In current neural network algorithms for nuclide identification in high-background,poor-resolution detectors,traditional network paradigms including back-propagation networks,convolutional neural networks,recurrent ne...In current neural network algorithms for nuclide identification in high-background,poor-resolution detectors,traditional network paradigms including back-propagation networks,convolutional neural networks,recurrent neural networks,etc.,have been limited in research on γ spectrum analysis because of their inherent mathematical mechanisms.It is difficult to make progress in terms of training data requirements and prediction accuracy.In contrast to traditional network paradigms,network models based on the transformer structure have the characteristics of parallel computing,position encoding,and deep stacking,which have enabled good performance in natural language processing tasks in recent years.Therefore,in this paper,a transformer-based neural network (TBNN) model is proposed to achieve nuclide identification for the first time.First,the Geant4 program was used to generate the basic single-nuclide energy spectrum through Monte Carlo simulations.A multi-nuclide energy spectrum database was established for neural network training using random matrices of γ-ray energy,activity,and noise.Based on the encoder–decoder structure,a network topology based on the transformer was built,transforming the 1024-channel energy spectrum data into a 32×32 energy spectrum sequence as the model input.Through experiments and adjustments of model parameters,including the learning rate of the TBNN model,number of attention heads,and number of network stacking layers,the overall recognition rate reached 98.7%.Additionally,this database was used for training AI models such as back-propagation networks,convolutional neural networks,residual networks,and long shortterm memory neural networks,with overall recognition rates of 92.8%,95.3%,96.3%,and 96.6%,respectively.This indicates that the TBNN model exhibited better nuclide identification among these AI models,providing an important reference and theoretical basis for the practical application of transformers in the qualitative and quantitative analysis of the γ spectrum.展开更多
BACKGROUND Mycoplasma hominis(M.hominis),which causes central nervous system infections in adults,is very rare.It is also relatively difficult to culture mycoplasma and culturing requires special media,resulting in a ...BACKGROUND Mycoplasma hominis(M.hominis),which causes central nervous system infections in adults,is very rare.It is also relatively difficult to culture mycoplasma and culturing requires special media,resulting in a high rate of clinical underdiagnosis.Therefore,clinicians often treat patients based on their own experience before obtaining pathogenic results and may ignore infections with atypical pathogens,thus delaying the diagnosis and treatment of patients and increasing the length of hospital stay and costs.CASE SUMMARY A 44-year-old man presented to the hospital complaining of recurrent dizziness for 1 year,which had worsened in the last week.After admission,brain magnetic resonance imaging(MRI)revealed a 7.0 cm×6.0 cm×6.1 cm lesion at the skull base,which was irregular in shape and had a midline shift to the left.Based on imaging findings,meningioma was our primary consideration.After lesion resection,the patient had persistent fever and a diagnosis of suppurative meningitis based on cerebrospinal fluid(CSF)examination.The patient was treated with the highest level of antibiotics(meropenem and linezolid),but the response was ineffective.Finally,M.hominis was detected by next-generation metagenomic sequencing(mNGS)in the CSF.Therefore,we changed the antibiotics to moxifloxacin 0.4 g daily combined with doxycycline 0.1 g twice a day for 2 wk,and the patient had a normal temperature the next day.CONCLUSION Mycoplasma meningitis after neurosurgery is rare.We can use mNGS to detect M.hominis in the CSF and then provide targeted treatment.展开更多
基金supported by the National Natural Science Foundation of China(No.42127807)Natural Science Foundation of Sichuan Province(Nos.2024NSFSC0422,23NSFSCC0116)Nuclear Energy Development Project(No.[2021]-88).
文摘In current neural network algorithms for nuclide identification in high-background,poor-resolution detectors,traditional network paradigms including back-propagation networks,convolutional neural networks,recurrent neural networks,etc.,have been limited in research on γ spectrum analysis because of their inherent mathematical mechanisms.It is difficult to make progress in terms of training data requirements and prediction accuracy.In contrast to traditional network paradigms,network models based on the transformer structure have the characteristics of parallel computing,position encoding,and deep stacking,which have enabled good performance in natural language processing tasks in recent years.Therefore,in this paper,a transformer-based neural network (TBNN) model is proposed to achieve nuclide identification for the first time.First,the Geant4 program was used to generate the basic single-nuclide energy spectrum through Monte Carlo simulations.A multi-nuclide energy spectrum database was established for neural network training using random matrices of γ-ray energy,activity,and noise.Based on the encoder–decoder structure,a network topology based on the transformer was built,transforming the 1024-channel energy spectrum data into a 32×32 energy spectrum sequence as the model input.Through experiments and adjustments of model parameters,including the learning rate of the TBNN model,number of attention heads,and number of network stacking layers,the overall recognition rate reached 98.7%.Additionally,this database was used for training AI models such as back-propagation networks,convolutional neural networks,residual networks,and long shortterm memory neural networks,with overall recognition rates of 92.8%,95.3%,96.3%,and 96.6%,respectively.This indicates that the TBNN model exhibited better nuclide identification among these AI models,providing an important reference and theoretical basis for the practical application of transformers in the qualitative and quantitative analysis of the γ spectrum.
文摘BACKGROUND Mycoplasma hominis(M.hominis),which causes central nervous system infections in adults,is very rare.It is also relatively difficult to culture mycoplasma and culturing requires special media,resulting in a high rate of clinical underdiagnosis.Therefore,clinicians often treat patients based on their own experience before obtaining pathogenic results and may ignore infections with atypical pathogens,thus delaying the diagnosis and treatment of patients and increasing the length of hospital stay and costs.CASE SUMMARY A 44-year-old man presented to the hospital complaining of recurrent dizziness for 1 year,which had worsened in the last week.After admission,brain magnetic resonance imaging(MRI)revealed a 7.0 cm×6.0 cm×6.1 cm lesion at the skull base,which was irregular in shape and had a midline shift to the left.Based on imaging findings,meningioma was our primary consideration.After lesion resection,the patient had persistent fever and a diagnosis of suppurative meningitis based on cerebrospinal fluid(CSF)examination.The patient was treated with the highest level of antibiotics(meropenem and linezolid),but the response was ineffective.Finally,M.hominis was detected by next-generation metagenomic sequencing(mNGS)in the CSF.Therefore,we changed the antibiotics to moxifloxacin 0.4 g daily combined with doxycycline 0.1 g twice a day for 2 wk,and the patient had a normal temperature the next day.CONCLUSION Mycoplasma meningitis after neurosurgery is rare.We can use mNGS to detect M.hominis in the CSF and then provide targeted treatment.
