-
Can Physician Judgment Enhance Model Trustworthiness? A Case Study on Predicting Pathological Lymph Nodes in Rectal Cancer
Authors:
Kazuma Kobayashi,
Yasuyuki Takamizawa,
Mototaka Miyake,
Sono Ito,
Lin Gu,
Tatsuya Nakatsuka,
Yu Akagi,
Tatsuya Harada,
Yukihide Kanemitsu,
Ryuji Hamamoto
Abstract:
Explainability is key to enhancing artificial intelligence's trustworthiness in medicine. However, several issues remain concerning the actual benefit of explainable models for clinical decision-making. Firstly, there is a lack of consensus on an evaluation framework for quantitatively assessing the practical benefits that effective explainability should provide to practitioners. Secondly, physici…
▽ More
Explainability is key to enhancing artificial intelligence's trustworthiness in medicine. However, several issues remain concerning the actual benefit of explainable models for clinical decision-making. Firstly, there is a lack of consensus on an evaluation framework for quantitatively assessing the practical benefits that effective explainability should provide to practitioners. Secondly, physician-centered evaluations of explainability are limited. Thirdly, the utility of built-in attention mechanisms in transformer-based models as an explainability technique is unclear. We hypothesize that superior attention maps should align with the information that physicians focus on, potentially reducing prediction uncertainty and increasing model reliability. We employed a multimodal transformer to predict lymph node metastasis in rectal cancer using clinical data and magnetic resonance imaging, exploring how well attention maps, visualized through a state-of-the-art technique, can achieve agreement with physician understanding. We estimated the model's uncertainty using meta-level information like prediction probability variance and quantified agreement. Our assessment of whether this agreement reduces uncertainty found no significant effect. In conclusion, this case study did not confirm the anticipated benefit of attention maps in enhancing model reliability. Superficial explanations could do more harm than good by misleading physicians into relying on uncertain predictions, suggesting that the current state of attention mechanisms in explainability should not be overestimated. Identifying explainability mechanisms truly beneficial for clinical decision-making remains essential.
△ Less
Submitted 14 December, 2023;
originally announced December 2023.
-
Sketch-based Medical Image Retrieval
Authors:
Kazuma Kobayashi,
Lin Gu,
Ryuichiro Hataya,
Takaaki Mizuno,
Mototaka Miyake,
Hirokazu Watanabe,
Masamichi Takahashi,
Yasuyuki Takamizawa,
Yukihiro Yoshida,
Satoshi Nakamura,
Nobuji Kouno,
Amina Bolatkan,
Yusuke Kurose,
Tatsuya Harada,
Ryuji Hamamoto
Abstract:
The amount of medical images stored in hospitals is increasing faster than ever; however, utilizing the accumulated medical images has been limited. This is because existing content-based medical image retrieval (CBMIR) systems usually require example images to construct query vectors; nevertheless, example images cannot always be prepared. Besides, there can be images with rare characteristics th…
▽ More
The amount of medical images stored in hospitals is increasing faster than ever; however, utilizing the accumulated medical images has been limited. This is because existing content-based medical image retrieval (CBMIR) systems usually require example images to construct query vectors; nevertheless, example images cannot always be prepared. Besides, there can be images with rare characteristics that make it difficult to find similar example images, which we call isolated samples. Here, we introduce a novel sketch-based medical image retrieval (SBMIR) system that enables users to find images of interest without example images. The key idea lies in feature decomposition of medical images, whereby the entire feature of a medical image can be decomposed into and reconstructed from normal and abnormal features. By extending this idea, our SBMIR system provides an easy-to-use two-step graphical user interface: users first select a template image to specify a normal feature and then draw a semantic sketch of the disease on the template image to represent an abnormal feature. Subsequently, it integrates the two kinds of input to construct a query vector and retrieves reference images with the closest reference vectors. Using two datasets, ten healthcare professionals with various clinical backgrounds participated in the user test for evaluation. As a result, our SBMIR system enabled users to overcome previous challenges, including image retrieval based on fine-grained image characteristics, image retrieval without example images, and image retrieval for isolated samples. Our SBMIR system achieves flexible medical image retrieval on demand, thereby expanding the utility of medical image databases.
△ Less
Submitted 6 March, 2023;
originally announced March 2023.
-
Decomposing Normal and Abnormal Features of Medical Images into Discrete Latent Codes for Content-Based Image Retrieval
Authors:
Kazuma Kobayashi,
Ryuichiro Hataya,
Yusuke Kurose,
Mototaka Miyake,
Masamichi Takahashi,
Akiko Nakagawa,
Tatsuya Harada,
Ryuji Hamamoto
Abstract:
In medical imaging, the characteristics purely derived from a disease should reflect the extent to which abnormal findings deviate from the normal features. Indeed, physicians often need corresponding images without abnormal findings of interest or, conversely, images that contain similar abnormal findings regardless of normal anatomical context. This is called comparative diagnostic reading of me…
▽ More
In medical imaging, the characteristics purely derived from a disease should reflect the extent to which abnormal findings deviate from the normal features. Indeed, physicians often need corresponding images without abnormal findings of interest or, conversely, images that contain similar abnormal findings regardless of normal anatomical context. This is called comparative diagnostic reading of medical images, which is essential for a correct diagnosis. To support comparative diagnostic reading, content-based image retrieval (CBIR), which can selectively utilize normal and abnormal features in medical images as two separable semantic components, will be useful. Therefore, we propose a neural network architecture to decompose the semantic components of medical images into two latent codes: normal anatomy code and abnormal anatomy code. The normal anatomy code represents normal anatomies that should have existed if the sample is healthy, whereas the abnormal anatomy code attributes to abnormal changes that reflect deviation from the normal baseline. These latent codes are discretized through vector quantization to enable binary hashing, which can reduce the computational burden at the time of similarity search. By calculating the similarity based on either normal or abnormal anatomy codes or the combination of the two codes, our algorithm can retrieve images according to the selected semantic component from a dataset consisting of brain magnetic resonance images of gliomas. Our CBIR system qualitatively and quantitatively achieves remarkable results.
△ Less
Submitted 23 March, 2021;
originally announced March 2021.
-
Learning Global and Local Features of Normal Brain Anatomy for Unsupervised Abnormality Detection
Authors:
Kazuma Kobayashi,
Ryuichiro Hataya,
Yusuke Kurose,
Amina Bolatkan,
Mototaka Miyake,
Hirokazu Watanabe,
Masamichi Takahashi,
Jun Itami,
Tatsuya Harada,
Ryuji Hamamoto
Abstract:
In real-world clinical practice, overlooking unanticipated findings can result in serious consequences. However, supervised learning, which is the foundation for the current success of deep learning, only encourages models to identify abnormalities that are defined in datasets in advance. Therefore, abnormality detection must be implemented in medical images that are not limited to a specific dise…
▽ More
In real-world clinical practice, overlooking unanticipated findings can result in serious consequences. However, supervised learning, which is the foundation for the current success of deep learning, only encourages models to identify abnormalities that are defined in datasets in advance. Therefore, abnormality detection must be implemented in medical images that are not limited to a specific disease category. In this study, we demonstrate an unsupervised learning framework for pixel-wise abnormality detection in brain magnetic resonance imaging captured from a patient population with metastatic brain tumor. Our concept is as follows: If an image reconstruction network can faithfully reproduce the global features of normal anatomy, then the abnormal lesions in unseen images can be identified based on the local difference from those reconstructed as normal by a discriminative network. Both networks are trained on a dataset comprising only normal images without labels. In addition, we devise a metric to evaluate the anatomical fidelity of the reconstructed images and confirm that the overall detection performance is improved when the image reconstruction network achieves a higher score. For evaluation, clinically significant abnormalities are comprehensively segmented. The results show that the area under the receiver operating characteristics curve values for metastatic brain tumors, extracranial metastatic tumors, postoperative cavities, and structural changes are 0.78, 0.61, 0.91, and 0.60, respectively.
△ Less
Submitted 8 May, 2021; v1 submitted 26 May, 2020;
originally announced May 2020.
-
Creation of nanometric magnetic skyrmions by global application of circularly polarized microwave magnetic field
Authors:
Masayuki Miyake,
Masahito Mochizuki
Abstract:
From a theoretical perspective, we demonstrate that nanometric magnetic skyrmions are created by application of a circularly polarized microwave magnetic field to a thin-plate Dzyaloshinskii-Moriya ferromagnet with fabricated rectangular holes. This phenomenon is caused by an effective steady magnetic field perpendicular to the microwave-polarization plane induced by the rotating magnetic field an…
▽ More
From a theoretical perspective, we demonstrate that nanometric magnetic skyrmions are created by application of a circularly polarized microwave magnetic field to a thin-plate Dzyaloshinskii-Moriya ferromagnet with fabricated rectangular holes. This phenomenon is caused by an effective steady magnetic field perpendicular to the microwave-polarization plane induced by the rotating magnetic field and the intense interference of spin waves excited by this magnetic field due to the hole structure, which causes reversals of local magnetizations and results in the formation of skyrmions. Our proposal provides a new option to write or create magnetic textures the sizes of which are much smaller than the spot size of the external stimulus such as magnetic field, light, and microwave.
△ Less
Submitted 16 March, 2020;
originally announced March 2020.
-
Development of a method for determining the search window for solar flare neutrinos
Authors:
K. Okamoto,
Y. Nakano,
S. Masuda,
Y. Itow,
M. Miyake,
T. Terasawa,
S. Ito,
M. Nakahata
Abstract:
Neutrinos generated during solar flares remain elusive. However, after $50$ years of discussion and search, the potential knowledge unleashed by their discovery keeps the search crucial. Neutrinos associated with solar flares provide information on otherwise poorly known particle acceleration mechanisms during solar flare. For neutrino detectors, the separation between atmospheric neutrinos and so…
▽ More
Neutrinos generated during solar flares remain elusive. However, after $50$ years of discussion and search, the potential knowledge unleashed by their discovery keeps the search crucial. Neutrinos associated with solar flares provide information on otherwise poorly known particle acceleration mechanisms during solar flare. For neutrino detectors, the separation between atmospheric neutrinos and solar flare neutrinos is technically encumbered by an energy band overlap. To improve differentiation from background neutrinos, we developed a method to determine the temporal search window for neutrino production during solar flares. Our method is based on data recorded by solar satellites, such as Geostationary Operational Environmental Satellite (GOES), Reuven Ramaty High Energy Solar Spectroscopic Imager (RHESSI), and GEOTAIL. In this study, we selected 23 solar flares above the X5.0 class that occurred between 1996 and 2018. We analyzed the light curves of soft X-rays, hard X-rays, $γ$-rays, line $γ$-rays from neutron capture as well as the derivative of soft X-rays. The average search windows are determined as follows: $4,178$ s for soft X-ray, $700$ s for derivative of soft X-ray, $944$ s for hard X-ray ($100$-$800$ keV), $1,586$ s for line $γ$-ray from neutron captures, and $776$ s for hard X-ray (above $50$ keV). This method allows neutrino detectors to improve their sensitivity to solar flare neutrinos.
△ Less
Submitted 16 September, 2020; v1 submitted 24 September, 2019;
originally announced September 2019.
-
Search for Proton Decay via $p \rightarrow νK^{+}$ using 260 kiloton$\cdot$year data of Super-Kamiokande
Authors:
The Super-Kamiokande Collaboration,
:,
K. Abe,
Y. Hayato,
K. Iyogi,
J. Kameda,
M. Miura,
S. Moriyama,
M. Nakahata,
S. Nakayama,
R. A. Wendell,
H. Sekiya,
M. Shiozawa,
Y. Suzuki,
A. Takeda,
Y. Takenaga,
K. Ueno,
T. Yokozawa,
H. Kaji,
T. Kajita,
K. Kaneyuki,
K. P. Lee,
K. Okumura,
T. McLachlan,
L. Labarga
, et al. (82 additional authors not shown)
Abstract:
We have searched for proton decay via $p \rightarrow νK^{+}$ using Super-Kamiokande data from April 1996 to February 2013, 260 kiloton$\cdot$year exposure in total. No evidence for this proton decay mode is found. A lower limit of the proton lifetime is set to $5.9 \times 10^{33}$ years at 90% confidence level.
We have searched for proton decay via $p \rightarrow νK^{+}$ using Super-Kamiokande data from April 1996 to February 2013, 260 kiloton$\cdot$year exposure in total. No evidence for this proton decay mode is found. A lower limit of the proton lifetime is set to $5.9 \times 10^{33}$ years at 90% confidence level.
△ Less
Submitted 6 August, 2014;
originally announced August 2014.
-
Calibration of the Super-Kamiokande Detector
Authors:
K. Abe,
Y. Hayato,
T. Iida,
K. Iyogi,
J. Kameda,
Y. Kishimoto,
Y. Koshio,
Ll. Marti,
M. Miura,
S. Moriyama,
M. Nakahata,
Y. Nakano,
S. Nakayama,
Y. Obayashi,
H. Sekiya,
M. Shiozawa,
Y. Suzuki,
A. Takeda,
Y. Takenaga,
H. Tanaka,
T. Tomura,
K. Ueno,
R. A. Wendell,
T. Yokozawa,
T. J. Irvine
, et al. (108 additional authors not shown)
Abstract:
Procedures and results on hardware level detector calibration in Super-Kamiokande (SK) are presented in this paper. In particular, we report improvements made in our calibration methods for the experimental phase IV in which new readout electronics have been operating since 2008. The topics are separated into two parts. The first part describes the determination of constants needed to interpret th…
▽ More
Procedures and results on hardware level detector calibration in Super-Kamiokande (SK) are presented in this paper. In particular, we report improvements made in our calibration methods for the experimental phase IV in which new readout electronics have been operating since 2008. The topics are separated into two parts. The first part describes the determination of constants needed to interpret the digitized output of our electronics so that we can obtain physical numbers such as photon counts and their arrival times for each photomultiplier tube (PMT). In this context, we developed an in-situ procedure to determine high-voltage settings for PMTs in large detectors like SK, as well as a new method for measuring PMT quantum efficiency and gain in such a detector. The second part describes the modeling of the detector in our Monte Carlo simulation, including in particular the optical properties of its water target and their variability over time. Detailed studies on the water quality are also presented. As a result of this work, we achieved a precision sufficient for physics analysis over a wide energy range (from a few MeV to above a TeV). For example, the charge determination was understood at the 1% level, and the timing resolution was 2.1 nsec at the one-photoelectron charge level and 0.5 nsec at the 100-photoelectron charge level.
△ Less
Submitted 20 December, 2013; v1 submitted 29 June, 2013;
originally announced July 2013.
-
Search for Proton Decay into Muon plus Neutral Kaon in Super-Kamiokande I, II, and III
Authors:
The Super-Kamiokande Collaboration,
:,
C. Regis,
K. Abe,
Y. Hayato,
K. Iyogi,
J. Kameda,
Y. Koshio,
Ll. Marti,
M. Miura,
S. Moriyama,
M. Nakahata,
S. Nakayama,
Y. Obayashi,
H. Sekiya,
M. Shiozawa,
Y. Suzuki,
A. Takeda,
Y. Takenaga,
K. Ueno,
T. Yokozawa,
H. Kaji,
T. Kajita,
K. Kaneyuki,
K. P. Lee
, et al. (83 additional authors not shown)
Abstract:
We have searched for proton into muon plus neutral kaon using data from a 91.7 kiloton-year exposure of Super-Kamiokande-I, a 49.2 kiloton-year exposure of Super-Kamiokande-II, and a 31.9 kiloton-year exposure of Super-Kamiokande-III. The number of candidate events in the data was consistent with the atmospheric neutrino background expectation and no evidence for proton decay in this mode was foun…
▽ More
We have searched for proton into muon plus neutral kaon using data from a 91.7 kiloton-year exposure of Super-Kamiokande-I, a 49.2 kiloton-year exposure of Super-Kamiokande-II, and a 31.9 kiloton-year exposure of Super-Kamiokande-III. The number of candidate events in the data was consistent with the atmospheric neutrino background expectation and no evidence for proton decay in this mode was found. We set a partial lifetime lower limit of 1.6x10^33 years at the 90% confidence level.
△ Less
Submitted 30 May, 2012;
originally announced May 2012.
-
Supernova Relic Neutrino Search at Super-Kamiokande
Authors:
The Super-Kamiokande Collaboration,
:,
K. Bays,
T. Iida,
K. Abe,
Y. Hayato,
K. Iyogi,
J. Kameda,
Y. Koshio,
L. Marti,
M. Miura,
S. Moriyama,
M. Nakahata,
S. Nakayama,
Y. Obayashi,
H. Sekiya,
M. Shiozawa,
Y. Suzuki,
A. Takeda,
Y. Takenaga,
K. Ueno,
K. Ueshima S. Yamada T. Yokozawa H. Kaji T. Kajita,
K. Kaneyuki,
T. McLachlan,
K. Okumura
, et al. (83 additional authors not shown)
Abstract:
A new Super-Kamiokande (SK) search for Supernova Relic Neutrinos (SRNs) was conducted using 2853 live days of data. Sensitivity is now greatly improved compared to the 2003 SK result, which placed a flux limit near many theoretical predictions. This more detailed analysis includes a variety of improvements such as increased efficiency, a lower energy threshold, and an expanded data set. New combin…
▽ More
A new Super-Kamiokande (SK) search for Supernova Relic Neutrinos (SRNs) was conducted using 2853 live days of data. Sensitivity is now greatly improved compared to the 2003 SK result, which placed a flux limit near many theoretical predictions. This more detailed analysis includes a variety of improvements such as increased efficiency, a lower energy threshold, and an expanded data set. New combined upper limits on SRN flux are between 2.8 and 3.0 nu_e cm^-2 s^-1 > 16 MeV total positron energy (17.3 MeV E_nu).
△ Less
Submitted 21 November, 2011;
originally announced November 2011.
-
A low-cost time-hop** impulse radio system for high data rate transmission
Authors:
Andreas F. Molisch,
Ye Geoffrey Li,
Yves-Paul Nakache,
Philip Orlik,
Makoto Miyake,
Yunnan Wu,
Sinan Gezici,
Harry Sheng,
S. Y. Kung,
H. Kobayashi,
H. Vincent Poor,
Alexander Haimovich,
**yun Zhang
Abstract:
We present an efficient, low-cost implementation of time-hop** impulse radio that fulfills the spectral mask mandated by the FCC and is suitable for high-data-rate, short-range communications. Key features are: (i) all-baseband implementation that obviates the need for passband components, (ii) symbol-rate (not chip rate) sampling, A/D conversion, and digital signal processing, (iii) fast acqu…
▽ More
We present an efficient, low-cost implementation of time-hop** impulse radio that fulfills the spectral mask mandated by the FCC and is suitable for high-data-rate, short-range communications. Key features are: (i) all-baseband implementation that obviates the need for passband components, (ii) symbol-rate (not chip rate) sampling, A/D conversion, and digital signal processing, (iii) fast acquisition due to novel search algorithms, (iv) spectral sha** that can be adapted to accommodate different spectrum regulations and interference environments. Computer simulations show that this system can provide 110Mbit/s at 7-10m distance, as well as higher data rates at shorter distances under FCC emissions limits. Due to the spreading concept of time-hop** impulse radio, the system can sustain multiple simultaneous users, and can suppress narrowband interference effectively.
△ Less
Submitted 9 February, 2005;
originally announced February 2005.
-
Direct evidence for ferromagnetic spin polarization in gold nanoparticles
Authors:
Y. Yamamoto,
T. Miura,
T. Teranishi,
M. Miyake,
H. Hori,
M. Suzuki,
N. Kawamura,
H. Miyagawa,
T. Nakamura,
K. Kobayashi
Abstract:
We report the first direct observation of ferromagnetic spin polarization of Au nanoparticles with a mean diameter of 1.9 nm using X-ray magnetic circular dichroism (XMCD). Owing to the element selectivity of XMCD, only the gold magnetization is explored. Magnetization of gold atoms estimated by XMCD shows a good agreement with the results obtained by conventional magnetometry. This result is ev…
▽ More
We report the first direct observation of ferromagnetic spin polarization of Au nanoparticles with a mean diameter of 1.9 nm using X-ray magnetic circular dichroism (XMCD). Owing to the element selectivity of XMCD, only the gold magnetization is explored. Magnetization of gold atoms estimated by XMCD shows a good agreement with the results obtained by conventional magnetometry. This result is evidence of intrinsic spin polarization in nano-sized gold.
△ Less
Submitted 24 March, 2004;
originally announced March 2004.
-
Diameter dependence of ferromagnetic spin moment in Au nanocrystals
Authors:
H. Hori,
Y. Yamamoto,
T. Iwamoto,
T. Miura,
T. Teranishi,
M. Miyake
Abstract:
Au nanoparticles exhibit ferromagnetic spin polarization and show diameter dependence in magnetization. The magnetic moment per Au atom in the particle attains its maximum value at a diameter of about 3 nanometer (nm) in the Magnetization-Diameter curve. Because Au metal is a typical diamagnetic material, its ferromagnetic polarization mechanism is thought to be quite different from the ferromag…
▽ More
Au nanoparticles exhibit ferromagnetic spin polarization and show diameter dependence in magnetization. The magnetic moment per Au atom in the particle attains its maximum value at a diameter of about 3 nanometer (nm) in the Magnetization-Diameter curve. Because Au metal is a typical diamagnetic material, its ferromagnetic polarization mechanism is thought to be quite different from the ferromagnetism observed in transition metals. The size effect strongly suggests the existence of some spin correlation effect at the nanoscale. The so-called ``Fermi hole effect'' is the most probable one given in the free electron gas system. Ferromagnetism in Au nanoparticles is discussed using this model.
△ Less
Submitted 12 March, 2004; v1 submitted 11 June, 2003;
originally announced June 2003.
