Yu Sa
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Yu Sa has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
- Ning Zhang
Department of Biomedical Engineering, Tianjin Key Lab of BME Measurement, Tianjin University, Tianjin, P.R. China
3 shared publications - Xin-Hua Hu
Institute for Advanced Optics, Hunan Institute of Science and Technology, Yueyang, Hunan 414006, China
3 shared publications - Yu Guo
Medical School, Tianjin University, China
3 shared publications - Yuanming Feng
Department of Biomedical Engineering, Tianjin University, Tianjin, China
3 shared publications Changrong Jin
Department of Biomedical Engineering, Tianjin University, Tianjin 300072, China
2 shared publicationsChengwen Yang
Department of Biomedical Engineering, Tianjin University, 92 Weijin Rd., Tianjin, 300072, China
2 shared publications
Latest Publications
- 23 April 2026
- Published by Springer Nature
- p. 487-499
Automatic Parameters Measurement System of Reflective Holographic Grating
- 23 April 2026
- Published by Springer Nature
- p. 487-499
Automatic Parameters Measurement System of Reflective Holographic Grating
- Open Access
- 16 July 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15  (1) , 1-10
Fusion of microscopic and diffraction images with VGG net for budding yeast recognition in imaging flow cytometry
Microscopic-Diffraction Imaging Flow Cytometry (MDIFC) is a high-throughput, stain-free technology that captures paired microscopic and diffraction images of cellular events, utilizing machine learning for the classification of cell subpopulations. However, MDIFC is still hindered by challenges related to limited accuracy, processing speed, and a lack of automation. To address this, we propose a novel approach that integrates image fusion techniques with a deep learning-based classification algorithm. Using budding yeast recognition as a model system, we categorized events into three groups: single cells, budding cells, and aggregated cells. Paired images were fused with varying weight factors to generate a comprehensive training dataset for a VGG-net-based Convolutional Neural Network (CNN). For comparison, Support Vector Machines (SVM) and Random Forests (RF) based on Grey-Level Co-occurrence Matrix (GLCM) features were employed. The results demonstrate that the VGG-net classifier achieved an impressive classification accuracy of 0.98 when trained on a dataset with a fusion weight of 0.2 for microscopic images and 0.8 for diffraction images. Furthermore, it demonstrated a high throughput of 260.42 cells per second, surpassing the performance of GLCM-based methods. These findings suggest that the combination of image fusion and deep learning algorithms significantly improves both the speed and accuracy of cell classification in MDIFC, offering substantial benefits for high-throughput cell analysis in biological and medical applications.
- 16 July 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15  (1) , 1-10
Open AccessFusion of microscopic and diffraction images with VGG net for budding yeast recognition in imaging flow cytometry
Microscopic-Diffraction Imaging Flow Cytometry (MDIFC) is a high-throughput, stain-free technology that captures paired microscopic and diffraction images of cellular events, utilizing machine learning for the classification of cell subpopulations. However, MDIFC is still hindered by challenges related to limited accuracy, processing speed, and a lack of automation. To address this, we propose a novel approach that integrates image fusion techniques with a deep learning-based classification algorithm. Using budding yeast recognition as a model system, we categorized events into three groups: single cells, budding cells, and aggregated cells. Paired images were fused with varying weight factors to generate a comprehensive training dataset for a VGG-net-based Convolutional Neural Network (CNN). For comparison, Support Vector Machines (SVM) and Random Forests (RF) based on Grey-Level Co-occurrence Matrix (GLCM) features were employed. The results demonstrate that the VGG-net classifier achieved an impressive classification accuracy of 0.98 when trained on a dataset with a fusion weight of 0.2 for microscopic images and 0.8 for diffraction images. Furthermore, it demonstrated a high throughput of 260.42 cells per second, surpassing the performance of GLCM-based methods. These findings suggest that the combination of image fusion and deep learning algorithms significantly improves both the speed and accuracy of cell classification in MDIFC, offering substantial benefits for high-throughput cell analysis in biological and medical applications.
- 1 March 2022
- Published by Bentham Science Publishers Ltd. in Current Medicinal Chemistry
- Vol. 29  (9) , 1622-1639
Identification of WHO II/III Gliomas by 16 Prognostic-related Gene Signatures using Machine Learning Methods
Background: It is found that the prognosis of gliomas of the same grade has large differences among World Health Organization (WHO) grade II and III in clinical observation. Therefore, a better understanding of the genetics and molecular mechanisms underlying WHO grade II and III gliomas is required, with the aim of developing a classification scheme at the molecular level rather than the conventional pathological morphology level. Method: We performed survival analysis combined with machine learning methods of Least Absolute Shrinkage and Selection Operator using expression datasets downloaded from the Chinese Glioma Genome Atlas as well as The Cancer Genome Atlas. Risk scores were calculated by the product of expression level of overall survival-related genes and their multivariate Cox proportional hazards regression coefficients. WHO grade II and III gliomas were categorized into the low-risk subgroup, medium-risk subgroup, and high-risk subgroup. We used the 16 prognostic-related genes as input features to build a classification model based on prognosis using a fully connected neural network. Gene function annotations were also performed. Results: The 16 genes (AKNAD1, C7orf13, CDK20, CHRFAM7A, CHRNA1, EFNB1, GAS1, HIST2H2BE, KCNK3, KLHL4, LRRK2, NXPH3, PIGZ, SAMD5, ERINC2, and SIX6) related to the glioma prognosis were screened. The 16 selected genes were associated with the development of gliomas and carcinogenesis. The accuracy of an external validation data set of the fully connected neural network model from the two cohorts reached 95.5%. Our method has good potential capability in classifying WHO grade II and III gliomas into low-risk, medium-risk, and high-risk subgroups. The subgroups showed significant (P<0.01) differences in overall survival. Conclusion: This resulted in the identification of 16 genes that were related to the prognosis of gliomas. Here we developed a computational method to discriminate WHO grade II and III gliomas into three subgroups with distinct prognoses. The gene expressionbased method provides a reliable alternative to determine the prognosis of gliomas.
- 1 March 2022
- Published by Bentham Science Publishers Ltd. in Current Medicinal Chemistry
- Vol. 29  (9) , 1622-1639
Identification of WHO II/III Gliomas by 16 Prognostic-related Gene Signatures using Machine Learning Methods
Background: It is found that the prognosis of gliomas of the same grade has large differences among World Health Organization (WHO) grade II and III in clinical observation. Therefore, a better understanding of the genetics and molecular mechanisms underlying WHO grade II and III gliomas is required, with the aim of developing a classification scheme at the molecular level rather than the conventional pathological morphology level. Method: We performed survival analysis combined with machine learning methods of Least Absolute Shrinkage and Selection Operator using expression datasets downloaded from the Chinese Glioma Genome Atlas as well as The Cancer Genome Atlas. Risk scores were calculated by the product of expression level of overall survival-related genes and their multivariate Cox proportional hazards regression coefficients. WHO grade II and III gliomas were categorized into the low-risk subgroup, medium-risk subgroup, and high-risk subgroup. We used the 16 prognostic-related genes as input features to build a classification model based on prognosis using a fully connected neural network. Gene function annotations were also performed. Results: The 16 genes (AKNAD1, C7orf13, CDK20, CHRFAM7A, CHRNA1, EFNB1, GAS1, HIST2H2BE, KCNK3, KLHL4, LRRK2, NXPH3, PIGZ, SAMD5, ERINC2, and SIX6) related to the glioma prognosis were screened. The 16 selected genes were associated with the development of gliomas and carcinogenesis. The accuracy of an external validation data set of the fully connected neural network model from the two cohorts reached 95.5%. Our method has good potential capability in classifying WHO grade II and III gliomas into low-risk, medium-risk, and high-risk subgroups. The subgroups showed significant (P<0.01) differences in overall survival. Conclusion: This resulted in the identification of 16 genes that were related to the prognosis of gliomas. Here we developed a computational method to discriminate WHO grade II and III gliomas into three subgroups with distinct prognoses. The gene expressionbased method provides a reliable alternative to determine the prognosis of gliomas.
- Open Access
Research Progress of Gliomas in Machine Learning
In the field of gliomas research, the broad availability of genetic and image information originated by computer technologies and the booming of biomedical publications has led to the advent of the big-data era. Machine learning methods were applied as possible approaches to speed up the data mining processes. In this article, we reviewed the present situation and future orientations of machine learning application in gliomas within the context of workflows to integrate analysis for precision cancer care. Publicly available tools or algorithms for key machine learning technologies in the literature mining for glioma clinical research were reviewed and compared. Further, the existing solutions of machine learning methods and their limitations in glioma prediction and diagnostics, such as overfitting and class imbalanced, were critically analyzed.
Open AccessResearch Progress of Gliomas in Machine Learning
In the field of gliomas research, the broad availability of genetic and image information originated by computer technologies and the booming of biomedical publications has led to the advent of the big-data era. Machine learning methods were applied as possible approaches to speed up the data mining processes. In this article, we reviewed the present situation and future orientations of machine learning application in gliomas within the context of workflows to integrate analysis for precision cancer care. Publicly available tools or algorithms for key machine learning technologies in the literature mining for glioma clinical research were reviewed and compared. Further, the existing solutions of machine learning methods and their limitations in glioma prediction and diagnostics, such as overfitting and class imbalanced, were critically analyzed.
- Open Access
- 17 June 2019
- Published by Springer Nature in BMC Medical Imaging
- Vol. 19  (1) , 1-9
A potential field segmentation based method for tumor segmentation on multi-parametric MRI of glioma cancer patients
Accurate segmentation of brain tumors is vital for the gross tumor volume (GTV) definition in radiotherapy. Functional MR images like apparent diffusion constant (ADC) and fractional anisotropy (FA) images can provide more comprehensive information for sensitive detection of the GTV. We synthesize anatomical and functional MRI for accurate and semi-automatic segmentation of GTVs and improvement of clinical efficiency. Four MR image sets including T1-weighted contrast-enhanced (T1C), T2-weighted (T2), apparent diffusion constant (ADC) and fractional anisotropy (FA) images of 5 glioma patients were acquired and registered. A new potential field segmentation (PFS) method was proposed based on the concept of potential field in physics. For T1C, T2 and ADC images, global potential field segmentation (global-PFS) was used on user defined region of interest (ROI) for rough segmentation and then morphologically processed for accurate delineation of the GTV. For FA images, white matter (WM) was removed using local potential field segmentation (local-PFS), and then tumor extent was delineated with region growing and morphological methods. The individual segmentations of multi-parametric images were ensembled into a fused segmentation, considered as final GTV. GTVs were compared with manually delineated ground truth and evaluated with segmentation quality measure (Q), Dice’s similarity coefficient (DSC) and Sensitivity and Specificity. Experimental study with the five patients’ data and new method showed that, the mean values of Q, DSC, Sensitivity and Specificity were 0.80 (±0.07), 0.88 (±0.04), 0.92 (±0.01) and 0.88 (±0.05) respectively. The global-PFS used on ROIs of T1C, T2 and ADC images can avoid interferences from skull and other non-tumor areas. Similarity to local-PFS on FA images, it can also reduce the time complexity as compared with the global-PFS on whole image sets. Efficient and semi-automatic segmentation of the GTV can be achieved with the new method. Combination of anatomical and functional MR images has the potential to provide new methods and ideas for target definition in radiotherapy.
- 17 June 2019
- Published by Springer Nature in BMC Medical Imaging
- Vol. 19  (1) , 1-9
Open AccessA potential field segmentation based method for tumor segmentation on multi-parametric MRI of glioma cancer patients
Accurate segmentation of brain tumors is vital for the gross tumor volume (GTV) definition in radiotherapy. Functional MR images like apparent diffusion constant (ADC) and fractional anisotropy (FA) images can provide more comprehensive information for sensitive detection of the GTV. We synthesize anatomical and functional MRI for accurate and semi-automatic segmentation of GTVs and improvement of clinical efficiency. Four MR image sets including T1-weighted contrast-enhanced (T1C), T2-weighted (T2), apparent diffusion constant (ADC) and fractional anisotropy (FA) images of 5 glioma patients were acquired and registered. A new potential field segmentation (PFS) method was proposed based on the concept of potential field in physics. For T1C, T2 and ADC images, global potential field segmentation (global-PFS) was used on user defined region of interest (ROI) for rough segmentation and then morphologically processed for accurate delineation of the GTV. For FA images, white matter (WM) was removed using local potential field segmentation (local-PFS), and then tumor extent was delineated with region growing and morphological methods. The individual segmentations of multi-parametric images were ensembled into a fused segmentation, considered as final GTV. GTVs were compared with manually delineated ground truth and evaluated with segmentation quality measure (Q), Dice’s similarity coefficient (DSC) and Sensitivity and Specificity. Experimental study with the five patients’ data and new method showed that, the mean values of Q, DSC, Sensitivity and Specificity were 0.80 (±0.07), 0.88 (±0.04), 0.92 (±0.01) and 0.88 (±0.05) respectively. The global-PFS used on ROIs of T1C, T2 and ADC images can avoid interferences from skull and other non-tumor areas. Similarity to local-PFS on FA images, it can also reduce the time complexity as compared with the global-PFS on whole image sets. Efficient and semi-automatic segmentation of the GTV can be achieved with the new method. Combination of anatomical and functional MR images has the potential to provide new methods and ideas for target definition in radiotherapy.
- 16 April 2018
- Published by Springer Nature in Apoptosis
- Vol. 23  (5-6) , 290-298
Feasibility study of stain-free classification of cell apoptosis based on diffraction imaging flow cytometry and supervised machine learning techniques
This study was to explore the feasibility of prediction and classification of cells in different stages of apoptosis with a stain-free method based on diffraction images and supervised machine learning. Apoptosis was induced in human chronic myelogenous leukemia K562 cells by cis-platinum (DDP). A newly developed technique of polarization diffraction imaging flow cytometry (p-DIFC) was performed to acquire diffraction images of the cells in three different statuses (viable, early apoptotic and late apoptotic/necrotic) after cell separation through fluorescence activated cell sorting with Annexin V-PE and SYTOX® Green double staining. The texture features of the diffraction images were extracted with in-house software based on the Gray-level co-occurrence matrix algorithm to generate datasets for cell classification with supervised machine learning method. Therefore, this new method has been verified in hydrogen peroxide induced apoptosis model of HL-60. Results show that accuracy of higher than 90% was achieved respectively in independent test datasets from each cell type based on logistic regression with ridge estimators, which indicated that p-DIFC system has a great potential in predicting and classifying cells in different stages of apoptosis.
- 16 April 2018
- Published by Springer Nature in Apoptosis
- Vol. 23  (5-6) , 290-298
Feasibility study of stain-free classification of cell apoptosis based on diffraction imaging flow cytometry and supervised machine learning techniques
This study was to explore the feasibility of prediction and classification of cells in different stages of apoptosis with a stain-free method based on diffraction images and supervised machine learning. Apoptosis was induced in human chronic myelogenous leukemia K562 cells by cis-platinum (DDP). A newly developed technique of polarization diffraction imaging flow cytometry (p-DIFC) was performed to acquire diffraction images of the cells in three different statuses (viable, early apoptotic and late apoptotic/necrotic) after cell separation through fluorescence activated cell sorting with Annexin V-PE and SYTOX® Green double staining. The texture features of the diffraction images were extracted with in-house software based on the Gray-level co-occurrence matrix algorithm to generate datasets for cell classification with supervised machine learning method. Therefore, this new method has been verified in hydrogen peroxide induced apoptosis model of HL-60. Results show that accuracy of higher than 90% was achieved respectively in independent test datasets from each cell type based on logistic regression with ridge estimators, which indicated that p-DIFC system has a great potential in predicting and classifying cells in different stages of apoptosis.
- Open Access
- 29 August 2017
- Published by Springer Nature in Scientific Reports
- Vol. 7  (1) , 1-10
A novel automatic quantification method for high-content screening analysis of DNA double strand-break response
High-content screening is commonly used in studies of the DNA damage response. The double-strand break (DSB) is one of the most harmful types of DNA damage lesions. The conventional method used to quantify DSBs is ÎłH2AX foci counting, which requires manual adjustment and preset parameters and is usually regarded as imprecise, time-consuming, poorly reproducible, and inaccurate. Therefore, a robust automatic alternative method is highly desired. In this manuscript, we present a new method for quantifying DSBs which involves automatic image cropping, automatic foci-segmentation and fluorescent intensity measurement. Furthermore, an additional function was added for standardizing the measurement of DSB response inhibition based on co-localization analysis. We tested the method with a well-known inhibitor of DSB response. The new method requires only one preset parameter, which effectively minimizes operator-dependent variations. Compared with conventional methods, the new method detected a higher percentage difference of foci formation between different cells, which can improve measurement accuracy. The effects of the inhibitor on DSB response were successfully quantified with the new method (p = 0.000). The advantages of this method in terms of reliability, automation and simplicity show its potential in quantitative fluorescence imaging studies and high-content screening for compounds and factors involved in DSB response.
- 29 August 2017
- Published by Springer Nature in Scientific Reports
- Vol. 7  (1) , 1-10
Open AccessA novel automatic quantification method for high-content screening analysis of DNA double strand-break response
High-content screening is commonly used in studies of the DNA damage response. The double-strand break (DSB) is one of the most harmful types of DNA damage lesions. The conventional method used to quantify DSBs is ÎłH2AX foci counting, which requires manual adjustment and preset parameters and is usually regarded as imprecise, time-consuming, poorly reproducible, and inaccurate. Therefore, a robust automatic alternative method is highly desired. In this manuscript, we present a new method for quantifying DSBs which involves automatic image cropping, automatic foci-segmentation and fluorescent intensity measurement. Furthermore, an additional function was added for standardizing the measurement of DSB response inhibition based on co-localization analysis. We tested the method with a well-known inhibitor of DSB response. The new method requires only one preset parameter, which effectively minimizes operator-dependent variations. Compared with conventional methods, the new method detected a higher percentage difference of foci formation between different cells, which can improve measurement accuracy. The effects of the inhibitor on DSB response were successfully quantified with the new method (p = 0.000). The advantages of this method in terms of reliability, automation and simplicity show its potential in quantitative fluorescence imaging studies and high-content screening for compounds and factors involved in DSB response.
- 24 July 2016
- Published by Frontiers Media SA in Experimental Biology and Medicine
- Vol. 241  (16) , 1751-1756
Original Research: Label-free detection for radiation-induced apoptosis in glioblastoma cells
Current flow cytometry (FCM) requires fluorescent dyes labeling cells which make the procedure costly and time consuming. This manuscript reports a feasibility study of detecting the cell apoptosis with a label-free method in glioblastoma cells. A human glioma cell line M059K was exposed to 8 Gy dose of radiation, which enables the cells to undergo radiation-induced apoptosis. The rates of apoptosis were studied at different time points post-irradiation with two different methods: FCM in combination with Annexin V-FITC/PI staining and a newly developed technique named polarization diffraction imaging flow cytometry. Totally 1000 diffraction images were acquired for each sample and the gray level co-occurrence matrix (GLCM) algorithm was used in morphological characterization of the apoptotic cells. Among the feature parameters extracted from each image pair, we found that the two GLCM parameters of angular second moment (ASM) and sum entropy (SumEnt) exhibit high sensitivities and consistencies as the apoptotic rates ( Pa) measured with FCM method. In addition, no significant difference exists between Pa and ASM_S, Pa and SumEnt_S, respectively ( P > 0.05). These results demonstrated that the new label-free method can detect cell apoptosis effectively. Cells can be directly used in the subsequent biochemical experiments as the structure and function of cells and biomolecules are well-preserved with this new method.
- 24 July 2016
- Published by Frontiers Media SA in Experimental Biology and Medicine
- Vol. 241  (16) , 1751-1756
Original Research: Label-free detection for radiation-induced apoptosis in glioblastoma cells
Current flow cytometry (FCM) requires fluorescent dyes labeling cells which make the procedure costly and time consuming. This manuscript reports a feasibility study of detecting the cell apoptosis with a label-free method in glioblastoma cells. A human glioma cell line M059K was exposed to 8 Gy dose of radiation, which enables the cells to undergo radiation-induced apoptosis. The rates of apoptosis were studied at different time points post-irradiation with two different methods: FCM in combination with Annexin V-FITC/PI staining and a newly developed technique named polarization diffraction imaging flow cytometry. Totally 1000 diffraction images were acquired for each sample and the gray level co-occurrence matrix (GLCM) algorithm was used in morphological characterization of the apoptotic cells. Among the feature parameters extracted from each image pair, we found that the two GLCM parameters of angular second moment (ASM) and sum entropy (SumEnt) exhibit high sensitivities and consistencies as the apoptotic rates ( Pa) measured with FCM method. In addition, no significant difference exists between Pa and ASM_S, Pa and SumEnt_S, respectively ( P > 0.05). These results demonstrated that the new label-free method can detect cell apoptosis effectively. Cells can be directly used in the subsequent biochemical experiments as the structure and function of cells and biomolecules are well-preserved with this new method.
- Open Access
- 1 February 2016
- Published by SPIE-Intl Soc Optical Eng in Optical Engineering
- Vol. 55  (2) , 023103
Quantitative assessment of image motion blur in diffraction images of moving biological cells
Motion blur (MB) presents a significant challenge for obtaining high-contrast image data from biological cells with a polarization diffraction imaging flow cytometry (p-DIFC) method. A new p-DIFC experimental system has been developed to evaluate the MB and its effect on image analysis using a time-delay-integration (TDI) CCD camera. Diffraction images of MCF-7 and K562 cells have been acquired with different speed-mismatch ratios and compared to characterize MB quantitatively. Frequency analysis of the diffraction images shows that the degree of MB can be quantified by bandwidth variations of the diffraction images along the motion direction. The analytical results were confirmed by the p-DIFC image data acquired at different speed-mismatch ratios and used to validate a method of numerical simulation of MB on blur-free diffraction images, which provides a useful tool to examine the blurring effect on diffraction images acquired from the same cell. These results provide insights on the dependence of diffraction image on MB and allow significant improvement on rapid biological cell assay with the p-DIFC method.
- 1 February 2016
- Published by SPIE-Intl Soc Optical Eng in Optical Engineering
- Vol. 55  (2) , 023103
Open AccessQuantitative assessment of image motion blur in diffraction images of moving biological cells
Motion blur (MB) presents a significant challenge for obtaining high-contrast image data from biological cells with a polarization diffraction imaging flow cytometry (p-DIFC) method. A new p-DIFC experimental system has been developed to evaluate the MB and its effect on image analysis using a time-delay-integration (TDI) CCD camera. Diffraction images of MCF-7 and K562 cells have been acquired with different speed-mismatch ratios and compared to characterize MB quantitatively. Frequency analysis of the diffraction images shows that the degree of MB can be quantified by bandwidth variations of the diffraction images along the motion direction. The analytical results were confirmed by the p-DIFC image data acquired at different speed-mismatch ratios and used to validate a method of numerical simulation of MB on blur-free diffraction images, which provides a useful tool to examine the blurring effect on diffraction images acquired from the same cell. These results provide insights on the dependence of diffraction image on MB and allow significant improvement on rapid biological cell assay with the p-DIFC method.