How To Track A Minimally Invasive Surgical Instrument With An Optical Sensor

Transcription

1 Sensors and Actuators A 126 (2006) TrEndo, a device for tracking minimally invasive surgical instruments in training setups Magdalena K. Chmarra, Niels H. Bakker, Cornelis A. Grimbergen, Jenny Dankelman Man Machine Systems Group, Faculty of Mechanical, Maritime, and Materials Engineering, Delft University of Technology, Delft, The Netherlands Received 28 June 2005; received in revised form 18 October 2005; accepted 27 October 2005 Available online 29 November 2005 Abstract A novel, four degrees of freedom, low-cost device for tracking minimally invasive surgical instruments (MIS) in training setups was developed. This device consists of a gimbal mechanism with three optical computer mouse sensors. The gimbal guides the MIS instrument, while optical sensors measure the movements of the instrument. To test the feasibility of using optical mouse sensors to track MIS instruments, the accuracy of these sensors was tested depending on three conditions: distance between lens and object, velocity of movements, and surface characteristics. The results of this study were used for developing a prototype TrEndo. Tests performed on TrEndo show that the smallest movement that can be recognized by sensors is 0.06 mm for translation and 1.27 for rotation of the MIS instrument around its axis. The smallest recognized angle for rotation around incision point is The accuracy of TrEndo is higher than 95%, and therefore allows tracking the movements of the MIS instrument Elsevier B.V. All rights reserved. Keywords: Optical sensor; Gimbal; Minimally invasive surgery; Training; Instrument tracking; Objective assessment 1. Introduction Minimally invasive surgery (MIS, endoscopic surgery) is performed using special long instruments that are inserted into the patient s body through small incisions in the skin. Visual feedback of the operating area is obtained by an endoscopic camera, which provides a 2D image on a monitor. MIS is thus a difficult surgical technique and it takes a relatively long learning curve to master it fully [1]. To guarantee safe use of instruments, training of the operative skills is very important. Training can be applied in box-trainers, virtual reality trainers (VR trainers), animal experiments, and in patients [2 4]. In all configurations, objective assessment of a trainee s skills remains a challenge. Presently, there is no widely used automatic method to evaluate technical MIS skills during training [5 7]. Mostly, the assessment requires a lot of the teacher s effort and is subjective [8 11]. Most VR trainers provide a scoring system based on information about the movements of the instruments [12 15]. They Corresponding author. Tel.: ; fax: address: j.dankelman@wbmt.tudelft.nl (J. Dankelman). also provide a database of the previously performed exercises, which can be used for motivating trainees. Unfortunately, VR trainers do not provide natural instrument tissue interaction and they do not allow using standard MIS instruments (e.g. graspers or scissors). To be able to automatically rate surgical performance in a training setup that uses standard MIS instruments, a device for tracking these instruments is required. Devices available on the market that can track and record movements of the MIS instruments are complex and expensive, some of them limit the freedom of movement, and most of them are intended to be used in VR environments only [16 19]. The aim of this study is to develop a simple and low-cost device that allows free manipulation of a standard MIS instrument either in a box-trainer or a VR trainer and that tracks its movements. 2. Requirements The design of a low-cost interface device that tracks the movements of MIS instruments should meet the following requirements: 1. Realistic manipulation of the MIS instrument in four degrees of freedom (DOFs): translation of the instrument along its /$ see front matter 2005 Elsevier B.V. All rights reserved. doi: /j.sna

2 M.K. Chmarra et al. / Sensors and Actuators A 126 (2006) axis (1st DOF), rotation of the instrument around its axis (2nd DOF), left right (3rd DOF) and forward backward (4th DOF) rotations of the instrument around the incision point. 2. Possibility to use with real MIS instruments (Ø 5 mm) in a box-trainer with or without combining with virtual reality environments. 3. The level of accuracy and sensitivity suitable to allow reliable assessment of trainees. 4. Low-cost and easy to produce to make a training facility affordable for every (training) hospital or private use. 5. Plug-and-play in PC ready to use. 6. Small size in order to be easily carried and mounted, e.g. on a box-trainer; it should fit on an average work desk. Based on these requirements, a prototype of the device was developed. 3. Prototype of the tracking system TrEndo The main principle of the developed prototype for tracking MIS instruments TrEndo is to mimic the incision (pivoting) point by a gimbal mechanism and to track the movements of the instrument by means of optical sensors. The gimbal is a simple mechanism that allows the desired degrees of freedom. The optical sensors are the same as those used in a computer mouse, being thus cheap and easily available Gimbal mechanism TrEndo (Tr, tracking; Endo, endoscopy) consists of a twoaxis gimbal mechanism in which three optical mouse sensors are incorporated (Fig. 1). The role of the gimbal mechanism in TrEndo is to guide the MIS instrument. The gimbal mechanism allows four DOFs. The 1st and the 2nd DOFs are introduced by designing a rectangular rigid body inner ring with a cylindrical hole. In this hole, standard MIS instruments ( 5 mm) can be inserted, retracted, and rotated in any way that is possible during MIS (Requirements 1, 2). The middle ring allows the left right rotation of the inner ring (3rd DOF). The outer (U-shaped) ring allows forward backward rotations of the middle ring with the instrument (4th DOF). The maximum allowable angles in sideways directions are approximately ± Optical sensors Optical computer mouse sensors enable the contactless measuring of the movements of standard instruments in TrEndo. In the rectangular rigid body of the gimbal mechanism two sensors are placed. Sensor 1 measures movements of the MIS instrument (translation along and rotation around its own axis 1st and 2nd DOFs). Sensor 2 is facing a fixed semicircular surface placed on the middle ring and measures rotations of the inner ring (3rd DOF). Sensor 3 is located on the outer ring of TrEndo. This sensor is facing the second fixed semicircular surface placed on the middle ring and measures the rotations of the middle ring (4th DOF). In Fig. 2, a schematic drawing of an optical mouse sensor system is presented. The sensor measures changes in position over a flat surface by taking surface pictures (frames) with a small CCD chip at high frame rate [20,21]. Images of the surface (Fig. 3) are acquired via lens and illumination system. A digital signal processor (DSP) by correlation determines from these images the direction and the magnitude of motions. A stream of x and y relative displacement values (given in number of sensor counts that the camera sees) generated by the DSP is sent to the serial peripheral interface. This signal is converted to a USB-signal, which means that optical mouse sensors system is plug-and-play (Requirement 5). The low-cost and ready to use requirements are met because the gimbal mechanism is easy to manufacture and the production costs are low. Furthermore, the computer mouse is an inexpensive and very common computer input device. The compact appearance of the sensors keeps the device manageable and portable (Requirement 6). Finally, the elements of the device, such as print circuit board (PCB) housing, can be designed in such way that the production is easy (Requirement 4). The accuracy and the sensitivity of the prototype have to be investigated in order to find out whether Requirement 3 is met. Fig. 1. TrEndo a prototype of a device for tracking the movements of the MIS instrument. Sensor 1 tracks the insertion-retraction, and the rotation of the MIS instrument around its own axis. Sensor 2 tracks the left right rotations of the instrument by recording movements of a semicircular surface (SCS 1) mounted on the middle ring. Sensor 3 tracks forward backward rotations by recording movements of another semicircular surface (SCS 2) mounted on the middle ring. Sensor 3 is located on the outer ring. Fig. 2. Schematic drawing of the optical mouse. PCB, print circuit board [21] ( Agilent Technologies, Inc., 2003; reproduced with permission, courtesy of Agilent Technologies, Inc.).

3 330 M.K. Chmarra et al. / Sensors and Actuators A 126 (2006) smallest measured displacement is 65 m. However, the resolution of the sensor changes when the conditions under which the sensor works change [21]. The following conditions were investigated: the distance between lens and the tracked surface; the velocity of the movements; the surface characteristics of the tracked object. Fig. 3. Example of what the optical mouse sensor sees on a surface (resolution pixels). Details from top left image are compared with details from left down image. The differences in positions between images give information about the position of the mouse. In this case, the mouse is moving down to the right (from: Logitech, The MX TM Optical Engine, epd/i tech.html). 4. Methods 4.1. Tests of the optical sensor Before the prototype was developed, several conditions that affect the accuracy of the optical sensor were investigated. For this reason, a series of tests using a standard optical computer mouse sensor, comparable to the smallest available optical mouse sensor (Agilent ADNS-2620), were carried out. According to the specification of the manufacturer, the ADNS-2620 sensor can track motions up to 12 in./s (305 mm/s) [21]. Default resolution of the ADNS-2620 sensor is approximately 400 counts/in. (16 counts/mm). This means that the These conditions were tested using the measuring setup shown in Fig. 4. A computer-driven CNC milling machine was used to accurately control the position of the sensor in x, y, and z directions (up to 1 m). The milling machine also allows making movements with different velocities during the measurements. The sensor was mounted on the arm of the milling machine and moved 50 mm forward and backward (movement from point A to B to A). In the ideal situation, the optical sensor measures the same movement in both directions. In reality, however, this does not occur. During this study the relative error is calculated as: Error = d dif 100 (%) (1) d total where d dif is the difference between the initial and return position measured by the sensor given in sensor counts and d total is the whole measured distance in sensor counts. All tests were repeated five times and data from the sensor were recorded on a PC with a sample frequency of 10 Hz Test: distance lens object In order to find the optimal distance lens object, the distance between tracked surface and lens z was changed within the range mm. The velocity of the movements during all measurements was 1000 mm/min. A sand paper with 1000 parts/in. 2 was used as tracked surface Test: velocity of movements Tests were performed for five different velocities: 200, 500, 1000, 3500, and 6000 mm/min. The distance lens object was Fig. 4. Measuring setup for testing the optical sensor. The left picture presents the arm of the computer-driven CNC milling machine, on which the optical sensor was mounted. The right picture presents a more detailed view.

4 2.4 mm. Sand paper with 1000 parts/in. 2 was used as tracked surface Test: surface characteristics Four different groups of surfaces were investigated: different metals (aluminium blasted with glass beads, stainless steal blasted with glass beads, and sandblasted brass), sand papers (2000, 1000, 400, 240, and 40 parts/in. 2 ), one-colour prints, and two-colour prints (printed on the laser printer). The distance lens object was 2.4 mm and the velocity of movements was 1000 mm/min. M.K. Chmarra et al. / Sensors and Actuators A 126 (2006) Tests of the TrEndo tracking system To keep the TrEndo device as small as possible, the smallest available optical sensors Agilent ADNS-2620 were used. The size of the ADNS-2620 footprint is 10 mm 12.5 mm. As referred in Section 4.1, the sensor s resolution depends on different factors (e.g. distance lens object, velocity of movements, surface characteristics) [21]. The calibration of every sensor in TrEndo is thus necessary. A special print circuit board and a base plate were designed so that they were smaller than those in the original Agilent design. The USB controller was separated from the sensor part (including the optical sensor, lens, and the necessary electrical components). The superfluous parts of the lens were removed and a smaller LED was used. The print circuit boards and the base plates were designed such that they can be used for all three sensors in the prototype, keeping production costs low. Normally, optical mouse sensors go in a sleeping mode after 1 s of no movement [21]. Getting the sensor awake takes a few milliseconds delay, which causes errors during measurements. Therefore, the sleeping mode of the sensor was turned off. Unfortunately, turning off the sleeping mode function caused a temperature rise of the sensors. To prevent this, holes in the PCB housing were necessary for cooling. During the tests the MIS instrument was simulated by a 5 mm diameter sandblasted brass stick. This stick was connected to the arm of the milling machine. In order to guarantee free movements of the stick, the prototype was mounted on a high and stiff box (Fig. 5). The box was clamped to the computer driven table of the milling machine. To determine the accuracy and sensitivity of TrEndo for the different movements the following tests were performed Test: 1st DOF (translation) The instrument was translated along its axis 50 mm up and 50 mm down, using three different velocities: 2000, 4000, and 6000 mm/min Test: 2nd DOF (rotation) The instrument was rotated around its axis with velocities: 10, 30, and 60 rotations/min in both clockwise and counterclockwise directions. A rotation of 360 was recorded and analysed. Fig. 5. A schematic view of the measuring setup for tests described in Sections (1) Arm of the milling machine, (2) universal joint, (3) instrument, (4) tracking system, (5) stiff box, and (6) table of the milling machine Test: 3rd DOF (left right) The instrument was rotated in left right direction around the incision point. Movements of 22 were applied for velocities: 2000, 4000, and 6000 mm/min Test: 4th DOF (forward backward) The left right rotation of the instrument around the incision point was evaluated in the same way as the left right rotation. All tests were repeated 20 times. During performing these tests, all three sensors were continuously turned on. Normally, the Windows System does not distinguish between multiple pointing devices at the application program level (which means that all pointer devices control the same cursor). During our tests we needed to differentiate signals from three mouse sensors in order to deliver the right information to the computer. Therefore, we used a CPNMouse driver and application programming interface (API) [22,23] to write a computer program for recording data from more than one pointing device sensor at the same time. Data from all sensors were recorded on a PC with a sample frequency of 100 Hz. During tests of the prototype the relative errors were determined in two ways: moving from point A to B (A B) and from point A to B to A (A B A). The relative errors of the tests were calculated by: Error = (d X real d X meas ) 100 (%) (2) d X real where d X real is the distance (A B and A B A, respectively, for the test) in millimeters given by the milling machine and d X meas is the distance measured by the sensor (A B and A B A, respectively, for the test) in millimeters. To compute the distances in millimeters, the resolutions of the sensors should be determined by a calibration. TrEndo tracks motions in four DOFs, so the calibration of sensors was performed independently for each DOF. Calibrations were performed on the first four tests and averaged over all velocities of motion. The mean and the standard deviation of the relative errors were computed on the rest (16) of the tests. The calculated mean errors and standard deviations for both types of tests can be interpreted differently. The mean error of test A B A gives the information about possible systematic errors. The mean error

5 332 M.K. Chmarra et al. / Sensors and Actuators A 126 (2006) of test A B is caused by a sum of two effects: the same phenomenon that caused the mean error in test A B A, and the use of a wrong resolution. The standard deviation gives in both cases information about the random error. 5. Results 5.1. Results of the optical sensor Test: distance lens object The results of this test are presented in Fig. 6. The relative error was smallest when the distance lens object was within the range mm Test: velocity of movements The dependency of the sensor s resolution on velocity is small. The largest difference in measured number of sensor counts while moving from A to B with different velocities was less than 1%. Fig. 7 shows the results of velocity test. The test shows that the mean relative error increases with the velocity. The highest obtained error was for the highest velocity. During minimally invasive surgery a velocity of 6000 mm/min is achieved only Fig. 7. Relative error (mean ± S.D.) as a function of the velocity of the movements. when the surgeon pulls out the instrument from the patient s body. In the normal working range (velocity of movement smaller than 6000 mm/min) the error is less than 0.5% Test: surface characteristics The results of surface characteristics test show that the smallest errors were obtained for sandblasted brass, sand papers, and two-coloured prints. Therefore, a sand paper (400 parts/in. 2 )was used as a surface for sensors 2 and 3 in the prototype. One-colour prints did not work at all. The reason is that the sensor measures changes in position over a flat surface by taking images and in the case of one-colour prints, every image taken by the camera looks the same. So, the sensor does not recognize any movements on one-colour prints without texture. Fig. 6. Relative errors (mean ± S.D.) as a function of the distance between lens and object Results of the TrEndo tracking system Results of the prototype test are presented in Table 1. Table 1 Results of testing tracking system Test Speed Distance A B a A B a A B A b Mean relative error ± S.D. (%) Mean relative error ± S.D. (%) Translation c 2000 (mm/min) 50 mm 2.56 ± ± (mm/min) 50 mm 1.54 ± ± (mm/min) 50 mm 1.05 ± ± 0.95 Clockwise and counter-clockwise rotation d 10 (rpm) ± (rpm) ± (rpm) ± 1.72 Left right rotation c 2000 (mm/min) ± ± (mm/min) ± ± (mm/min) ± ± 0.75 Forward backward rotation c 2000 (mm/min) ± ± (mm/min) ± ± (mm/min) ± ± 0.95 a Tests A B: movement from point A to B in the shortest possible way. b Tests A B A: movement from point A to B to A in the shortest possible way. c Mean was calculated on 16 repeated tests. d Mean was calculated on 32 repeated tests.

6 M.K. Chmarra et al. / Sensors and Actuators A 126 (2006) Fig. 8. Examples of results of tests A B and A B A. Presented results A B (left) and A B A (right) were performed for the translation of the instrument along its axis with velocity of 4000 mm/min. Each test was repeated 20 times Test: 1st DOF (translation) For the translation of the instrument along its axis the calibration resulted in a resolution of 17 counts/mm. This means that the smallest displacement that can be seen by TrEndo is 60 m. The maximal mean error during translation was for test A B and occurred for the lowest velocity (Table 1). Examples of the results of tests are shown in Fig Test: 2nd DOF (rotation) For the rotation of the instrument around its axis the calibration resulted in a sensor resolution of 459 cpi. This means that the smallest rotation that can be observed by TrEndo is Test: 3rd DOF (left right) During left right rotation of the instrument around the incision point no clear relation between velocity and error was found. Mean relative errors of test A B are small (less than 0.5%). The calibration for this test resulted in a sensor resolution of 475 cpi. This means that the smallest angle recognized by TrEndo in left right direction is Test: 4th DOF (forward backward) The test results for forward backward rotation of the instrument around the incision point do not show any trends. The calibration for this test resulted in a sensor resolution of 535 cpi. This means that TrEndo recognizes angles bigger than 0.23 in forward backward direction. 6. Discussion The objective of this study was to design and build a device (TrEndo) that tracks MIS instruments during training of MIS skills. The tests of the optical mouse sensor have shown that the following characteristics give the smallest errors (totally lower than 5%): 1. a distance of mm between lens and tracked object; 2. a velocity of movements up to 5000 mm/min; 3. a rough surface or a patterned surface. Furthermore, results of the prototype test (presented in Table 1) show that errors are smaller than 5%, resulting in an accuracy of more than 95%. The design based on the combination of the gimbal mechanism with three optical mouse sensors results in a simple and inexpensive system, which can follow and record movements of the standard MIS instrument in four DOFs. During performing MIS, small movements in the incision point can result in large movements of the tip of the MIS instrument. Therefore, sensors that record instrument s movements at the incision point should be able to recognize small movements. The smallest movement in 1st DOF (translation of the instrument along its axis) that can be recognized by the sensor in TrEndo is 0.06 mm. The smallest angle in 2nd DOF (rotation of the instrument around its axis) that can be recognized by TrEndo is These results were obtained using a 5 mm diameter sandblasted brass stick. Normally, the surface of MIS instruments is shiny and black. Therefore, modification of this surface is necessary. This can be easily done by roughing the surface with sand paper. Additionally, the diameter of standard MIS instruments does not always equal 5 mm. For example the diameter of an endoscopic camera equals 10 mm. Moreover, many times the diameter of MIS instruments is not constant over the whole working length of the instrument. This could affect the measurements of the movements because the resolution of the sensor depends on the distance lens object. A solution to this problem can be to design a tube, in which the instrument will be placed and fixed. If this tube has a rough surface, no modification of the instrument surface to improve the performance of the sensor will be needed. Moreover, using such a tube has one more advantage; since the diameter of the tube will be bigger than the diameter of the MIS instrument, smaller angles than 1.27 in the 2nd DOF will be recognized by TrEndo. Under different conditions the optical sensors track with varying resolution. To avoid these difficulties, a calibration of the device before using it in a training setup is necessary. Our results indicate that this calibration is only necessary once. In the literature we found that there are large differences between trajectories of experienced and inexperienced surgeons [24]. Hence, the sensitivity of the device is expected to be high enough.

7 334 M.K. Chmarra et al. / Sensors and Actuators A 126 (2006) Conclusions A novel low-cost tracking device that can be used during training of minimally invasive surgical skills TrEndo was developed. This device allows free manipulation of standard MIS instruments in a training setup, such as box and virtual reality. Due to the use of optical sensors from a computer mouse, TrEndo is simple to use and gives easy access to the computer via a standard USB port. The experiments show that MIS instrument can be accurately tracked by the system. Acknowledgements The authors would like to thank Medical Technology Development department from Academic Medical Center Amsterdam for help in designing and manufacturing of the prototype. References [1] M.J. Moore, C.L. Bennett, The learning curve for laparoscopic cholecystectomy, Am. J. Surg. 170 (1995) [2] A.M. Pearson, et al., Evaluation of structured and quantitative training methods for teaching intracorporeal knot tying, Surg. Endosc. 16 (2002) [3] R. Aggarwal, et al., Laparoscopic skills training and assessment, Br. J. Surg. 91 (2004) [4] Y. Munz, et al., Laparoscopic virtual reality and box trainers: is one superior to the other? Surg. Endosc. 18 (2004) [5] K.R. Wanzel, et al., Teaching the surgical craft: from selection to certification, Curr. Prob. Surg. 39 (2002) [6] L.S. Feldman, et al., Using simulators to assess laparoscopic competence: ready for widespread use? Surgery 135 (2004) [7] A. Darzi, et al., The challenge of objective assessment of surgical skill, Am. J. Surg. 181 (2001) [8] A. Darzi, et al., Assessing operative skill needs to be become more objective, Br. Med. J. 318 (1999) [9] J.A. Martin, et al., Objective structured assessment of technical skills (OSATS) for surgical residents, Br. J. Surg. 84 (1997) [10] K. Moorthy, et al., Objective assessment of technical skills in surgery, Br. Med. J. 327 (2003) [11] J. Shah, A. Darzi, Simulation and skills assessment, in: IEEE Proceedings of the International Workshop on Medical Imaging and Augmented Reality (MIAR 01), 2001, pp [12] J. Torkington, et al., The role of the basic surgical skills course in the acquisition and retention of laparoscopic skills, Surg. Endosc. 15 (2001) [13] A.G. Gallagher, et al., Objective psychomotor skills assessment of experienced, junior, and novice laparoscopists with virtual reality, World J. Surg. 25 (2001) [14] A.G. Gallagher, et al., Discriminative validity of the minimally invasive surgical trainer in virtual reality (MIST-VR) using criteria levels based on expert performance, Surg. Endosc. 18 (2004) [15] M. Schijven, J. Jakimowicz, Virtual reality surgical laparoscopic simulators. How to choose, Surg. Endosc. 17 (2003) [16] G. Lacey, A Surgical Training Simulator, Patent Number IE [17] D. Demirtas, et al., Joystick, Patent WO [18] L.B. Rosenberg, Physically Realistic Computer Simulation of Medical Procedures, United States Patent Number 6,654,000 B2. [19] L.B. Rosenberg, et al., Computer Apparatus Including Linkage Having Flex, United States Patent Application Number A1. [20] T.W. Ng, The optical mouse as a two-dimensional displacement sensor, Sens. Actuators A Phys. 107 (2003) [21] Agilent, ADNS-2620 Optical Mouse Sensor Data Sheet, literature.agilent.com/litweb/pdf/ en.pdf. [22] CPN Group, University of Aarhus, [23] M. Westergaard, Supporting Pointing Devices in Microsoft Windows, Microsoft Summer Workshop for Faculty and PhDs, Cambridge, England, 2002, [24] N. Stylopoulos, et al., Computer-enhanced laparoscopic training system (CELTS), Surg. Endosc. 18 (2004) Biographies Magdalena K. Chmarra was born in Warsaw, Poland, in She received her MSc degree in mechatronics from the Warsaw University of Technology in 2002, with specialisation in biocybernetic and biomedical engineering. She is currently a PhD student at Man Machine Systems Group, Faculty of Mechanical, Maritime, and Materials Engineering at the Delft University of Technology. Her research concerns training and assessment of the technical skills in minimally invasive surgery. Niels H. Bakker was born in Dormaa Ahenkro, Ghana, in After obtaining his MSc in mechanical engineering at the Delft University of Technology in the Man Machine Systems Group in 1995 he worked for a year as research scientist at the Department of Transportation and Logistic. In 1996, he started a four-year PhD project in mechanical engineering in cooperation with the TNO Human Factors Research Institute resulting in a degree in February From 2001, he worked as a postdoc at the Medical Technology Group of the Academic Medical Centre in Amsterdam. Currently he is working as a clinical scientist at Philips Medical Systems in Best. Cornelis A. Grimbergen was born in Leiden, The Netherlands, in After receiving his Ir. degree in electrical engineering at the Delft University of Technology he was active in research in solid-state physics, semiconductor technology, resulting in a PhD degree in 1977 from the University of Groningen. From 1977, he has been with the Laboratory of Medical Physics of the Faculty of Medicine of the University of Amsterdam working as an assistant professor. From 1991, he is a part-time professor at the Measurement & Control Department of the Faculty of Mechanics of the Delft University of Technology. From 1995, he is also professor in Medical Technology at the Academic Medical Center of the University of Amsterdam (AMC). His interests and research projects are in the fields of medical image processing, minimally invasive techniques and safety and training. Jenny Dankelman was born in Ommen, The Netherlands, in She obtained her degree in mathematics, with specialisation in system and control engineering, in 1984 at the University of Groningen. Her PhD degree was obtained in 1989 at the Man Machine Systems Group, Faculty of Mechanical, Maritime, and Materials Engineering, Delft University of Technology (DUT) on her research of the dynamics of the coronary circulation. This work was performed in close co-operation with the Department of Medical Physics of the Academic Medical Center, University of Amsterdam. She continued her research on the coronary circulation for 3 years as postdoc at both universities. Since 1992, she is researcher at Man Machine Systems Group, DUT, and since 2001 professor in biomedical engineering. Her interests and research projects are in the fields of medical instruments, patient safety, and minimally invasive techniques.