Masahiro Yakami
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Masahiro Yakami has a total of 89 co-authors. Here are the top co-authors by number of shared publications:
Kaori Togashi
Department of Diagnostic Imaging and Nuclear Medicine Kyoto University Graduate School of Medicine Shogoin Sakyo-ku Kyoto Kyoto Japan
6 shared publications- Tomohiro Kuroda
Graduate School of Informatics, Kyoto University, Kyoto, Japan
4 shared publications - Kazumasa Kishimoto
Kyoto University
3 shared publications - Mizuho Nishio
Department of Radiology, Kobe University Graduate School of Medicine, Kobe, Japan
3 shared publications - Osamu Sugiyama
Kindai University
3 shared publications Takeshi Kubo
Kyoto University Hospital
3 shared publications
Latest Publications
- Open Access
- 12 November 2025
- Published by JMIR Publications Inc. in JMIR Medical Informatics
- Vol. 13, e35987
Detection of Polyphonic Alarm Sounds From Medical Devices Using Frequency-Enhanced Deep Learning: Simulation Study
Background: Although an increasing number of bedside medical devices are equipped with wireless connections for reliable notifications, many nonnetworked devices remain effective at detecting abnormal patient conditions and alerting medical staff through auditory alarms. Staff members, however, can miss these notifications, especially when in distant areas or other private rooms. In contrast, the signal-to-noise ratio of alarm systems for medical devices in the neonatal intensive care unit is 0 dB or higher. A feasible system for automatic sound identification with high accuracy is needed to prevent alarm sounds from being missed by the staff. Objective: The purpose of this study was to design a method for classifying multiple alarm sounds collected with a monaural microphone in a noisy environment. Methods: Features of 7 alarm sounds were extracted using a mel filter bank and incorporated into a classifier using convolutional and recurrent neural networks. To estimate its clinical usefulness, the classifier was evaluated with mixtures of up to 7 alarm sounds and hospital ward noise. Results: The proposed convolutional recurrent neural network model was evaluated using a simulation dataset of 7 alarm sounds mixed with hospital ward noise. At a signal-to-noise ratio of 0 dB, the best-performing model (convolutional neural network 3+bidirectional gate recurrent unit) achieved an event-based F 1 -score of 0.967, with a precision of 0.944 and a recall of 0.991. When the venous foot pump class was excluded, the classwise recall of the classifier ranged from 0.990 to 1.000. Conclusions: The proposed classifier was found to be highly accurate in detecting alarm sounds. Although the performance of the proposed classifier in a clinical environment can be improved, the classifier could be incorporated into an alarm sound detection system. The classifier, combined with network connectivity, could improve the notification of abnormal status detected by unconnected medical devices.
- 12 November 2025
- Published by JMIR Publications Inc. in JMIR Medical Informatics
- Vol. 13, e35987
Open AccessDetection of Polyphonic Alarm Sounds From Medical Devices Using Frequency-Enhanced Deep Learning: Simulation Study
Background: Although an increasing number of bedside medical devices are equipped with wireless connections for reliable notifications, many nonnetworked devices remain effective at detecting abnormal patient conditions and alerting medical staff through auditory alarms. Staff members, however, can miss these notifications, especially when in distant areas or other private rooms. In contrast, the signal-to-noise ratio of alarm systems for medical devices in the neonatal intensive care unit is 0 dB or higher. A feasible system for automatic sound identification with high accuracy is needed to prevent alarm sounds from being missed by the staff. Objective: The purpose of this study was to design a method for classifying multiple alarm sounds collected with a monaural microphone in a noisy environment. Methods: Features of 7 alarm sounds were extracted using a mel filter bank and incorporated into a classifier using convolutional and recurrent neural networks. To estimate its clinical usefulness, the classifier was evaluated with mixtures of up to 7 alarm sounds and hospital ward noise. Results: The proposed convolutional recurrent neural network model was evaluated using a simulation dataset of 7 alarm sounds mixed with hospital ward noise. At a signal-to-noise ratio of 0 dB, the best-performing model (convolutional neural network 3+bidirectional gate recurrent unit) achieved an event-based F 1 -score of 0.967, with a precision of 0.944 and a recall of 0.991. When the venous foot pump class was excluded, the classwise recall of the classifier ranged from 0.990 to 1.000. Conclusions: The proposed classifier was found to be highly accurate in detecting alarm sounds. Although the performance of the proposed classifier in a clinical environment can be improved, the classifier could be incorporated into an alarm sound detection system. The classifier, combined with network connectivity, could improve the notification of abnormal status detected by unconnected medical devices.
- Open Access
- 30 March 2025
- Published by JMIR Publications Inc. in JMIR Preprints
Detection of Alarm Status of Various Medical Devices to Enhance Patient Safety: Deep Alarm Sound Detection (Preprint)
BACKGROUND Although an increasing number of bedside medical devices have wireless connections, providing feasible and reliable notification, many devices without any connections perform very well in detecting abnormal status and alerting medical staff with various sounds. Staff members, however, can miss these notifications, especially when in distant areas or other private rooms. In contrast, the signal-to-noise ratio (SNR) of alarm systems for medical devices in the neonatal intensive care unit is 0 dB or more. A feasible system for automatic sound identification with high accuracy is needed to prevent alarm sounds from being missed by the staff. OBJECTIVE The purpose of this study was to design a method for classifying multiple alarm sounds collected with a monaural microphone in a noisy environment. METHODS Features of seven alarm sounds were extracted with a Mel filter bank and incorporated into a classifier using convolutional and recurrent neural networks. To estimate its clinical usefulness, the classifier was evaluated with mixtures of up to seven alarm sounds and hospital ward noise. RESULTS At an SNR of 0 dB, the recall performance of the classifier was 0.976 with an F score of 0.945. When foot pump was excluded, the class-wise recall the classifier ranged from 0.990 to 1.000. CONCLUSIONS The proposed classifier was found to be highly accurate in detecting alarm sounds. Although the performance of the proposed classifier in a clinical environment can be improved, the classifier could be incorporated into an alarm sound detection system. The classifier, combined with network connectivity, could improve the notification of abnormal status detected by unconnected medical devices.
- 30 March 2025
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessDetection of Alarm Status of Various Medical Devices to Enhance Patient Safety: Deep Alarm Sound Detection (Preprint)
BACKGROUND Although an increasing number of bedside medical devices have wireless connections, providing feasible and reliable notification, many devices without any connections perform very well in detecting abnormal status and alerting medical staff with various sounds. Staff members, however, can miss these notifications, especially when in distant areas or other private rooms. In contrast, the signal-to-noise ratio (SNR) of alarm systems for medical devices in the neonatal intensive care unit is 0 dB or more. A feasible system for automatic sound identification with high accuracy is needed to prevent alarm sounds from being missed by the staff. OBJECTIVE The purpose of this study was to design a method for classifying multiple alarm sounds collected with a monaural microphone in a noisy environment. METHODS Features of seven alarm sounds were extracted with a Mel filter bank and incorporated into a classifier using convolutional and recurrent neural networks. To estimate its clinical usefulness, the classifier was evaluated with mixtures of up to seven alarm sounds and hospital ward noise. RESULTS At an SNR of 0 dB, the recall performance of the classifier was 0.976 with an F score of 0.945. When foot pump was excluded, the class-wise recall the classifier ranged from 0.990 to 1.000. CONCLUSIONS The proposed classifier was found to be highly accurate in detecting alarm sounds. Although the performance of the proposed classifier in a clinical environment can be improved, the classifier could be incorporated into an alarm sound detection system. The classifier, combined with network connectivity, could improve the notification of abnormal status detected by unconnected medical devices.
- Open Access
- 22 August 2024
- Published by IOS Press
- Vol. 316, 1746-1747
Investigation of Radiologist Diagnostic Difficulty Prediction Without CT Images
For better collaboration among radiologists, the interpretation workload should be evaluated, considering the difference in difficulty for interpreting each case. However, objective evaluation of difficulty is challenging. This study proposes a multimodal classifier of structural and textual data to predict difficulty based on order information and patient data without using images. The classifier showed performance with a specificity of 0.9 and an accuracy of 0.7.
- 22 August 2024
- Published by IOS Press
- Vol. 316, 1746-1747
Open AccessInvestigation of Radiologist Diagnostic Difficulty Prediction Without CT Images
For better collaboration among radiologists, the interpretation workload should be evaluated, considering the difference in difficulty for interpreting each case. However, objective evaluation of difficulty is challenging. This study proposes a multimodal classifier of structural and textual data to predict difficulty based on order information and patient data without using images. The classifier showed performance with a specificity of 0.9 and an accuracy of 0.7.
- 7 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 67-68
Investigation of Radiologist Diagnostic Workload Prediction without CT Images Using Multimodal Deep Learning
For radiologists to collaborate, work must be distributed equally to prevent any sense of unfairness. This study proposes a model to predict the workload of radiologists based on order information and patient data without CT images. The multimodal multiclass classifier showed the highest performance when combining structured data and text data. This suggested that features contributing to workload prediction include not only in the text data but also in the structured data.
- 7 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 67-68
Investigation of Radiologist Diagnostic Workload Prediction without CT Images Using Multimodal Deep Learning
For radiologists to collaborate, work must be distributed equally to prevent any sense of unfairness. This study proposes a model to predict the workload of radiologists based on order information and patient data without CT images. The multimodal multiclass classifier showed the highest performance when combining structured data and text data. This suggested that features contributing to workload prediction include not only in the text data but also in the structured data.
- Open Access
- 18 March 2022
- Published by Association for Computing Machinery (ACM)
- p. 58-62
Investigation on continual training of computer-aided diagnosis systems by semi-supervised learning
- 18 March 2022
- Published by Association for Computing Machinery (ACM)
- p. 58-62
Open AccessInvestigation on continual training of computer-aided diagnosis systems by semi-supervised learning
- Open Access
- 15 July 2021
- Published by MDPI AG in Diagnostics
- Vol. 11  (7) , 1267
A Proposed Dedicated Breast PET Lexicon: Standardization of Description and Reporting of Radiotracer Uptake in the Breast
Dedicated breast positron emission tomography (dbPET) is a new diagnostic imaging modality recently used in clinical practice for the detection of breast cancer and the assessment of tumor biology. dbPET has higher spatial resolution than that of conventional whole body PET systems, allowing recognition of detailed morphological attributes of radiotracer accumulation within the breast. 18F-fluorodeoxyglucose (18F-FDG) accumulation in the breast may be due to benign or malignant entities, and recent studies suggest that morphology characterization of 18F-FDG uptake could aid in estimating the probability of malignancy. However, across the world, there are many descriptors of breast 18F-FDG uptake, limiting comparisons between studies. In this article, we propose a lexicon for breast radiotracer uptake to standardize description and reporting of image findings on dbPET, consisting of terms for image quality, radiotracer fibroglandular uptake, breast lesion uptake.
- 15 July 2021
- Published by MDPI AG in Diagnostics
- Vol. 11  (7) , 1267
Open AccessA Proposed Dedicated Breast PET Lexicon: Standardization of Description and Reporting of Radiotracer Uptake in the Breast
Dedicated breast positron emission tomography (dbPET) is a new diagnostic imaging modality recently used in clinical practice for the detection of breast cancer and the assessment of tumor biology. dbPET has higher spatial resolution than that of conventional whole body PET systems, allowing recognition of detailed morphological attributes of radiotracer accumulation within the breast. 18F-fluorodeoxyglucose (18F-FDG) accumulation in the breast may be due to benign or malignant entities, and recent studies suggest that morphology characterization of 18F-FDG uptake could aid in estimating the probability of malignancy. However, across the world, there are many descriptors of breast 18F-FDG uptake, limiting comparisons between studies. In this article, we propose a lexicon for breast radiotracer uptake to standardize description and reporting of image findings on dbPET, consisting of terms for image quality, radiotracer fibroglandular uptake, breast lesion uptake.
- Open Access
- 6 October 2020
- Published by Springer Nature in Journal of Digital Imaging
- Vol. 33  (6) , 1543-1553
Adaptive Voxel Matching for Temporal CT Subtraction
Temporal subtraction (TS) technique calculates a subtraction image between a pair of registered images acquired from the same patient at different times. Previous studies have shown that TS is effective for visualizing pathological changes over time; therefore, TS should be a useful tool for radiologists. However, artifacts caused by partial volume effects degrade the quality of thick-slice subtraction images, even with accurate image registration. Here, we propose a subtraction method for reducing artifacts in thick-slice images and discuss its implementation in high-speed processing. The proposed method is based on voxel matching, which reduces artifacts by considering gaps in discretized positions of two images in subtraction calculations. There are two different features between the proposed method and conventional voxel matching: (1) the size of a searching region to reduce artifacts is determined based on discretized position gaps between images and (2) the searching region is set on both images for symmetrical subtraction. The proposed method is implemented by adopting an accelerated subtraction calculation method that exploit the nature of liner interpolation for calculating the signal value at a point among discretized positions. We quantitatively evaluated the proposed method using synthetic data and qualitatively using clinical data interpreted by radiologists. The evaluation showed that the proposed method was superior to conventional methods. Moreover, the processing speed using the proposed method was almost unchanged from that of the conventional methods. The results indicate that the proposed method can improve the quality of subtraction images acquired from thick-slice images.
- 6 October 2020
- Published by Springer Nature in Journal of Digital Imaging
- Vol. 33  (6) , 1543-1553
Open AccessAdaptive Voxel Matching for Temporal CT Subtraction
Temporal subtraction (TS) technique calculates a subtraction image between a pair of registered images acquired from the same patient at different times. Previous studies have shown that TS is effective for visualizing pathological changes over time; therefore, TS should be a useful tool for radiologists. However, artifacts caused by partial volume effects degrade the quality of thick-slice subtraction images, even with accurate image registration. Here, we propose a subtraction method for reducing artifacts in thick-slice images and discuss its implementation in high-speed processing. The proposed method is based on voxel matching, which reduces artifacts by considering gaps in discretized positions of two images in subtraction calculations. There are two different features between the proposed method and conventional voxel matching: (1) the size of a searching region to reduce artifacts is determined based on discretized position gaps between images and (2) the searching region is set on both images for symmetrical subtraction. The proposed method is implemented by adopting an accelerated subtraction calculation method that exploit the nature of liner interpolation for calculating the signal value at a point among discretized positions. We quantitatively evaluated the proposed method using synthetic data and qualitatively using clinical data interpreted by radiologists. The evaluation showed that the proposed method was superior to conventional methods. Moreover, the processing speed using the proposed method was almost unchanged from that of the conventional methods. The results indicate that the proposed method can improve the quality of subtraction images acquired from thick-slice images.
- Open Access
- 16 November 2018
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 13  (11) , e0207661
Automatic inference model construction for computer-aided diagnosis of lung nodule: Explanation adequacy, inference accuracy, and experts’ knowledge
We aimed to describe the development of an inference model for computer-aided diagnosis of lung nodules that could provide valid reasoning for any inferences, thereby improving the interpretability and performance of the system. An automatic construction method was used that considered explanation adequacy and inference accuracy. In addition, we evaluated the usefulness of prior experts’ (radiologists’) knowledge while constructing the models. In total, 179 patients with lung nodules were included and divided into 79 and 100 cases for training and test data, respectively. F-measure and accuracy were used to assess explanation adequacy and inference accuracy, respectively. For F-measure, reasons were defined as proper subsets of Evidence that had a strong influence on the inference result. The inference models were automatically constructed using the Bayesian network and Markov chain Monte Carlo methods, selecting only those models that met the predefined criteria. During model constructions, we examined the effect of including radiologist’s knowledge in the initial Bayesian network models. Performance of the best models in terms of F-measure, accuracy, and evaluation metric were as follows: 0.411, 72.0%, and 0.566, respectively, with prior knowledge, and 0.274, 65.0%, and 0.462, respectively, without prior knowledge. The best models with prior knowledge were then subjectively and independently evaluated by two radiologists using a 5-point scale, with 5, 3, and 1 representing beneficial, appropriate, and detrimental, respectively. The average scores by the two radiologists were 3.97 and 3.76 for the test data, indicating that the proposed computer-aided diagnosis system was acceptable to them. In conclusion, the proposed method incorporating radiologists’ knowledge could help in eliminating radiologists’ distrust of computer-aided diagnosis and improving its performance.
- 16 November 2018
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 13  (11) , e0207661
Open AccessAutomatic inference model construction for computer-aided diagnosis of lung nodule: Explanation adequacy, inference accuracy, and experts’ knowledge
We aimed to describe the development of an inference model for computer-aided diagnosis of lung nodules that could provide valid reasoning for any inferences, thereby improving the interpretability and performance of the system. An automatic construction method was used that considered explanation adequacy and inference accuracy. In addition, we evaluated the usefulness of prior experts’ (radiologists’) knowledge while constructing the models. In total, 179 patients with lung nodules were included and divided into 79 and 100 cases for training and test data, respectively. F-measure and accuracy were used to assess explanation adequacy and inference accuracy, respectively. For F-measure, reasons were defined as proper subsets of Evidence that had a strong influence on the inference result. The inference models were automatically constructed using the Bayesian network and Markov chain Monte Carlo methods, selecting only those models that met the predefined criteria. During model constructions, we examined the effect of including radiologist’s knowledge in the initial Bayesian network models. Performance of the best models in terms of F-measure, accuracy, and evaluation metric were as follows: 0.411, 72.0%, and 0.566, respectively, with prior knowledge, and 0.274, 65.0%, and 0.462, respectively, without prior knowledge. The best models with prior knowledge were then subjectively and independently evaluated by two radiologists using a 5-point scale, with 5, 3, and 1 representing beneficial, appropriate, and detrimental, respectively. The average scores by the two radiologists were 3.97 and 3.76 for the test data, indicating that the proposed computer-aided diagnosis system was acceptable to them. In conclusion, the proposed method incorporating radiologists’ knowledge could help in eliminating radiologists’ distrust of computer-aided diagnosis and improving its performance.
- 10 February 2015
- Published by Oxford University Press (OUP) in Radiation Protection Dosimetry
- Vol. 168  (1) , 61-71
Optimised paediatric CT dose at a tertiary children's hospital in Japan: a 4-y single-centre analysis
Since diagnostic reference levels (DRLs) for children are not currently established in Japan, the authors determined local DRLs for the full range of paediatric CT examinations in a single tertiary care children's hospital. A retrospective review of 4801 CT performance records for paediatric patients (vol and DLP for head, chest and abdomen increased as a function of age. Benchmarking of the results showed that CTDIvol, DLP and effective dose for chest and abdomen examinations in this hospital were below average, whereas those for the head tended to be at or slightly above average of established DRL values from five countries. The results suggest that CT examinations as performed in a tertiary children's hospital in Japan are well optimised.
- 10 February 2015
- Published by Oxford University Press (OUP) in Radiation Protection Dosimetry
- Vol. 168  (1) , 61-71
Optimised paediatric CT dose at a tertiary children's hospital in Japan: a 4-y single-centre analysis
Since diagnostic reference levels (DRLs) for children are not currently established in Japan, the authors determined local DRLs for the full range of paediatric CT examinations in a single tertiary care children's hospital. A retrospective review of 4801 CT performance records for paediatric patients (vol and DLP for head, chest and abdomen increased as a function of age. Benchmarking of the results showed that CTDIvol, DLP and effective dose for chest and abdomen examinations in this hospital were below average, whereas those for the head tended to be at or slightly above average of established DRL values from five countries. The results suggest that CT examinations as performed in a tertiary children's hospital in Japan are well optimised.
- 1 November 2012
- Published by Oxford University Press (OUP) in British Journal of Radiology
- Vol. 85  (1019) , e1004-e1008
Evaluation of CT angiography for visualisation of the lenticulostriate artery: difference between normotensive and hypertensive patients
Objective: High-resolution CT angiography (CTA) is currently available using multidetector row CT (MDCT); however, its use for small artery visualisation has been limited. To evaluate its capability, we investigated CTA visualisation for difference in number of the lenticulostriate artery (LSA) branches between normotensive and hypertensive patients, because hypertension is a major cause of LSA damage. Methods: This was a retrospective study evaluating cerebrovascular CTA at our hospital conducted from February 2008 to June 2009 under approval of the institutional review board. 117 patients (39 males and 78 females, 19–88 years old) were included. CTA was conducted using a 64 channel MDCT. Total numbers of LSA branches were examined for differences by age with regression analysis and the presence or absence of hypertension and/or aneurysm using two-sample t-tests. A p-value <0.016 was considered statistically significant after correction for multiple comparisons. A multiple variable analysis of three factors was also conducted. Results: The average number of LSA branches was 3.6 [95% confidence interval (CI) 3.0–4.1] and 4.4 (95% CI 4.1–4.7), respectively, for a patient with and without history of hypertension, and the difference was statistically significant (p=0.013). The difference was approximately one branch in the multiple variable analysis. No significant correlation was observed for age and no significant difference was observed for the presence or absence of aneurysms. Conclusions: Contrast-enhanced CTA can visualise significant differences in the number of LSA branches among patients with and without hypertension. Advances in knowledge: Current high-resolution CTA can visualise LSA well, which enables finding a difference in the LSA between normotensive subjects and hypertensive patients.
- 1 November 2012
- Published by Oxford University Press (OUP) in British Journal of Radiology
- Vol. 85  (1019) , e1004-e1008
Evaluation of CT angiography for visualisation of the lenticulostriate artery: difference between normotensive and hypertensive patients
Objective: High-resolution CT angiography (CTA) is currently available using multidetector row CT (MDCT); however, its use for small artery visualisation has been limited. To evaluate its capability, we investigated CTA visualisation for difference in number of the lenticulostriate artery (LSA) branches between normotensive and hypertensive patients, because hypertension is a major cause of LSA damage. Methods: This was a retrospective study evaluating cerebrovascular CTA at our hospital conducted from February 2008 to June 2009 under approval of the institutional review board. 117 patients (39 males and 78 females, 19–88 years old) were included. CTA was conducted using a 64 channel MDCT. Total numbers of LSA branches were examined for differences by age with regression analysis and the presence or absence of hypertension and/or aneurysm using two-sample t-tests. A p-value <0.016 was considered statistically significant after correction for multiple comparisons. A multiple variable analysis of three factors was also conducted. Results: The average number of LSA branches was 3.6 [95% confidence interval (CI) 3.0–4.1] and 4.4 (95% CI 4.1–4.7), respectively, for a patient with and without history of hypertension, and the difference was statistically significant (p=0.013). The difference was approximately one branch in the multiple variable analysis. No significant correlation was observed for age and no significant difference was observed for the presence or absence of aneurysms. Conclusions: Contrast-enhanced CTA can visualise significant differences in the number of LSA branches among patients with and without hypertension. Advances in knowledge: Current high-resolution CTA can visualise LSA well, which enables finding a difference in the LSA between normotensive subjects and hypertensive patients.