nature.com

In vivo evaluation of tissue damage from varying grasping forces using the Saroa surgical system

Abstract

Robot-assisted surgery can help to reduce patient burden and operator stress by enabling precise manipulations with multiple joint motions, but may also cause complications due to the lack of tactile sensation. The Saroa surgical system was developed with a haptic feedback function, and allows operators to adjust grasping forces as desired. In this study, we investigated tissue damage from varying grasping forces using the Saroa surgical system, and assessed the utility of this system. The grasping forceps of the Saroa system were used to grasp the lungs, esophagus, aorta, liver, spleen, small intestine, and large intestine of six beagle dogs with forces of 1, 2, and 3 N for durations of 1, 2, and 4 min. The effects of different grasping forces and durations on tissue damage were histologically evaluated. Histological evaluations showed that grasping force caused tissue damage in the lung and liver, but not the other organs. These results showed the lung and liver were more vulnerable to grasping forces, and exhibited more severe tissue damage at higher forces. These findings suggest that the haptic feedback function of the Saroa system could help to reduce intraoperative organ damage, especially in the fields of lung and liver surgery.

Introduction

The first “robotic surgeon” used on a human patient was the Programmable Universal Machine for Assembly (PUMA) 200, which was employed by Kwoh et al. in 1985 to perform a brain tumor biopsy1. However, this robot required pre-programming with fixed anatomical landmarks in each patient, and could not be used for dynamic surgical targets. In addition to such pre-programmable robotic surgeons, master-slave robots and robotic endoscope positioning systems have also been developed2,3,4,5. Robot-assisted surgery (RAS) has since become widely used in various surgical specialties, especially cancer surgeries6. The Da Vinci Surgical System (Intuitive Surgical, CA, USA) was launched as a master-slave robot in 19997, and has been employed in over 14 million patients worldwide8. In the field of thoracic surgery, a meta-analysis found that RAS produced short-term and long-term outcomes that were comparable to those of video-assisted surgery in patients who underwent lobectomy or segmentectomy for non-small cell lung cancer9. Although RAS is reported to reduce patient burden and operator stress by enabling precise manipulations with multiple joint motions3,10,11, traditional open or endoscopic surgeries continue to be widely preferred due to RAS’s high equipment costs, need for wide operating spaces, need for additional training and certifications, and lack of tactile sensation12. In particular, the lack of tactile sensation can prolong operative times and increase learning curves, thereby elevating the risk of surgical errors13,14. Moreover, robotic systems can generate forces that exceed tissue tolerance, resulting in tissue damage and potentially serious adverse events15,16.

To compensate for the lack of tactile sensation, the Surgical Assist Robot by Air (Saroa) was developed with a haptic feedback function. This function is achieved through the use of pneumatically driven multi-degree-of-freedom forceps with a rigid-link mechanism and reduced motion interference between joints17. This system facilitates external force estimation by utilizing the high back-drivability of pneumatic cylinders, which allows operators to freely set upper limits to grasping forces and manipulate tissue with appropriate force17. Haptic feedback refers to the real-time re-creation and transmission of forces exerted on surgical instruments to the operating surgeon. To achieve this, it is necessary to quantify the force imparted on the forceps tips, transduce this mechanical load into electrical signals, relay these signals, and re-create the original forceps tip force at the master controller. Although achieving realistic haptic feedback remains a challenge in clinical applications, two principal methodologies have emerged as potential solutions. The first involves embedding miniaturized force sensors directly into surgical instrument tips to quantify localized forces18. The second entails quantifying the forces imparted along the instrument by the robotic arm itself, which can reduce instrument size and lower manufacturing costs relative to the first approach. This technique was pioneered in the Saroa surgical system, which became the first RAS platform integrated with a haptic force feedback mechanism for clinical use17.

Studies have described various RAS-related complications resulting from issues such as patient positioning, robot malfunction, and lack of tactile sensation13,15,19,20,21,22,23,24,25. Our previous study demonstrated that tissue could be grasped with lower forces when using the haptic feedback function of the Saroa surgical system26. The present study investigated the degree of tissue damage for various organs according to grasping force and duration using the Saroa surgical system. Herein, we assess the usefulness of low grasping force for reducing tissue damage, and examine the potential utility of this system.

Materials and methods

Robotic system

Saroa is a pneumatically actuated robotic system developed by Riverfield Inc. (Tokyo, Japan), and was designed to provide real-time haptic feedback to the operating surgeon. Pneumatic cylinders within the robotic arms are utilized to actuate wires housed in each instrument, and these wires are subsequently linked to a fenestrated grasping jaw located at the instrument’s tip (Fig. 1). The grasping force exerted at the jaw is generated by the pressure within these cylinders. This pressure is continuously monitored, with measured values translated into electrical signals that are transmitted to the hand controller; this facilitates real-time control and feedback of force. Saroa accurately estimates grasping forces with errors of less than 20% in comparison to the maximum output forces. Moreover, the practical sensitivity of external force estimation is > 0.5 N. Grasping force is displayed on the console’s sub-monitor and can be adjusted in real time, allowing the operator to vary the load while performing manipulations within the range of applicable force.

Fig. 1

figure 1

Overview of the Saroa surgical system’s feedback mechanism. This system can be grasped with lower forces when the haptic feedback function is engaged.

Full size image

Animal experimental protocol

The study used six adult male beagle dogs weighing 8.06 to 9.08 kg provided from Kitayama Labes Co., Ltd. All animals received humane care in compliance with the “Guide for the Care and Use of Laboratory Animals” published by the US National Institutes of Health (NIH Publication No. 85-3, revised 1985). the study is reported in accordance with ARRIVE guidelines (https://arriveguidelines.org). The experimental protocol was approved by the Animal Experimental Committee of Fukuoka University (Approval No. 2214090). Each dog was sedated with an intramuscular injection of 0.05 mg/kg atropine sulfate, 10 mg/kg ketamine hydrochloride, and 7 mg/kg xylazine before being intubated with an 8-mm endotracheal tube. Anesthesia during mechanical ventilation was maintained using inhalational sevoflurane. After thoracotomy and laparotomy, the fenestrated grasping forceps of the Saroa surgical system were used to grasp the lungs, esophagus, aorta, liver, spleen, small intestine, and large intestine with forces of 1, 2, and 3 N for durations of 1, 2, and 4 min (Fig. 2). The dogs were euthanized deep anesthesia with an overdose of thiopental Sodium. The effects of these different grasping forces on the various tissues were examined. Surgically resected specimens were fixed in 10% formalin (room temperature for 24 h) and embedded in paraffin blocks. Tissue sections were cut at a thickness of 4 mm. Hematoxylin and eosin (H&E) staining, Elastica van Gieson (EVG) staining, and Periodic acid-Schiff (PAS) staining were performed.

Fig. 2

figure 2

Animal experiment. The photograph shows the lung of a beagle dog being grasped by the fenestrated grasping forceps of the Saroa surgical system.

Full size image

Histological evaluation of tissue damage

Tissue damage was assessed under a microscope by two pathologists. The lungs were examined for the presence or absence of hemorrhage and congestion, and changes in alveoli, pleural elastic layers, and collagen layers. The liver and spleen were examined for the presence or absence of hemorrhage and congestion, and degeneration. The gastrointestinal tract was examined for the presence or absence of hemorrhage and congestion, and mucosal erosion. Aortic tissue was examined for the presence or absence of hemorrhage and congestion, and changes in elastic layers.

Measurement of areas of tissue damage

The areas of tissue damage were imaged using microscope and measured using BX-analyzer (BZ-X700, KEYENCE, Osaka, Japan). Organ parenchyma was recognized in yellow, erythrocytes in blue, pleural elastic layer in blue, and pleural collagen layer in blue. For lung tissue, we measured the area of hemorrhage and/ or congestion, red blood cell (RBC) count, and thickness of the pleural elastic layers and collagen layers. For liver and spleen tissue, we measured the area of hemorrhage and/ or congestion, RBC count, and area of degeneration. For gastrointestinal tract tissue, we measured the area of hemorrhage and/ or congestion, RBC count, and area of mucosal erosion. For aortic tissue, we measured the RBC count and elastic layer thickness in the intima, media, and adventitia.

Statistical analysis

Linear regression analyses were conducted to evaluate the associations between tissue damage and grasping forces. The regression models were constructed with tissue damage as the dependent variable and grasping force as the independent variable. The associations were evaluated by calculating standardized regression coefficients (β) and R2 value. Statistical analyses were performed using JMP version 14.0 statistical software package for Windows (SAS Institute, NC, USA). A p-value of less than 0.05 was considered statistically significant. In addition, the post hoc power analysis was performed to evaluate the adequacy of the sample size. A power exceeding 0.8 was considered indicative of a sufficient sample size.

Results

Macroscopic findings

Macroscopic examination of the lung and liver revealed hemorrhage and congestion in grasped areas that increased with higher grasping forces (Fig. 3), but not with longer grasping durations. No macroscopic differences were detected in the esophagus, aorta, spleen, small intestine, or large intestine at the different grasping forces. In addition, no major differences in macroscopic findings were noted among the different grasping durations in all organs.

Fig. 3

figure 3

Macroscopic findings after grasping the (A) Lung and (B) Liver. The grasped areas of the lung and liver show hemorrhage and congestion that increased at higher grasping forces.

Full size image

Histological evaluation of tissue damage

Figure 4 shows the histological changes in the lung and liver after the grasping experiments. The lung tissue exhibited hemorrhage and congestion in the grasped areas (Fig. 4A). No morphological irregularities were detected in the alveoli, pleural elastic layers, or collagen layers; however, the thickness of the pleural elastic and collagen layers increased in the grasped areas at higher grasping forces (Fig. 4B). Similarly, the liver tissue exhibited hemorrhage, congestion, and degenerative changes in the grasped areas (Fig. 4C). The degenerative changes included hepatocyte atrophy, impaired adhesion, reduced eosinophilic granularity, pyknosis, and karyorrhexis. A reduction in glycogen was also observed in the grasped areas of the liver (Fig. 4D). No histological changes were observed in the esophagus, aorta, spleen, small intestine, or large intestine at the different grasping forces (Supplementary Fig. 1). Furthermore, no major differences in histological findings were noted among the different grasping durations in all organs.

Fig. 4

figure 4

Histological images of grasped areas in the lung (A, B) and liver (C, D). A Lung tissue shows hemorrhage and congestion in the grasped areas (H&E staining). B Thickness of the pleural elastic and collagen layers increased after grasping (EVG staining). C Left half (grasped area) of the liver tissue shows hemorrhage, congestion, and hepatocyte degeneration (H&E staining). D The same grasped area in the left half of C shows reduced glycogen (PAS staining).

Full size image

Associations between lung tissue damage and grasping forces

In grasped lung tissue, the area of hemorrhage and/ or congestion, and RBC count were observed to increase with higher grasping forces (Fig. 5). In addition, the thickness of the pleural elastic layers and collagen layers increased slightly at higher grasping forces. The regression analysis showed that grasping force was significantly and positively associated (p < 0.0001) with lung tissue damage (Fig. 6). Specifically, grasping force was strongly associated with the area of hemorrhage and/ or congestion (β = 0.8015, R2 = 0.6424), and RBC count (β = 0.8538, R2 = 0.7291), and weakly associated with the thickness of the pleural elastic layers (β = 0.4260, R2 = 0.1815) and collagen layers (β = 0.3482, R2 = 0.1213). Post hoc power analysis for the sample size of lung tissue damage, all of parameter were confirmed to exceed 0.8.

Fig. 5

figure 5

Histological images of grasped areas in the lung. A–D EVG staining. E–H Enhanced images of RBCs. Blue: RBCs, Yellow: lung parenchyma. I–L Enhanced images of pleural elastic/collagen layers. Blue: elastic layer, Red: collagen layer, Yellow: lung parenchyma.

Full size image

Fig. 6

figure 6

Results of linear regression analysis of lung tissue damage at different grasping forces. The graphs show the associations between grasping forces and A Area of hemorrhage and/ or congestion, B RBC count, C Pleural elastic layer thickness, and D Pleural collagen layer thickness.

Full size image

Associations between liver tissue damage and grasping forces

In grasped liver tissue, the area of hemorrhage and/ or congestion, RBC count, and area of hepatocyte degeneration were observed to increase with higher grasping forces (Fig. 7). The regression analysis showed that grasping force was significantly and positively associated (p < 0.0001) with liver tissue damage (Fig. 8). Specifically, grasping force was strongly associated with the area of hemorrhage and/ or congestion (β = 0.9733, R2 = 0.9475), RBC count (β = 0.9622, R2 = 0.9260), and area of degeneration (β = 9153, R2 = 0.8378). Post hoc power analysis for the sample size of liver tissue damage, all of parameter were confirmed to exceed 0.8.

Fig. 7

figure 7

Histological images of grasped areas in the liver. A–D Loupe images in which the dotted lines indicate areas of tissue degenerative change that became more severe as grasping force increased (H&E staining). E–H Low-power magnification of the areas shown in A to D. F to H correspond to the areas within the dotted lines shown in B to D. I–L Enhanced images of RBCs. Blue: RBCs, Yellow: liver parenchyma.

Full size image

Fig. 8

figure 8

Results of linear regression analysis of liver tissue damage at different grasping forces. The graphs show the associations between grasping forces and (A) Area of hemorrhage and/ or congestion, (B) RBC count, and (C) Area of degeneration.

Full size image

Discussion

RAS has been employed in various surgical fields, and the number of RAS cases has steadily increased6,8. The advantages of RAS include 3D visualization, elimination of the fulcrum effect and physiological tremors, improvements in ergonomic position, as well as the abilities to scale movements into micromotions, construct microanastomoses, and perform telesurgery12. However, these advantages are offset by the absence of tactile sensation, high start-up and running costs, and requirements for additional staff. Furthermore, RAS-associated organ damage can occur due to complications such as incorrect patient positioning, robot malfunction, and lack of tactile sensation25. While surgeons may attempt to overcome the lack of tactile sensation through visual cues and other techniques, these abilities are likely to vary among individuals. As a result, organ damage can occur during RAS when tissue is grasped with excessive force16. We have previously reported that tissue can be grasped with lower forces during surgical operations using the haptic feedback function of the Saroa surgical system, and that operators can adjust the grasping force as desired to ensure smoothly operation26. In that study, we demonstrated that the grasping force exerted on a pig lung was significantly reduced when haptic feedback was engaged. In addition, we showed that procedures such as interlobar dissection, vascular dissection, bronchial dissection, and lymph node dissection could be safely performed with a grasping force of ≤ 3 N. On the other hand, Saroa have haptic feedback of grasping forces, Da Vinci 5 and Senhance have feedback functions of propulsive and traction force, and the reduction of complications is expected even in surgeries using for these functions27,28. In addition, not only robotic surgery but also laparoscopic studies are being conducted to introduce tactile feedback and contribute to the reduction of complications29.

Our present study found significant associations between grasping force and tissue damage in the lung and liver. Although there were associations between grasping force and the thickness of the pleural elastic and collagen layers, the standardized regression coefficients and R2 values were small and the associations were weak. This may have been due to the inherent thinness of these layers, resulting in a small range of values.

The liver showed an increasing degree of tissue damage, such as hemorrhage and congestion, and degenerative changes, as grasping force increased. Liver injuries can usually be repaired as it is a regenerative organ, and hepatocytes have well-established mechanisms of regeneration when they have normal proliferative capacity. However, severe acute or chronic liver injury can impair the organ’s regenerative capabilities. Impaired native hepatocyte regeneration may result in fibrosis, transdifferentiation, and malignant transformation due to DNA damage30,31. In order to promote the normal regeneration of hepatocytes, it is important to perform surgeries with the minimum necessary amount of grasping force.

Our experiments did not detect any notable tissue damage in the esophagus, aorta, spleen, small intestine, or large intestine at different grasping forces, which may be due to the mechanical strength of these tissues. For example, the gastrointestinal tract is thought to have robust tissues because of its well-developed muscular layer, the spleen is maintained in its tissue architecture by the splenic column, and the aorta is rich in elastic fibers. These characteristics may explain the lack of differences in the degree of tissue damage at grasping forces ≤ 3 N. Since Saroa grasps with a force of ≤ 3 N when the haptic feedback function is engaged, we consider that this system can be safely used to operate on these organs.

A limitation of this study is that we were unable to perform subsequent analyses of the damaged tissue. This was because closing the chest after the initial experiment and performing re-thoracotomy would be highly invasive and traumatic to the animal. Accordingly, this was not allowed by the Animal Experimental Committee with consideration to animal welfare.

Our study showed that lung and liver tissue damage occurred even at grasping forces ≤ 3 N, and that higher grasping forces resulted in more severe tissue damage. On the other hand, the haptic feedback function of the Saroa surgical system was shown to enable grasping at a reduced force. These functions could contribute to reduce intraoperative organ damage during lung and liver surgeries.

Data availability

The datasets generated during and/ or analysed during the current study are available from the corresponding author on reasonable request.

Abbreviations

EVG:

Elastica van Gieson

H&E:

Hematoxylin and eosin

PAS:

Periodic acid-Schiff

RAS:

Robot-assisted surgery

RBC:

Red blood cell

Saroa:

Surgical Assist Robot by Air

References

Kwoh, Y. S., Hou, J., Jonckheere, E. A. & Hayati, S. A robot with improved absolute positing accuracy for CT guided stereotactic brain surgery. IEEE Trans. Boimed Eng. 35, 153–160 (1988).

CASGoogle Scholar

Diana, M. & Marescaux, J. Robotic surgery. Br. J. Surg. 102, e15–e28 (2015).

CASPubMedMATHGoogle Scholar

Leal Ghezzi, T. & Campos Corleta, O. 30 years of robotic surgery. World J. Surg. 40, 2550–2557 (2016).

PubMedMATHGoogle Scholar

Lane, T. A short history of robotic surgery. Ann. R Coll. Surg. Engl. 100, 5–7 (2018).

PubMedPubMed CentralMATHGoogle Scholar

Marescaux, J. & Seeliger, B. Robotic surgery: A time of change. Update Surg. 75, 793–794 (2023).

MATHGoogle Scholar

Hoeppner, J. Robotic cancer surgery. Cancers (Basel) 13, 3 (2021).

Google Scholar

Guthart, G. S. & Salisbury, J. K. The Intuitive/sup TM/telesurgery system: Overview and application. In Proceedings. Millennium Conference. IEEE International Conference on Robotics and Automation 2000 618–621. (2000).

Intuitive Surgical. Annual Report 3–4 (Sunnyvale (CA)). (2023).

Ma, J. et al. Robot-assisted thoracic surgery versus video-assisted thoracic surgery for lung lobectomy or segmentectomy in patients with non-small cell lung cancer: A meta-analysis. BMC Cancer. 21, 498 (2021).

PubMedPubMed CentralMATHGoogle Scholar

Cerfolio, R., Louie, B. E., Farivar, A. S., Onaitis, M. & Park, B. J. Consensus statement on definitions and nomenclature for robotic thoracic surgery. J. Thorac. Cardiovasc. Surg. 154, 1065–1069 (2017).

PubMedGoogle Scholar

Du, H. & Li, H. Robotic thoracic surgery: S1 + 2 segmentectomy of the left upper lobe: Advantage of robotic assisted thoracic surgery. J. Thorac. Dis. 9, E973 (2017).

PubMedPubMed CentralGoogle Scholar

Lanfranco, A. R., Castellanos, A. E., Desai, J. P. & Meyers, W. C. Robotic surgery: A current perspective. Ann. Surg. 239, 14–21 (2004).

PubMedPubMed CentralMATHGoogle Scholar

Toledo, L., Gossot, D., Fritsch, S., Revillon, Y. & Reboulet, C. Study of sustained forces and the working space of endoscopic surgery instruments. Ann. Chir. 53, 587–597 (1999).

CASPubMedGoogle Scholar

Karponis, D., Koya, Y., Miyazaki, R., Kanno, T. & Kawashima, K. Evaluation of a pneumatic surgical robot with dynamic force feedback. J. Robot Surg. 13, 413–421 (2019).

PubMedGoogle Scholar

Trejos, A. L., Patel, R. V. & Naish, M. D. Force sensing and its application in minimally invasive surgery and therapy: A survey. Proc. Inst. Mech. Eng. C. 224, 1435–1454 (2010).

MATHGoogle Scholar

Mucksavage, P. et al. Differences in grip forces among various robotic instruments and Da Vinci surgical platforms. J. Endourol. 25, 523–528 (2011).

PubMedGoogle Scholar

Zhou, D., Tadano, K. & Haraguchi, D. Motion control and external force estimation of a pneumatically driven multi-dof robotic forceps. Appl. Sci. 10, 3679 (2020).

CASMATHGoogle Scholar

Wottawa, C. R. et al. Evaluating tactile feedback in robotic surgery for potential clinical application using an animal model. Surg. Endosc. 30, 3198–3209 (2016).

PubMedMATHGoogle Scholar

Chen, C. H., Chen, H. H. & Liu, W. M. Complication reports for robotic surgery using three arms by a single surgeon at a single institution. J. Minim. Access. Surg. 13, 22–28 (2017).

PubMedPubMed CentralMATHGoogle Scholar

Maerz, D. A., Beck, L. N., Sim, A. J. & Gainsburg, D. M. Complications of robotic-assisted laparoscopic surgery distant from the surgical site. Br. J. Anaesth. 118, 492–503 (2017).

CASPubMedGoogle Scholar

Ackerman, S. J. et al. Comparison of complication and conversion rates between robotic-assisted and laparoscopic rectal resection for rectal cancer: Which patients and providers could benefit most from robotic-assisted surgery? J. Med. Econ. 21, 254–261 (2018).

PubMedGoogle Scholar

Kapur, A. & Kapur, V. Robotic surgery: Anaesthesiologist’s contemplation. Malays J. Med. Sci. 27, 143–149 (2020).

PubMedPubMed CentralMATHGoogle Scholar

Fotiou, A. & Iavazzo, C. Gynecologic robotic surgery: Intraoperative complication and conversion rates. J. Invest. Surg. 35, 916–917 (2022).

PubMedGoogle Scholar

Buli, B., Diriba, T., Tesfaye, B. & M. & Magnitude of position-related soft tissue injuries and associated factors among elective adult surgical patients at Tikur Anbessa specialized hospital, addis Ababa, Ethiopia. Ann. Med. Surg. (Lond). 82, 104592 (2022).

PubMedGoogle Scholar

Jara, R. D., Guerrón, A. D. & Portenier, D. Complications of robotic surgery. Surg. Clin. North. Am. 100, 461–468 (2020).

PubMedMATHGoogle Scholar

Ueda, Y. et al. Impact of a pneumatic surgical robot with haptic feedback function on surgical manipulation. Sci. Rep. 13, 22615 (2023).

ADSCASPubMedPubMed CentralMATHGoogle Scholar

Gamal, A. et al. Comparing the technological and intraoperative performances of Da Vinci Xi and davinci 5 robotic platforms in patients undergoing robotic-assisted radical prostatectomy. Eur. Urol. Open. Sci. 69, 1–4 (2024).

PubMedPubMed CentralMATHGoogle Scholar

Fujii, T. et al. Comparison of short- and mid-term outcomes between the Senhance digital laparoscopic system and laparoscopic colectomy: A propensity score matching study. Surg. Endosc. 39, 1153–1159 (2025).

PubMedMATHGoogle Scholar

Hao, Y. et al. Development of force sensing techniques for robot-assisted laparoscopic surgery: A review. IEEE Trans. Med. Rob. Bionics. 6, 868–887 (2024).

MATHGoogle Scholar

Gadd, V. L., Aleksieva, N. & Forbes, S. J. Epithelial plasticity during liver injury and regeneration. Cell. Stem Cell. 27, 557–573 (2020).

CASPubMedGoogle Scholar

Deng, X. et al. Chronic liver injury induces conversion of biliary epithelial cells into hepatocytes. Cell Stem Cell 23 114–122.e3 e113 (2018).

CASPubMedMATHGoogle Scholar

Download references

Acknowledgements

The authors thank Yasushi Tanaka of Riverfield Inc. for contributing to the analysis and assisting in the preparation of the manuscript.

Funding

This work was partly supported by a fund for collaborative research between Fukuoka University and Riverfield Inc.

Author information

Authors and Affiliations

Department of Thoracic, Breast and Endocrine, and Pediatric Surgery, Fukuoka University School of Medicine, 7-45-1 Nanakuma, Jonan-ku, Fukuoka, 814-0180, Japan

Hiroyasu Nakashima, Yuichiro Ueda, Yoko Miyanari, So Miyahara, Keita Tokuishi, Ryuichi Waseda, Takeshi Shiraishi & Toshihiko Sato

Riverfield Inc., Tokyo, Japan

Teruyuki Nishihara

Department of Pathology, Fukuoka University School of Medicine, Fukuoka, Japan

Makoto Hamasaki

Department of Pathology, Kitasato University Kitasato Institute Hospital, Tokyo, Japan

Makoto Ohbu

Department of Information Physics and Computing, School of Information Science and Technology, The University of Tokyo, Tokyo, Japan

Kenji Kawashima

Department of Medical Statistics, Satista Co., Ltd., Kyoto, Japan

Hajime Yamakage

Authors

Hiroyasu Nakashima

View author publications