Skin Research and Technology 2014; 20: 307–314 Printed in Singapore All rights reserved doi: 10.1111/srt.12120
© 2013 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd Skin Research and Technology
Prototype tactile feedback system for examination by skin touch O. Lee1, K. Lee2, C. Oh3, K. Kim4 and M. Kim2 1
Department of Radiological Science, College of Nursing and Health Science, Gimcheon University, Gimcheon City, Gyunbuk, Korea, 2 Department of Electronics and Information Engineering, 3D Information Processing Laboratory, Korea University, Seoul, Korea, 3 Department of Dermatology, Korea University College of Medicine, Korea University Guro Hospital, Seoul, Korea and 4Department of Electrical and Electronic Engineering, Institute of BioMed-IT, Energy-IT and Smart-IT Technology (BEST), Yonsei University, Seoul, Korea
Background/purpose: Diagnosis of conditions such as psoriasis and atopic dermatitis, in the case of induration, involves palpating the infected area via hands and then selecting a ratings score. However, the score is determined based on the tester’s experience and standards, making it subjective. To provide tactile feedback on the skin, we developed a prototype tactile feedback system to simulate skin wrinkles with PHANToM OMNI. Methods: To provide the user with tactile feedback on skin wrinkles, a visual and haptic Augmented Reality system was developed. First, a pair of stereo skin images obtained by a stereo camera generates a disparity map of skin wrinkles. Second, the generated disparity map is sent to an implemented tactile rendering algorithm that computes a reaction force according to the user’s interaction with the skin image. Results: We first obtained a stereo image of skin wrinkles from the in vivo stereo imaging system, which has a baseline
of 50.8 lm, and obtained the disparity map with a graph cuts algorithm. The left image is displayed on the monitor to enable the user to recognize the location visually. The disparity map of the skin wrinkle image sends skin wrinkle information as a tactile response to the user through a haptic device. Conclusion: We successfully developed a tactile feedback system for virtual skin wrinkle simulation by means of a commercialized haptic device that provides the user with a single point of contact to feel the surface roughness of a virtual skin sample.
HE BODY region and area and the severity of plaque characteristics comprise the standards of the Psoriasis Area and Severity Index (PASI), which is widely used for marking the severity or progression of psoriasis. The highest score for PASI score is 72. PASI shows reduced accuracy when the score is below 12. As the score decreases, the accuracy decreases strongly. When applying PASI to faces, despite applying the “rule of fours” and the fact that the highest score for PASI for facial psoriasis is 72, most cases show a PASI score of 10 or less, meaning that the test is flawed because, in judging the condition or treatment effect, the discerning power is low. This is because the six stages for the infected area (1–9%, 10–29%, 30–49%, 50–69%, 70–89%, 90–100%) and the four stages of severity (slight, moderate, severe, very severe) that comprise the PASI score are
mostly at the lower stages, therefore lowering the discerning power of the PASI score (1). The “rule of fours” mentioned above is a ratio that indicates how much area each part of the face occupies over the entire face in multiples of four. For example, the forehead is set as a multiple of 24%, one cheek as 20%, the perioral area as 8%, one periorbital as 4%, and one aspect of an ear as 4%. In reality, infected areas of the face are less than 1% for 79.5% of patients, meaning that they are grade 1. Therefore, these patients show very little differences in the entire PASI score. Moreover, most of the redness, thickness, and scaling, which are factors constituting severity, remain between grade 1 and grade 2, therefore generating no discernible characteristics between the patients (1). In terms of atopic dermatitis, the Eczema Area and Severity Index (EASI) score is widely
T
Key words: skin – tactile feedback – haptics – psoriasis – disparity
Ó 2013 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd Accepted for publication 19 October 2013
307
Lee et al.
used for evaluating the degree of symptoms, along with the SCORing Atopic Dermatitis (SCORAD) index. The lesion area and the factors of severity (erythema, induration, excoriation, lichenification) are the main indicators of the EASI score, which measures the area distribution of the inflammation according to the skin area (2). These diagnostic methods, in the case of induration (thickness), involve palpating the infected area by means of the hands and then selecting the ratings score. However, the score is determined based on the tester’s experience and standards, meaning that it is highly subjective. As such, when conducting research on the texture of skin and infected areas, the objectification of palpation based on subjective judgment will contribute greatly to the field. In this aspect, studies that have attempted to analyze texture recognition and to apply texture materialization equipment suggest a new direction for the objectification of palpation in dermatology. Asamura et al. used a method that discerns vibratory stimulation and pressure and stimulates them by utilizing the characteristics of a mechanoreceptor, which recognizes touch (3). Essick conducted a cognitive-physiological study on, in the case of feeling the quality of a material by rubbing with a finger, the influence of the rubbing methods and the characteristics of the material (4). Lederman conducted a study on active touching and passive touching (5), and Shimoho et al. studied the influence that pin intervals and shapes have on tactile recognition by utilizing materialization equipment able to replicate the shape of the specific item inserted into it (6). Haptic Technology is a feedback technology that takes advantage of the sense of touch by providing users with forces or vibrations by controlling motors (force feedback) and vibratory actuators (tactile feedback). During the last two decades, haptic technology has advanced rapidly and has enabled investigation of how touch feedback can assist in manipulating virtual objects, in enhancing the remote control of machines, and in improving user interfaces of electronic devices. As a result, haptic technology is playing an increasingly key role in game and mobile devices, tele-machine operation, 4D movies, and various medical applications such as palpation, robot surgery, needle insertion and simulations for training medical students.
308
In terms of skin diagnosis, tactile information is also crucial for, among other things, understanding the aging process and detecting cancer and tumors underneath the skin. For this reason, many researchers use an estimated roughness of the skin surface because this provides physiological features of the skin. A common technology used to obtain skin roughness is the camera. To provide tactile feedback on skin wrinkles, we developed a prototype tactile feedback system to simulate skin wrinkles with PHANToM OMNI. The developed system provides the user with tactile feedback on skin wrinkles captured by a stereo camera while the user strokes the surface of the skin sample with the stylus of the haptic device. This system allows dermatologists to touch the skin surface directly to feel the roughness of skin wrinkles. With the developed system, the cost of retaining skin samples long term may be reduced, and the learning performance of students may be improved.
Materials and Methods Development of a tactile feedback system for skin wrinkles To provide the user with tactile feedback on skin wrinkles, a visual and haptic Augmented Reality (AR) system was developed. With the developed system, a 2D skin wrinkle image is displayed on a standard PC monitor, and tactile force feedback on the skin image is delivered through a commercialized haptic device (PHANToM OMNI, 3 degrees-of-freedom force feedback, 6 lm position resolution, 3.3 N and 2.31 N/mm for max force and stiffness, respectively, manufactured by Sensable Technologies, Wilmington, MA, USA) in response to user’s interaction with the skin image. The haptic device and the PC monitor were installed and synchronized on a personal computer (CPU Intel i5-2400, RAM 8 GB). As seen in Fig. 1(a), the system consists of two stages. First, a pair of stereo skin images obtained by a stereo camera generates a disparity map of skin wrinkles. Second, the generated disparity map is sent to an implemented tactile rendering algorithm (see section ‘visual and haptic rendering for skin wrinkles’), which computes a reaction force according to the user’s interaction with the skin image. The computed force is eventually delivered to the user’s
Tactile feedback for skin
hand via the haptic device without latency, while the 2D skin image is displayed on a PC monitor. Note that the disparity map computed from the stereo skin images is not visible, but is touchable by the user, which is a common technology in AR. The haptic device not only provides force feedback but also senses the position of the end-effector, which is a virtual finger in the 3D virtual space. Because of this capability, the tactile rendering algorithm detects a collision when the user touches the skin surface, and sends the reaction force back to the user. For realistic haptic rendering of feedback with the haptic device, the tactile force feedback is updated well above 1 KHz, while the visual display is updated above 30 Hz. A scenario of a user employing the developed system is depicted in Fig. 1(b). The user, for
example a dermatologist, grasps the pen stylus of the haptic device and strokes the surface of the skin image to feel the various levels of roughness containing crucial tactile information on skin wrinkles.
Disparity map generation of skin wrinkles Stereo vision involves the calculation of disparity, which is the locational difference of two corresponding points that occur after a point in a three-dimensional space is projected on a stereo image. To obtain 3D information that is more accurate and quantitative than using stereo images in the medical field, a system with the following conditions is required (7). First, it must be able to take left and right images at the same time and must not require a replication (a)
(b)
Fig. 1. The developed skin tactile feedback system: (a) The simplified architecture incorporating 3D skin images and skin wrinkle tactile feedback; (b) A scenario of skin wrinkle diagnosis using tactile feedback on skin wrinkles.
309
Lee et al.
process. Second, fine baseline control must be possible in high-resolution videos to obtain the appropriate baseline for biosurface stereo images. Third, it must be possible to obtain non-convergence stereo images to enable quick and accurate matching without the process of rectification. Lee et al. conducted a study to construct a system that would satisfy the conditions mentioned above (7), and from this imaging system, we obtained stereo images. By a stereo-matching algorithm, the obtained stereo images generate disparity on each corresponding pixel, and the intensity of the disparity forms a disparity map. More specifically, to calculate disparity, the similarity of the corresponding point between two images must be evaluated. Therefore, the right image is inspected based on the left image to obtain this similarity. We used a graph cuts method to obtain a disparity map because graph cuts are known to extract excellent results from many matching algorithms. Biosurface imagery was obtained using the stereo imaging system shown in Fig. 2. As can be seen in the figure, the stereo image was obtained by locating the main body near the biosurface that we wanted to measure, before loading it into the computer. The appropriate Field of View (FOV) for the imaging was between the minimum of 5 9 3.8 mm and the maximum of 80 9 60.4 mm. Visual and haptic rendering for skin wrinkles The visual and haptic rendering algorithms for skin wrinkles were implemented via Visual C++ with Chai3D and OpenGL libraries. As Fig. 3
illustrates, the visual and haptic rendering algorithms were incorporated and synchronized in the world coordinate system, which forms a virtual 3D space in which the virtual haptic finger and the skin wrinkle image are displayed, to provide the user with natural interactions (e.g. touching the skin wrinkles with the haptic device). Specifically, a disparity map is first generated from a stereo skin image and then converted into a tactile map (see Eq. 1) after filtering out disparity noise and tuning the scale factor. The tactile map is an invisible data structure (height map) intended to provide touch feedback upon a collision. To achieve this, the generated tactile map is transparently superimposed on the displayed skin image and used only for collision detection of the haptic rendering algorithm (see Fig. 4). Tactile map equation : Tðx; yÞ ¼ S FðDðx; yÞÞ ð1Þ x; yOSR;R size where S is the scale factor, F is the filter to remove noise and D is the disparity map Computing force equation : KðYv YÞ; 0;
Y Yv Y [ Yv
ð2Þ
As illustrated in Fig. 4, when the virtual finger (haptic end-effector avatar) contacts the invisible tactile map, which comprises actual 3D skin wrinkles, a reaction force is computed via Eq. 2. Yv and Y are the virtual finger position at the collision and the actual position of the virtual haptic end-effector pushed down by the user, respectively. The larger the difference between Yv and Y, the larger the reaction force that will be delivered to the user’s hand. K (N/mm) is the stiffness and is set to 0.4 N/mm, approximated for the skin (further study is needed to refine this approximation).
Results and Discussion
Fig. 2. In vivo stereo imaging system.
310
We first obtained a stereo image of skin wrinkles from the in vivo stereo imaging system, which has a baseline of 50.8 lm, and obtained the disparity map with the graph cuts algorithm. The conditions for the graph cuts were lambda 10 and disparity range 50. As shown in Fig. 5, the left image was displayed on the monitor so that the user was able
Tactile feedback for skin
Fig. 3. Visual and haptic rendering blocks for skin wrinkle images. SL,R denote a stereo image pair (left and right), and (x,y,z) represents the position from the haptic device in the 3D space where a collision occurs between the haptic end-effector and the virtual skin wrinkle surface. F denotes a computed reaction force to be delivered through the haptic device to the user’s finger.
Fig. 4. Skin wrinkle tactile rendering.
to recognize the location visually. We chose the left image from the stereo images because it precisely overlapped the disparity map and is the standard in stereo matching. The name, age, gender, patient number, and image acquisition time were recorded on the upper right corner of the skin wrinkle image, whereas information such as FOV, baseline, lambda, display range, and sphere size, which are acquisition conditions for image information, were displayed on the lower left corner to enable the user to confirm the information. When magnifying or contracting the image, a scale bar was placed on the right of the image to facilitate comparisons. The disparity map of the skin wrinkle image that satisfied these conditions, i.e. the 3D information, sent skin wrinkle information as a tactile response to the user through the haptic device. The sphere located within the skin image indicated the location of the probe of the haptic device. Psoriasis, which is a skin disease expected to be capable of auxiliary application in diagnoses
Fig. 5. A prototype tactile feedback system for examination by skin touch.
that use this type of developed system, is a chronic recurrent illness that causes erythematous papules covered with silver-white scales. Over time, these papules grow larger or often merge and turn into large plaque lesions. There are many methods of determining the severity of psoriasis. The oldest and most commonly used subjective method involves categorizing the severity into mild, moderate, and severe according to the infected area. Usually, mild indicates an infected area of less than 5%, moderate indicates an infected area between 5% and 30%, and severe indicates an infected area of 30% or more (8, 9). In recent years, there have been cases in which the standard used for the infected area has been reduced, with mild for an area below 2%, moderate for 2–10%, and severe for 10% or more. Moreover, while the infected area is important for the determination of severity, the activity of the psoriasis is also important. Therefore, the severity is sometimes determined as mild when there are no major
311
Lee et al.
changes within 1 month, as moderate when the psoriasis spreads around the plaque lesions, and as severe when new lesions rapidly appear around existent lesions or in cases of generalized pustular psoriasis or exfoliative psoriasis (10). There is also a global assessment scale that categorizes levels as clear, minimal, mild, average, severe, and very severe (11). We calculated 3D information about the skin with the stereo imaging system, and constructed a system that sends this information to the user through a haptic device. The diagnoses of skin diseases where induration is an important piece of information, such as psoriasis and atopic dermatitis, are difficult to achieve through palpation, because of wounds and itching. However, we confirmed that tactile recognition without direct palpation is possible, and we expect this to contribute to the relevant research fields and to diagnoses. In terms of stereo vision, which we used to calculate 3D information about the skin, researchers in the field are active mainly in stereo matching and in its medical application (7, 12–14).Although there is a nonlinear relationship between disparity and depth, the calibration of these two factors has not become a great issue in the general research of stereo images, because of the lack of necessity or the simplification of calculations. However, for quantification in medical application, studies of calibration are necessary. The parameter of the graph cuts (disparity range, lambda) stereo-matching algorithm that we applied must be appropriately selected for the images that will be applied with the graph cuts, through repeated tests. Although a large disparity range can extract good results because it can divide the range into levels in detail, a range that is too large will lower the image’s brightness, and a range that is too small cannot obtain 3D information about the disparity. Moreover, the sensitivity of the disparity map can be adjusted by changing the lambda. If the lambda is too small, incorrect information such as noise can enter the disparity map, and a lambda that is too large will simplify the disparity of the object in the image and severely crush it, also providing incorrect information about the disparity. The number of iterations in the graph cuts can also be selected. Three iterations are recommended. This is because the process of repetition is intended to determine the global
312
minimum, which can generally be found in three iterations. This is why this algorithm takes more time than other algorithms to conduct the matching calculations mentioned above. Haptic feedback in medical applications is an emerging field because of the importance of life-saving and cost-saving technology. For instance, surgical education requires extensive practice on patients under faculty supervision and is often limited because of the high cost. To address this problem, Moody et al. (15) demonstrated the effect of force feedback in the training and assessment of surgeons. They developed a visuo-haptic surgical training system with PHANToM Desktop and simulated suturing to trainees. A total of 20 participants took the experiment, and the results showed that force feedback significantly reduced stitching time. Since then, many researchers have shown the effect of haptic feedback in simulations such as virtual venipuncture training, needle insertion, bone surgery, and telepresence surgery (16–19). The most successful example of tactile feedback in medicine is palpation, which is a method of feeling the presence or absence of physiological features or abnormalities on/ inside the human body. A practitioner may press on an area of interest with his or her finger and collect tactile information by moving the finger. In this case, assessment by tactile feedback outperforms examination by visual feedback. For this reason, many tactile feedback systems have been developed for palpation by employing haptic technology. Studies (20–21) demonstrate the effectiveness of training in detection of tumors and cancer on several human organs such as neck and breast. Most interestingly, Baillie et al. (22) developed a virtual cow system using PHANToM for bovine rectal palpation training. They demonstrated that training with the virtual cow simulator significantly improved students’ palpation skills compared with a group of students trained in a traditional way. These studies support the effect of adding tactile feedback, particularly to existing medical systems used for examining tactile information solely by vision. Of the five human sensorial modalities, two main ones are vision and touch. The sense of touch often becomes important when touch interactions are required or when vision is weak because of a limited environment (e.g. no
Tactile feedback for skin
light and touching invisible objects). For instance, learning tactile information on a skin surface for a diagnosis can be achieved by using either vision or touch. A subsequent question is then which modality is superior or more useful for measuring skin surface roughness, because this can inform dermatologists to choose a better skin diagnosis system. Kim’s (23) study attempted to answer a similar question, about the sensitivity of the two modalities (vision and touch) in detecting bumps on 3D surface patches as the roughness values are varied. He demonstrated that touch is more sensitive to detecting bumps on surface patches of low roughness, whereas vision becomes dominant on a surface patch of high roughness. His study indicates that the sense of touch sometimes plays an important role in investigating tactile information on a skin surface that may have differing roughness values. Nonetheless, until now, most skin studies have focused on vision. The main reason for this may be a lack of suitable haptic devices and technologies, as compared with vision, that can transparently deliver natural tactile feeling to the user’s fingertip, which has primary mechanoreceptors to sense cutaneous information. In this study, we therefore developed a tactile feedback system for virtual skin wrinkle simulation by using a commercialized haptic device (PHANToM OMNI) that provides the user with a single point of contact to feel the surface roughness of a virtual skin sample. It may be argued that this single point contact would decrease the discrimination of skin roughness and that the perceptibility would also be decreased by the impedance created by the rigid mechanical link (PHANToM OMNI) between the skin and user’s hand, as compared with using a bare finger. In general, 3D shape perception with the hand requires a more kinesthetic cue (i.e. the configuration of finger joints) than cutaneous information (tactile cue from fingertips), whereas 3D texture is the opposite (24, 25). Using a single contact haptic device is therefore sufficient for the user to study the roughness of a skin sample. A challenge in developing a skin wrinkle simulation system is how to accurately mimic bare finger touching with a haptic device; Klatzky and Lederman (26) show that a single haptic probe can sufficiently deliver 3D texture information, but that bare finger touching still outperforms in the perception of surface roughness. We believe
that our developed system is a good approach to addressing this problem, and we intend to continue our research to advance our technology. Our developed system provides several benefits in the field of skin medical applications. First, any stereoscopic skin images can be easily converted into 3D haptic data for tactile feedback. In this way, dermatologists can touch the real-like skin surface by using their hands. Second, the virtual skin wrinkle image is touchable at all times and is not destroyed by touching to diagnose disease. Third, the system can be used to train medical students alongside other applications (15–22, 27, 28). Fourth, the system makes remote diagnosis possible to benefit people living in places far from hospitals. To the best of our knowledge, our developed system is the first example of haptic feedback for skin wrinkles. Although this development shows promise and is a good candidate for skin simulation, some research problems must be solved in the near future. First, modeling human skin material is imperative to provide real-like touch feedback on virtual skin. Second, an evaluation study must follow to provide actual evidence of the effect of tactile feedback on skin disease detection and the improvement of students’ learning. Finally, and most importantly, we will continue to develop a skin tactile feedback system that enables users to touch virtual skin with their bare fingers instead of grasping a haptic device, because the former is a more natural form of interaction.
Acknowledgements This study was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2013R1A1A4A0 1006895, 2012R1A1A2043685), the Ministry of Science, ICT and Future Planning (2012R1 A1A2006556), the Foundation of Dongil Scholarship, and a Korea University Grant. The corresponding author (KK) has been supported by Institute of BioMed-IT, Energy-IT and Smart-IT Technology (BEST), a Brain Korea 21 plus program, Yonsei University.
Conflict of interest The authors declare that there are no conflicts of interest.
313
Lee et al.
References 1. Kwon H, Kwon I, Jo S, Yoon H, Youn J. Facial psoriasis log-based area and severity index (fPLASI): construct validity of a new facial psoriasis measurement tool. Br J Dermatol 2011; 165: 203–204. 2. Hanifin J et al. The eczema area and severity index (EASI): assessment of reliability in atopic dermatitis. Exp Dermatol 2001; 10: 11–18. 3. Asamura N, Yokoyama N, Shinoda H. Selectively stimulating skin receptors for tactile display. IEEE Comput Graphics Appl 1998; 18: 32–37. 4. Essick G. Factors affecting direction discrimination of moving tactile stimuli. Adv Psychol 1998; 127: 1–54. 5. Lederman SJ. The perception of surface roughness by active and passive touch. Bull Psychon Soc 1981; 18: 253–255. 6. Shimojo M, Shinohara M, Fukui Y. Human shape recognition performance for 3D tactile display. IEEE Trans Syst Man Cybern Part A: Syst Hum 1999; 29: 637–644. 7. Lee O et al. An optimized in vivo multiple-baseline stereo imaging system for skin wrinkles. Opt Commun 2010; 283: 4840–4845. 8. Farber E, Nall L. Epidemiology: natural history and genetics. Basic Clin Dermatol 1998; 16: 107–158. 9. Roenigk H, Epstein E, Maibach H. Skin manifestations of psoriasis and eczematous psoriasis: maturation. Basic Clin Dermatol 1998; 16: 3–12. 10. Haftek M, Gli nski W, Jabło nska S, Obałek S. T lymphocyte E rosette function during photochemotherapy (PUVA) of psoriasis. J Invest Dermatol 1979; 72: 214–218. 11. Berth-Jones J et al. A study examining inter-and intrarater reliability of three scales for measuring severity of psoriasis: psoriasis area and severity index, physician’s global assessment and lattice system physician’s global assessment. Br J Dermatol 2006; 155: 707–713. 12. Lee O et al. Improved quantitative evaluation of atopic dermatitis by
314
13.
14.
15.
16.
17.
18.
19.
20.
21.
camera calibration and rectification of a stereo image. Skin Res Technol 2011; 17: 11–19. Lee O et al. Effect of pedunculagin investigated by non-invasive evaluation on atopic-like dermatitis in NC/Nga mice. Skin Res Technol 2010; 16: 371–377. Kim JE et al. The efficacy of stereoimage optical topometry to evaluate depressed acne scar treatment using cultured autologous fibroblast injection. Dermatol Surg 2011; 37: 1304–1313. Moody CL, Baber C, Arvanitis TN. The role of haptic feedback in the training and assessment of surgeons using a virtual environment. Proceedings of eurohaptics, Birmingham, UK, 1–4 July, 2001. Smith SP, Todd S. ‘Collaborative evaluation of a haptic-based medical virtual environment.’ Proceedings of the 4th INTUITION International Conference on Virtual Reality and Virtual Environments, 4-5 October 2007, 102–110. Okrainec A et al. Development of a virtual reality haptic Veress needle insertion simulator for surgical skills training. Stud Health Technol Inform 2009; 142: 233–238. Morris D et al. Visuohaptic simulation of bone surgery for training and evaluation. IEEE Comput Graphics Appl 2006; 26: 48–57. Boulanger P et al. Hapto-audiovisual environments for collaborative training of ophthalmic surgery over optical network. IEEE International Workshop on Haptic Audio Visual Environments and their Applications, 2006. Stalfors J, Kling-Petersen T, Rydmark M, Westin T. Studies in health technology and informatics, Vol. 81, IOS Press Ebooks, 2011. Alhalabi MO et al. Medical training simulation for palpation of subsurface tumor using HIRO. Eurohaptics Conference, 2005, and Symposium on Haptic Interfaces for Virtual Environment and Teleoperator Systems, 2005. World
22.
23.
24.
25. 26.
27.
28.
Haptics 2005. First Joint, 2005. IEEE: 623–624. Baillie S et al. Validation of a bovine rectal palpation simulator for training veterinary students. Stud Health Technol Inform 2005; 111: 33–36. Kim K. Ch.4 Rougness-adaptive 3D visuohaptic watermarking. Dissertation , West Lafayette, IN: University of Purdue, pp 74–94. Klatzky RL et al. Haptic identification of objects and their depictions. Percept Psychophys 1993; 54: 170– 178. Kim K et al. Visuohaptic discrimination of 3D gross shape. Seeing Perceiving 2012; 25: 3–4. Klatzky RL, Lederman SJ. Tactile roughness perception with a rigid link interposed between skin and surface. Percept Psychophys 1999; 61: 591–607. Bello F. et al. Imagine-S: imaging guided interventional needle simulation. Proceedings of the Eurographics 2009-Medical Prize, 2009: 5–8. Ra J et al. Spine needle biopsy simulator using visual and force feedback. Comput Aided Surg 2002; 7: 353–363.
Addresses: K. Kim Human Computer Interaction Group Microsoft Research Asia No. 5, Danling Street Haidian District Beijing China Tel: +86 156 5279 1432 Fax: +86 10 8286 8526 e-mail:
[email protected] M. Kim 3D Information Processing Lab Korea University 126-1 Anam-dong 5ga, Seongbuk-gu Seoul 136-701 Korea Tel: +82 2 3290 3977 Fax: +82 2 3290 3977 e-mail:
[email protected]