Hisashi Suzuki
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Hisashi Suzuki has a total of 47 co-authors. Here are the top co-authors by number of shared publications:
S. Arimoto
11 shared publicationsSuguru Arimoto
Faculty of Engineering, The University of Tokyo
8 shared publications- Atsushi Nishikawa
Graduate School of Engineering Science, Osaka University, Toyonaka, Japan
2 shared publications H. Onishi
2 shared publicationsHiroyuki Onishi
Glory Ltd. (Japan)
2 shared publicationsKin-Ya Sugimoto
Graduate School of Science and Engineering, Chuo University, Bunkyo-ku, Tokyo 112-8551, Japan
2 shared publications
Latest Publications
- Open Access
- 1 December 2025
- Published by AME Publishing Company in Quantitative Imaging in Medicine and Surgery
- Vol. 15 (12) , 12386-12397
Surgical instrument segmentation and classification in transcanal endoscopic ear surgery video using Segment Anything Model 2
Background: Accurate automatic recognition of surgical instruments is important for enhancing the safety and efficiency of transcanal endoscopic ear surgery (TEES). Segment Anything Model 2 (SAM 2) is a state-of-the-art artificial intelligence (AI) model featuring “zero-shot segmentation”, which allows it to identify objects without extensive training data. The purpose of this study was to be the first to apply SAM 2 to TEES videos to evaluate its performance in segmenting and classifying surgical instruments. Methods: From videos of three clinical TEES cases (tympanoplasty), we created a 684-frame evaluation video that simulated the sequential use of four types of surgical instruments: cupped forceps, pick, alligator forceps, and circular knife. SAM 2 was instructed to identify the instruments using only 12 reference images. Performance was quantitatively evaluated using four metrics—Precision, Recall, dice similarity coefficient (DSC), and intersection over union (IoU)—as well as a confusion matrix to assess frame-level classification accuracy by comparing the model’s output to manually created ground-truth annotations. Results: For segmentation without classification, SAM 2 demonstrated excellent performance, with a mean Precision of 0.98, a mean Recall of 0.89, a mean DSC of 0.93, and a mean IoU of 0.87. When classifying the four specific instrument types, performance varied, with mean Precision ranging from 0.68 to 0.98, mean Recall from 0.79 to 0.86, mean DSC from 0.62 to 0.89, and mean IoU from 0.58 to 0.84. Notably, false-positive detections in non-instrument regions were extremely rare. A trend of lower classification accuracy was observed for articulating instruments with moving parts (cupped forceps and alligator forceps). Conclusions: This study demonstrated that SAM 2 can accurately identify surgical instruments in TEES videos without the need to prepare large, costly training datasets. This zero-shot segmentation characteristic significantly lowers the barrier to clinical implementation of AI, suggesting it is a promising foundational technology for future surgical navigation and robotic automation. Key future challenges include improving classification accuracy for articulating instruments and accelerating processing speed for practical, real-time use.
- 1 December 2025
- Published by AME Publishing Company in Quantitative Imaging in Medicine and Surgery
- Vol. 15 (12) , 12386-12397
Open AccessSurgical instrument segmentation and classification in transcanal endoscopic ear surgery video using Segment Anything Model 2
Background: Accurate automatic recognition of surgical instruments is important for enhancing the safety and efficiency of transcanal endoscopic ear surgery (TEES). Segment Anything Model 2 (SAM 2) is a state-of-the-art artificial intelligence (AI) model featuring “zero-shot segmentation”, which allows it to identify objects without extensive training data. The purpose of this study was to be the first to apply SAM 2 to TEES videos to evaluate its performance in segmenting and classifying surgical instruments. Methods: From videos of three clinical TEES cases (tympanoplasty), we created a 684-frame evaluation video that simulated the sequential use of four types of surgical instruments: cupped forceps, pick, alligator forceps, and circular knife. SAM 2 was instructed to identify the instruments using only 12 reference images. Performance was quantitatively evaluated using four metrics—Precision, Recall, dice similarity coefficient (DSC), and intersection over union (IoU)—as well as a confusion matrix to assess frame-level classification accuracy by comparing the model’s output to manually created ground-truth annotations. Results: For segmentation without classification, SAM 2 demonstrated excellent performance, with a mean Precision of 0.98, a mean Recall of 0.89, a mean DSC of 0.93, and a mean IoU of 0.87. When classifying the four specific instrument types, performance varied, with mean Precision ranging from 0.68 to 0.98, mean Recall from 0.79 to 0.86, mean DSC from 0.62 to 0.89, and mean IoU from 0.58 to 0.84. Notably, false-positive detections in non-instrument regions were extremely rare. A trend of lower classification accuracy was observed for articulating instruments with moving parts (cupped forceps and alligator forceps). Conclusions: This study demonstrated that SAM 2 can accurately identify surgical instruments in TEES videos without the need to prepare large, costly training datasets. This zero-shot segmentation characteristic significantly lowers the barrier to clinical implementation of AI, suggesting it is a promising foundational technology for future surgical navigation and robotic automation. Key future challenges include improving classification accuracy for articulating instruments and accelerating processing speed for practical, real-time use.
- 17 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-7
Algebraic Reproduction of Triage Decision Making on Simple Heuristics of Boolean Multivalued Logic
In this research, we consider a simplified disaster situation for a thought experiment where priorities must be assigned when treating victims. We model egalitarian decision making, utilitarian decision making, and triage decision making (which combines both egalitarian and utilitarian decision makings) using a quantum computing-like multivalued logic that forms a Boolean algebra in contrast to popular multivalued logics that lack algebraic consistency. As a result, this article reports that some simple heuristic rules can quantitatively and algebraically reproduce and control human decisions regarding the prioritization of treatment for victims that reflects bioethical considerations.
- 17 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-7
Algebraic Reproduction of Triage Decision Making on Simple Heuristics of Boolean Multivalued Logic
In this research, we consider a simplified disaster situation for a thought experiment where priorities must be assigned when treating victims. We model egalitarian decision making, utilitarian decision making, and triage decision making (which combines both egalitarian and utilitarian decision makings) using a quantum computing-like multivalued logic that forms a Boolean algebra in contrast to popular multivalued logics that lack algebraic consistency. As a result, this article reports that some simple heuristic rules can quantitatively and algebraically reproduce and control human decisions regarding the prioritization of treatment for victims that reflects bioethical considerations.
- 1 September 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 14-18
Experimental development of a multi-view stereo endoscope system
This paper reports on developing experimentally a multi-view stereo endoscope system that reconstructs once 3D data from each stereo image handled in the medical field, enhances the 3D data, and renders a set of multi-view images being available to an arbitrary multi-view glasses-free 3D display in contrast to a conventional stereo endoscope system being unavailable to multi-view displays. The developed system is similar to a CT reconstructing 3D data but is different in the respects that the former inputs one-shot stereo images with visible light to observe the surface of object in real-time while the latter inputs multidirectional projections with X-ray to observe the inside of object in non-real-time.
- 1 September 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 14-18
Experimental development of a multi-view stereo endoscope system
This paper reports on developing experimentally a multi-view stereo endoscope system that reconstructs once 3D data from each stereo image handled in the medical field, enhances the 3D data, and renders a set of multi-view images being available to an arbitrary multi-view glasses-free 3D display in contrast to a conventional stereo endoscope system being unavailable to multi-view displays. The developed system is similar to a CT reconstructing 3D data but is different in the respects that the former inputs one-shot stereo images with visible light to observe the surface of object in real-time while the latter inputs multidirectional projections with X-ray to observe the inside of object in non-real-time.
- 14 August 2015
- Published by Taylor & Francis in Neurocase
- Vol. 22 (2) , 220-224
Wandering behavior of a severely demented patient with frontotemporal dementia
Although wandering is a common behavioral problem in patients with dementia, information about wandering in patients with severe dementia is scarce. This study analyzed data about the wandering behavior of a 77-year-old woman with severe frontotemporal dementia. Due to severe atrophy of the anterior cortex, she was totally unable to act spontaneously. Nevertheless, she walked with her husband every day. Her walking pattern varied on a daily basis but it appeared to be random. However, a mathematical analysis showed that her walking pattern was not random. Our case suggests that even patients with severe atrophy in the anterior cortex may be able to orient themselves within their neighborhood if their posterior cortex is preserved.
- 14 August 2015
- Published by Taylor & Francis in Neurocase
- Vol. 22 (2) , 220-224
Wandering behavior of a severely demented patient with frontotemporal dementia
Although wandering is a common behavioral problem in patients with dementia, information about wandering in patients with severe dementia is scarce. This study analyzed data about the wandering behavior of a 77-year-old woman with severe frontotemporal dementia. Due to severe atrophy of the anterior cortex, she was totally unable to act spontaneously. Nevertheless, she walked with her husband every day. Her walking pattern varied on a daily basis but it appeared to be random. However, a mathematical analysis showed that her walking pattern was not random. Our case suggests that even patients with severe atrophy in the anterior cortex may be able to orient themselves within their neighborhood if their posterior cortex is preserved.
- 1 January 2012
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 370-375
Boolean complex logic
This article, by describing atoms and logic formulae with complex functions, defines a new multivalued logic such that the set of logic formulae forms a Boolean algebra in contrast to popular multivalued logics such as fuzzy logics, which we call a Boolean complex logic. This article shows also some intuitively understandable examples demonstrating that the proposed logic can process deductive inferences syntactically by evaluating ranges of truth values.
- 1 January 2012
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 370-375
Boolean complex logic
This article, by describing atoms and logic formulae with complex functions, defines a new multivalued logic such that the set of logic formulae forms a Boolean algebra in contrast to popular multivalued logics such as fuzzy logics, which we call a Boolean complex logic. This article shows also some intuitively understandable examples demonstrating that the proposed logic can process deductive inferences syntactically by evaluating ranges of truth values.
- 1 August 2011
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 109-114
New Boolean Multivalued Logic System Simplifying Inferences in Flexible Styles
This article, by regarding the relative number of affirmative bits on an arbitrary finite sequence of bits as the truth value, defines a new multivalued logic system such that the set of logic formulae forms a Boolean algebra in contrast to other non-Boolean multivalued logic systems, which we call a Boolean multivalued logic system. This article also, for demonstrating compactly that the Boolean properties (in contrast to other multivalued logic systems) simplify multivalued inferences in flexible styles, shows several intuitively-understandable examples tracing biotic diversity such that, on the proposed logic system, we can easily handle semantic inferences verifying values of all bits of each logic formula, inductive inferences counting the relative numbers of affirmative bits for determination of truth values, and deductive inferences syntactically narrowing ranges of truth values.
- 1 August 2011
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 109-114
New Boolean Multivalued Logic System Simplifying Inferences in Flexible Styles
This article, by regarding the relative number of affirmative bits on an arbitrary finite sequence of bits as the truth value, defines a new multivalued logic system such that the set of logic formulae forms a Boolean algebra in contrast to other non-Boolean multivalued logic systems, which we call a Boolean multivalued logic system. This article also, for demonstrating compactly that the Boolean properties (in contrast to other multivalued logic systems) simplify multivalued inferences in flexible styles, shows several intuitively-understandable examples tracing biotic diversity such that, on the proposed logic system, we can easily handle semantic inferences verifying values of all bits of each logic formula, inductive inferences counting the relative numbers of affirmative bits for determination of truth values, and deductive inferences syntactically narrowing ranges of truth values.
- 1 August 2009
- Published by Schweizerbart in Neues Jahrbuch für Geologie und Paläontologie - Abhandlungen
- Vol. 253 (2-3) , 293-311
Upper Triassic and Middle Jurassic radiolarians from the ophiolitic melange of the Dinaridic Ophiolite Belt, SW Serbia
- 1 August 2009
- Published by Schweizerbart in Neues Jahrbuch für Geologie und Paläontologie - Abhandlungen
- Vol. 253 (2-3) , 293-311
Upper Triassic and Middle Jurassic radiolarians from the ophiolitic melange of the Dinaridic Ophiolite Belt, SW Serbia
- 1 September 2008
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 248-253
Path planning in an unknown environment on the basis of observations of occluded areas
In order to get to a destination in an unknown environment, one must make efficient observations of the environment while moving toward the destination. This report begins with a classification of the conditions under which an observerpsilas view is blocked (occlusion). Once the agent has classified the visible occlusions in its environment, it determines the next point from which to carry out observations. A method is proposed here for planning the path to a given destination by planning a path to successive observation points. In doing so, the proposed path to the destination passes through a series of observation points. A simple algorithm is used to determine the location of the next observation point. The method allows planning calculations to be performed in real time and also provides sufficient flexibility for mapping of obstacles with curved surfaces. This search scheme also separates and alternates between the tasks of moving and taking observations, which facilitates the programming of real robots to use this algorithm. Simulations of operations using this algorithm in a variety of environments are described in this report.
- 1 September 2008
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 248-253
Path planning in an unknown environment on the basis of observations of occluded areas
In order to get to a destination in an unknown environment, one must make efficient observations of the environment while moving toward the destination. This report begins with a classification of the conditions under which an observerpsilas view is blocked (occlusion). Once the agent has classified the visible occlusions in its environment, it determines the next point from which to carry out observations. A method is proposed here for planning the path to a given destination by planning a path to successive observation points. In doing so, the proposed path to the destination passes through a series of observation points. A simple algorithm is used to determine the location of the next observation point. The method allows planning calculations to be performed in real time and also provides sufficient flexibility for mapping of obstacles with curved surfaces. This search scheme also separates and alternates between the tasks of moving and taking observations, which facilitates the programming of real robots to use this algorithm. Simulations of operations using this algorithm in a variety of environments are described in this report.
- 1 July 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 270-275
Curved Voronoi Diagrams Consisting of Influence Areas with Differentiable Boundaries
In this paper, we incorporate the concept of differentiability into Voronoi diagrams. We consider the influence area diagrams consisting of areas with differentiable boundaries. The diagram is called a curved Voronoi diagram. Any level of differentiability can be used in the definition of the curved Voronoi diagram depending on its applications. One of the applications is path planning for car-like robots in an area with obstacles. Carlike robots can easily drive along a path consisting of clothoid curve segments, circular arcs, and line segments. Therefore, we discuss a method for constructing a clothoid curve, circle, and line (CCL) diagram.
- 1 July 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 270-275
Curved Voronoi Diagrams Consisting of Influence Areas with Differentiable Boundaries
In this paper, we incorporate the concept of differentiability into Voronoi diagrams. We consider the influence area diagrams consisting of areas with differentiable boundaries. The diagram is called a curved Voronoi diagram. Any level of differentiability can be used in the definition of the curved Voronoi diagram depending on its applications. One of the applications is path planning for car-like robots in an area with obstacles. Carlike robots can easily drive along a path consisting of clothoid curve segments, circular arcs, and line segments. Therefore, we discuss a method for constructing a clothoid curve, circle, and line (CCL) diagram.