Touch-sensitive robot aids tumour detection in minimally invasive
28 August 2009
Canadian researchers have created a touch-sensitive robot that
detects tougher tumour tissue in half the time, and with 40% more
accuracy than a human. The technique also minimises tissue damage.
Malignant tissue is usually stiffer than the surrounding tissue.
Oncologists use scanning techniques such as magnetic resonance imaging
(MRI) and computed tomography (CT) to scan pre-operatively to identify
lesions. But tissues may shift during surgery, making it hard to rely on
the position identified by the scan.
So instead surgeons use gentle pressure (palpation) to confirm where
the tumour is, or to locate further tumours not picked up through
scanning. But in using minimally invasive surgical (MIS) techniques this
can be very tricky due to access difficulties — as the surgeon must
attempt to feel for harder tissue using long, slim instruments via a
very small incision.
An alternative is to relay touch (haptic) cues via an instrument.
Haptic cues include kinaesthetic information, relating to movement,
which helps determine the shape and stiffness of an object, and tactile
cues about surface textures. A variety of handheld sensors and grasping
instruments have been developed since the mid 1990s for use in surgery,
but these have the drawback that they do not in themselves control the
amount of pressure used, nor do they position themselves correctly. Many
are also too large for use in MIS.
Enter the robot-controlled palpating device: With cows' livers
standing in for human tissue and 10mm and 5mm blobs of glue wrapped in
wire representing tumours, the researchers compared palpation by
surgeons, non-surgeons and the robot in the blinded trials. The
researchers used a torque sensor to measure the force of the palpations.
Using tactile MIS sensing instruments under robotic control reduces
the maximum force applied to the tissue by over 35% compared to a human
controlling the same instrument. Accuracy in detecting the tumours was
also far greater with the robot — between 59 and 90% depending on the
robot control method used for palpation.
Unlike humans, the robot applies consistent force in each step, and
moves over the tissue systematically. This produces a complete map,
equivalent to one large pad applying ideal levels of force to the whole
sample. (Similar to tactile sensors that have been developed to detect
Humans do not know from one palpation to the next exactly how much
force they are applying. This means some features are only highlighted
because the surgeon is applying more force, or because the human user
has changed the angle slightly between the instrument and the tissue. It
is also easier to miss a tumour due to applying slightly lower force.
In fact both surgeons and non-surgeons were more likely to cause
tissue damage than the robot. When a subject observed increased pressure
on the visual display, they tended to focus on the area and apply even
more force to see if what they had observed was a tumour. In the case of
MIS, only a very small area can be palpated, which makes it challenging
to compare adjacent areas and search for a tumour manually.
If developed further, the authors suggest that this type of
instrument would particularly benefit surgeons performing lung tumour
resection, where tissue often shifts significantly.
To develop the prototype robot for use in real MIS, the researchers
plan to incorporate a design upgrade to include a flexible rotating head
and a remote centre of motion. They would also add an improved interface
to help surgeons overcome any fears about using robots in this type of
surgery, and to allow them to increase the number of palpations around a
This means using robots during MIS to detect tumours is "not only
feasible, but results in reduced tissue trauma and increased tumour
detection," according to lead author Analuisa Trejos.
1. A.L. Trejos, J. Jayender, M.T. Perri, M.D. Naish, R.V. Patel and R.A.
Malthaner. Robot-assisted Tactile Sensing for Minimally Invasive Tumor
Localization. International Journal of Robotics Research. 2009; 28; 1118 originally published online May 19, 2009;
Bookmark this page