Mechatronic Resectoscope Emulator for a Surgery Simulation Training

System of the Prostate

M.A. Padilla, F. Altamirano, F. Arambula & J. Marquez

Abstract— In this work is presented the development ofa mechatronic interface for a surgery simulation system fortraining Transurethral Resection of the Prostate (TURP). Theinterface emulates a real resectosope and allows to performthe most important movements of the surgical tool during aTURP. The interface is able to work in conjuntion with avirtual reality software with a deformable tissue model of theprostate, in order to simulate tissue resection and deformation.The current prototype has five degrees of freedom, which areenough to have a realistic simulation of the surgery movements.The results show that the interface is suitable for a real timesurgery simulation training system of the prostate without forcefeedback.

I. INTRODUCTION

The prostate gland is located next to the bladder in

human males, with the urethra running from the bladder

neck through the prostate to the penile urethra (Fig. 1). A

frequent condition in men above 50 years old is the benign

enlargement of the prostate known as Benign Prostatic Hy-

perplasia (BHP), which in some cases results in significant

blockage of the urinary flow. The standard surgical procedure

to treat a hypertrophied prostate gland is the Transurethral

Resection of the Prostate (TURP). It essentially consists of

the removal of the inner lobes of the prostate in order to

relieve urinary outflow obstruction. TURP is one of the most

performed surgical procedures in urology, however mastering

the TURP technique requires a highly developed hand-eye

coordination which enables the surgeon to orientate inside

the prostate, using only the monocular view of the lens of

the resectoscope. Currently TURP is taught through example

from an experienced surgeon, but training introduces risk

of severe haemorrhage for the patients. As a consequence,

the resident of urology has very restricted opportunities to

practice the procedure, but must perform hundreds of training

sessions during a period of four years before acquiring the

skills needed.

An alternative consists of the development of computer

simulation systems for surgery training that bring the urology

residents the opportunity to practice an unlimited of times,

with less risk for the patients, and in an economically

suitable manner, before practicing with real patients. More-

over, simulation of several physiological phenomena, like

bleeding, coagulation, and charring, among others, could be

incorporated in order to enhance the realism of the training

process.

M.A. Padilla, F. Altamirano, F. Arambula & J.Marquez are with ImageAnalysis and Visualization Lab, CCADET, Universidad Nacional Autonomade Mexico, P.O.Box 70-186, Mexico, D.F., 04510 {miguel.padilla,fernando.arambula}@ccadet.unam.mx

Fig. 1. Prostate gland position.

Gomes et al. [3] reported an interesting hybrid computer-

assisted training system for TURP, that use a resectable

physical phantom and a computer model of the prostate.

The aim of using traditional physical phantoms is to pro-

vide the surgeon with more realistic haptic feedback than

virtual reality techniques can provide. Positional feedback is

provided by an optical tracker and the 3D computer model

of the prostate. However, the 3D model used is a rigid

geometric model of the prostate capsule, without deformation

and resection simulation. Additionally, a phantom based

simulator requires continuous expense on replacements.

Manyak et al. [4] reported the construction of a virtual

reality surgery simulation system of the lower urinary track.

They considered only the surface of the urinary track, re-

constructed from the Visible Human Project dataset [5] with

texture mapping for visual realism. However, the prostatic

urethra behaviour depends on the conditions of the tissue

from the capsule to the urethra. As a consequence, a vol-

umetric model of the prostate should be consider in order

to simulating realistic TURP procedures. Sweet, et al. [6]

reported the experience of using the TURP virtual training

system described in [7]. They studied the effectiveness of

translating the skills acquired in their virtual environment to

the operating room. The surgery simulation system for TURP

reported in [7] uses an image-based approach for simulating

bleeding when the resecting loop contacts surface vessels.

The loop triggers precalculated movies of blood flow, that

is then oriented and mapped on the virtual environment.

However, how they model tissue deformation is not clear;

moreover as in [4] they used only the surface representation

of the urethra.

In [1] and [8] we reported the development a 3D de-

Proceedings of the 29th Annual InternationalConference of the IEEE EMBSCité Internationale, Lyon, FranceAugust 23-26, 2007.

FrA09.5

1-4244-0788-5/07/$20.00 ©2007 IEEE 1750

formable volumetric model of the prostate for TURP simula-

tion that involves tissue deformation and resection, consider-

ing the gland as a viscoelastic solid. In this work we describe

the development of the virtual resectoscope interface for our

simulator. Section 2 of this paper describes the development

of the mechatronic interface that emulates the resectoscope;

section 3 briefly describes the interaction scheme between the

virtual resectoscope and the tissue model; finally in section

5 we present the conclusions and future perspectives of this

work.

II. MECHATRONIC DESIGN

In order to obtain a realistic simulation of the most

important movements of the surgeon during a TURP, a

mechanism was designed based on a disk-ring array (Fig.

2). Due to difficulties in repreducing the seven degrees of

freedom, we decided at the moment to reproduce only the

five most important degrees (Fig. 2.a). In this manner the

mechanism provides five axes of movement. Three of these

axes are rotational and the other two are linear displacements

of the resectoscope. We ignore for the moment two additional

translation degrees of the resectoscope sheath. However, in

opinion of an expert urologist, the five movements repro-

duced are enough for having a realistic reproduction of the

real movements during cistoscopies and TURP procedures.

The disk-ring array is mounted in a plastic box. Inside

the box is placed and registered a phantom of the prostate

constructed from the 3D model of the gland [8], this phantom

is the physical reference of the 3D computer model and also

allows to limit the resecting volume during the simulation.

The correct localization of the phantom is important for the

interface, since it does not provide for the moment reacting

forces, and as a consecquence, the trainee could easily miss

the spatial orientation inside the prostate at the early stages

of the training process. Registration of the phantom is made

by calculating the orientation and position with respect to

the reference system of the interface, by doing the principal

component analysis of the phantom shape, where every shape

point is sampled with the head of the resectoscope.

Fig. 2. Resectoscope degrees of freedom. a) Linear (a and b) and rotationalmovements (c, d and e). b) Digital encoders for rotational axes.

A. Axes movement sensors

1) Rotational axes : Optical encoders from USDigital

(HEDS-9040 and HEDS-9140 models) [10] are used in order

to sense each of the three rotational movements. Each en-

coder is placed in each rotation axe, so with this arrangement

we can measure the direction and the angle rotated by the

user on each axis (Fig. 2.b).

2) Linear axes : The linear displacement (in/out) of

the surgical tool is measured with a linear precision multi-

turn potentiometer. The shaft of the potentiometer has a

metallic disk with a rubber band which is in contact with

the cylindrical body of the resectoscope. The output voltage

of the potentiometer varies according to the position of the

resectoscope from 0 to 23 cm, which corresponds to the

useful resectoscope displacement range. The resecting loop

has a linear movement of 36 mm, this distance is sensed with

an array of two linear Hall effect sensors and two permanent

magnets, as shown in Fig. 3.

Fig. 3. Hall effect sensors array for controling the resecting loop.

B. Sensor Data Processing

For monitoring we used, as described in next section a set

of microcontroller for real time sensing. Since translational

sensors produced analog signals, is easy and fast to use one

microcontoller for sensing and processing all translations:

we used an LP3500 card [11], which is a low-power, single-

board computer with a Rabbit 3000 8 bit microcontroller at

7,4 MHz and programmed in C language with the Dynamic

C compiler [2]. On the other hand, since the rotation sensing

produces digital signals, the complexity of processing it is

higher, so we used a set of PIC microcontrollers of Microchip

Technology [12] for monitoring in parallel rotational move-

ments.

1) Optical Sensors: Each optical sensor give us three

output signals: CH. A, CH. B and CH. I (as shown in Fig. 4).

We ignored signal CH. I, since this signal is useful only when

the optical sensor gives a complete turn, however our three

sensors will never turn completely. To determine the position

and direction of rotation, channels A and B are enough.

For each rotational axe a program was developed on a PIC

microcontroller for calculating from the two out of phase

signals, the angular position of each rotational axis. The

program is based on storing the previous value given by

the sensor to the microcontroller and comparing it with the

present value, with the aim to detect any change in the phase

between CH. A and CH. B. The angular distance of rotation

is determined by counting the number of pulses of CH. A.

2) Multi-Turn Potentiometer : The acquisition of the po-

tentiometer output signal is made through one of the analog

channels of the LP3500 microcontroller. The configuration of

the potentiometer is a voltage divider whose output measures

the resecting loop movements.

The signal is in the interval from 3.23 to 4.50V that

corresponds to the movement of the resectoscope from 0 to

1751

Fig. 4. Signals of each digital encoder for rotational movements.

23cm. The mathematical model of this sensor corresponds to

a linear device. For this reason, efficient linear interpolation

of the linear device model was directly programmed as a

routine of the LP3500.

3) Hall Effect Sensors : The output signal of the Hall

effect sensors array is measured through two analog channels

of the LP3500, the program in C is in charge of signal acqui-

sition, addition and final displacement calculation, through

interpolation using a lookup table. The mathematical model

obtained is nonlinear and at off-line stage, a lookup table

with 148 interpolated values that corresponds to a resolution

of 0.5mm of the non-linear curve were calculated, in order

to obtain in real-time the displacement of the resecting loop.

Finner displacements behind 0.5mm are calculated in real-

time by doing linear interpolation of the precalculated values.

III. INTERACTION WITH THE VIRTUAL MODEL

As we mention in the last section, monitoring of the

resectoscope movements are performed with an embebed

electronic system consisting on five microcontrollers: four

PIC microcontrollers for digital signal monitoring of the

optical sensors, for rotations; the LP3500 card for the analog

signals of the multi-turn linear potentiometer and hall-effect

sensor, for translations.

The five microcontrollers in the embebed system run

together in parallel in order to monitoring the movements

in real time; the system sends the movements information

in the form of moving commands to the virtual model. The

real-time resectoscope movements consequently reflect the

interaction between the surgeon and the tissue model. The

interactions (tissue deformation and resection) between the

virtual tool and the prostate model are consequence of the

collision between them (See section III.B).

A. Resectoscope movements monitoring

Sensors information is send to the model in 3D continu-

ously in asynchronous way through serial port at 57600 bits

per second (BPS). The information flow in the electronic

system is as follows (Fig. 5): The card LP3500 makes

a request of digital data to the master PIC, in parallel

while the data of the digital sensors are received in the

reading bus of a serial port C, the signals of the analogical

sensors are acquired with the DAC of the LP3500 card

and stored in an shipment data array. Data request done

by the LP3500 is received by the master PIC, which asks

for the information of each of the digital sensors to the

corresponding slave PIC, while the slave PIC continuously

calculate in parallerl the rotational axes position. Therefore

when they recived the data request the shipment is made

immediately with the last sensed data. These requests and

data reception are made by the PIC serial port. After the

master PIC receives the complete information of the digital

sensors, sends it to the LP3500 by its serial port C, which

stores the arriving information in the shipment data array.

When the incoming array is full the LP3500 card sends byte

by bye the information of the resectoscope position to the

3D model through its B serial port.

Fig. 5. Microcontrollers embebed system.

B. Virtual model interaction

We modeled the virtual resectoscope as a 3D surface

mesh. The prostate model were inplemented as a deformable

mass-spring system. We developed a collision detection

mechanism for our simulator based on the representation of

triangular meshes with hierarchical sphere-trees. Our imple-

mentation detects all collisions between the resectoscope and

the prostate in real-time and allows the hierarchical structure

updating for mesh modification after tissue cuttings. After

a collision is detected the soft tissue must slightly deform

before the tissue resection occurs. Tissue deformations result

from the reacting forces due to collisions, where reacting

1752

forces are the forces needed to separate the colliding objects

and depends on the penetration depth of the resectoscope

inside the prostate, and the mechanical properties of the soft

body (prostate). For the moment, tissue resection is done

without local mesh refinement, so tissue cutting is performed

as two operations: removing triangles from surface mesh and

adding the inner triangles exposed by the cut. More details

of the prostate model and the collision detection mechanism

are available in [8] and [9] respectively.

IV. CONCLUSION AND FUTURE WORK

In this paper we have presented the development of a

virtual resectoscope interface for a surgery simulation system

of the prostate, without force feedback for the moment. In

Fig. 6 a view of the mechatronic device and the virtual model

could be seen.

Fig. 6. View of the mechatronic device and the virtual model

We performed metrological tests by using a ZEISS MC850

Coordinates Measuring Machine with an accuracy of ±3 µm,

with 96% of confidence [13], and by comparing the data

measured by our interface and the ones measured by the

coordinate machine, for the same positions of the resecto-

scope head. After doing metrological evaluation we obtained

accurate specifications of the mechatronic interface. The tests

made reveals that as the resectoscope is moving inside the

virtual urethra, starting from the mechanical reference point,

the displacement error increase until reaching an acceptable

statistical error of 2.14 mm when the resecting head is

located around 7 cm from the reference point. This is an

acceptable error rate for simulation, but more effort must

be made for stablishing the real approximate position of

the prostate inside a patient with respect of an anatomic

reference, in order to setup the virtual prostate position

correctly in the simulator. For greater distances, the error

increase drastically until reaching 7mm approximately, so

mechanical adjustments must be made in order to reduce

errors of movements when the resecting head is located

farther from origin. The response time of the movements

monitoring is 44 ms (23 Hz approximately). Surgery simula-

tions systems must run at least at a response rate of 20 Hz in

order to be visually realistic, at least as a low cinema movie.

If we consider that the mininimal correct response rate of the

full system must include collision, deformation, cutting and

rendering, which are highly demanding computing tasks that

together could hardly run at 28-30 Hz, the electronic system

must be enhanced in order to monitoring movements with at

more acceptable rate from 60 to 100 Hz. The mechatronic

prototype has been evaluated by an urology specialists and

in his opinion visual feedback is more important than haptic

feedback and does not seem mandatory for TURP simulation,

however we are also planning in the near future to include

force feedback to the mechatronic device. We are also

planning to evaluates the usefulness of the system with a

group of residents in urology, as soon as the full virtual

simulation system will be finished.

V. ACKNOWLEDGMENTS

The authors are grateful to Dr. Sergio Duran, Urology

specialist from Instituto Nacional de Rehabilitacion, for their

useful comments and insight provided on the mechatronic

interface and TURP.

REFERENCES

[1] Arambula-Cosio F., Padilla-Castaneda M., Sevilla-Martinez P., Com-puter assisted prostate surgery training, International Journal of Hu-

manoid Robotics, Vol. 3, No. 4, pp. 485498[2] Fox (LP3500) C-Programmable Single-Board Computer, Users Man-

ual (2003), Zworld Inc, California USA.[3] M.P.S.F. G., A.R.W. B., A.G. T., B.L. D., A computer assisted

training/monitoring system for turp structure and design, IEEE Trans-

actions on Information Technology in Biomedicine 3, 4, 1999, pp.242-250

[4] MJ M., K S., J H., R K., Carleton T H. X., RJ W., Virtual realitysurgical simulation for lower urinary tract endoscopy and procedures,Journal of Endourology 16, 3, 2002, pp. 185-190

[5] of Medicine T. N. L., The Visible Human Project, National Institutesof Health

[6] Sweet R., Kowalewski T., Oppenheimer P., Weghorst S., Satava R.,Face content and construct validity of the university of washingtonvirtual reality transurethral prostate resection trainer, Journal of Urol-

ogy 172, 5, 2004, pp. 1953-1957[7] Oppenheimer P., Gupta A., Weghorst S., Sweet R., Porter J., The

representation of blood flow in endourologic surgical simulations, Stud

Health Technol Inform 81, 2001, pp. 365-371[8] Castaneda M. P., Cosio F. A., Deformable model of the prostate for

turp surgery simulation, Computers and Graphics 28, 2004, pp. 767-777

[9] Miguel Angel Padilla Castaneda, Felipe Altamirano del Monte, Fer-nando Arambula Cosıo, Virtual Resectoscope Interface for a SurgerySimulation Training System of the Prostate, 3rd Workshop in Virtual

Reality Interaction and Physical Simulation VRIPHYS06, Madrid,Espana, november 2006, pp. 101-108

[10] Three Channel Optical, Incremental Encoder Modules HEDS-9040 &HEDS-9140; Hewlett Packard; Technical Data 5965-5870E, pp. 2-53

[11] http://www.zworld.com[12] PIC16F87XA Data Sheet 28/40/44-Pin Enhanced Flash Micro-

controllers. Microchip Technology Inc. DS39582B, 2003, p. 234(http://ww1.microchip.com/downloads/en/DeviceDoc/39582b.pdf)

[13] http://www.zeiss.com

1753