Part I: a review of current literature . J. Med . Eng. Technol. 24 : 5 - 13 , 2000 .
2Alexander, K. , and P. J. Clarkson . Good design practice for medical devices and equipment, Part II: design for validation . J. Med . Eng. Technol. 24 : 53 - 62 , 2000 .
3AO Foundation. Distal Femur Fracture-Reduction and Fixation , 2015 .
4Blandford , A. Semi-structured qualitative studies . In: The Encyclopedia of Human-Computer Interaction , Vol. 2 , edited by M. Soegaard , and R. F. Dam . Aarhus: The Interaction Design Foundation , 2013 , p. 53 .
5Boakes , E. , B. Norris , and S. Scobie . The role of design and human factors in medical device patient safety incidents . In: Proceedings of Improving Patient Safety conference ''From safe design to safe practice'' , 2008 .
6Bridgelal Ram, M. , P. R. Grocott , and H. Weir . Issues and challenges of involving users in medical device development . Heal. Expect . 11 : 63 - 71 , 2008 .
7Buckle, P. , et al. Patient safety, systems design and ergonomics . Appl. Ergon . 37 : 491 - 500 , 2006 .
8Dagnino, G. et al. Design and real-time control of a robotic system for fracture manipulation . In: 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) , pp. 1 - 4 , 2015 .
9Dagnino, G. et al. Force-torque measurement system for fracture surgery . In: International Society for Computer-Assisted Orthopaedic Surgery ConferenceCAOS, 2015 .
10Dagnino, G. et al. Image-based robotic system for enhanced minimally invasive intra-articular fracture surgeries . In: Proceedings of the 2016 IEEE International Conference on Robotics and Automation (ICRA) , pp. 696 - 701 ( 2016 ).
11Dagnino, G. , et al. Navigation system for robot-assisted intra-articular lower-limb fracture surgery . Int. J. Comput. Assist. Radiol. Surg . 11 : 1831 - 1843 , 2016 .
12Dagnino, G. , et al. Vision-based real-time position control of a semi-automated system for robot-assisted joint fracture surgery . Int. J. Comput. Assist. Radiol. Surg . 11 : 437 - 455 , 2016 .
13Dagnino, G. , et al. Image-guided surgical robotic system for percutaneous reduction of joint Fractures . Ann. Biomed. Eng. 2017 . https://doi.org/10.1007/s10439-017- 1901-x.
14Dagnino, G. , et al. Intra-operative fiducial-based CT/fluoroscope image registration framework for image-guided robot-assisted joint fracture surgery . Int. J. Comput. Assist. Radiol. Surg . 12 : 1383 - 1397 , 2017 .
15DeFranco, J. , et al. The nonfunctional requirement focus in medical device software: a systematic mapping study and taxonomy . Innov. Syst. Softw. Eng . 13 : 81 - 100 , 2017 .
16Dick, J. , E. Hull , and K. Jackson . Requirements Engineering . Heidelberg: Springer, 2017 .
17Eppinger, S. , and K. Ulrich . Product Design and Development . New York: McGraw-Hill Higher Education , 2015 .
18Forsberg, K. , and H. Mooz . The relationship of system engineering to the project cycle . INCOSE Int. Symp . 1 : 57 - 65 , 1991 .
19Georgilas , I. et al. Preliminary analysis of force-torque measurements for robot-assisted fracture surgery . In: 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) , 2015 .
20Graham, M. J. , et al. Heuristic evaluation of infusion pumps: implications for patient safety in Intensive Care Units . Int. J. Med . Inform. 73 : 771 - 779 , 2004 .
21Hagedorn, T. J. , I. R. Grosse , and S. Krishnamurty . A concept ideation framework for medical device design . J. Biomed. Inf . 55 : 218 - 230 , 2015 .
22Hagedorn, T. J., S. Krishnamurty , and I. R. Grosse . An information model to support user-centered design of medical devices . J. Biomed. Inf . 62 : 181 - 194 , 2016 .
23Hyman, W. A. Errors in the use of medical equipment . In: Human Error in Medicine, edited by M. S. Bogner . Hillsdale: Lawrence Erlbaum Assoc, 1994 , pp. 327 - 347 .
24ISO 9241 - 210 : 2010 . Ergonomics of Human System Interaction-Part 210: Human-Centred Design for Interactive Systems . Switzerland: Int. Stand. Organ. (ISO) , 2009 .
25Jackson , A. E. , et al. Dual robot system for upper limb rehabilitation after stroke: the design process . Proc. Inst. Mech. Eng. C 221 : 845 - 857 , 2007 .
26Korotchenko , A. B. H. K. , and P. E. A. Mihailidis . Power mobility with collision avoidance for older adults: user, caregiver, and prescriber perspectives . J. Rehabil. Res. Dev . 50 : 1287 , 2013 .
27Lockton , D. , D. Harrison , and N. A. Stanton . The design with intent method: a design tool for influencing user behaviour . Appl. Ergon . 41 : 382 - 392 , 2010 .
28Martin, J. L. , and J. Barnett . Integrating the results of user research into medical device development: insights from a case study . BMC Med . Inf. Decis. Mak. 12 : 74 , 2012 .
29Martin, J. L. , et al. Medical device development: the challenge for ergonomics . Appl. Ergon . 39 : 271 - 283 , 2008 .
30Martin, J. L. , et al. A user-centred approach to requirements elicitation in medical device development: a case study from an industry perspective . Appl. Ergon . 43 : 184 - 190 , 2012 .
31McHugh, M. et al. An agile v-model for medical device software development to overcome the challenges with plan-driven software development lifecycles . In: 2013 5th International Workshop on Software Engineering in Health Care (SEHC) , pp. 12 - 19 , 2013 .
32Privitera, M. B. , M. Evans , and D. Southee . Human factors in the design of medical devices-approaches to meeting international standards in the European Union and USA . Appl. Ergon. 59 : 251 - 263 , 2017 .
33Raabe , D. , S. Dogramadzi , and R. Atkins . Semi-automatic percutaneous reduction of intra-articular joint fractures-an initial analysis . In: 2012 IEEE International Conference on Robotics and Automation (ICRA) , pp. 2679 - 2684 ( 2012 ). https://doi.org/10.1109/icra. 2012 . 6225076 .
34U.S. Food and Drug Administration . Design control guidance for medical device manufacturers . Cent. Devices Radiol . Heal., 1997 .
35Zhang, J. , et al. Using usability heuristics to evaluate patient safety of medical devices . J. Biomed. Inf . 36 : 23 - 30 , 2003 .