Tsung-Chun Lee
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
Tsung-Chun Lee has a total of 46 co-authors. Here are the top co-authors by number of shared publications:
Hsiu-Po Wang
National Taiwan University Hospital and College of Medicine, National Taiwan University
3 shared publications- Hsuan-Ting Chang
National Yunlin University of Science and Technology
3 shared publications - Clara Lavita Angelina
Department of Electronic Engineering, Politeknik Manufaktur Negeri Bangka Belitung
2 shared publications - Hsiu-Po Wang
Department of Internal Medicine, College of Medicine, National Taiwan University, Taipei, Taiwan
2 shared publications - Ju-Yeh Yang
Division of Nephrology, Department of Internal Medicine, Far Eastern Memorial Hospital, New Taipei City, Taiwan
2 shared publications - Andre Gerald Buret
Departments of Biological Sciences, Inflammation Research Network, University of Calgary, Calgary, Canada
1 shared publication
Latest Publications
- Open Access
- 3 September 2023
- Published by Association for Computing Machinery (ACM)
- p. 1
Multimodality-aware Intra-dialytic Hypotension Classifier: A Bilingual NLP Approach to Classify Dialysis Records
Nursing records, rich in free-text data, encapsulate valuable insights into patients' physiological status and physicians' diagnoses, offering more holistic perspectives than mere tabular data. The scrutiny of these free-text descriptions often enables physicians to rapidly assimilate the patients' historical and current health states, thereby facilitating accurate diagnosis and treatment. One particularly pressing issue is the identification of intra-dialytic hypotension (IDH) symptoms within these records. IDH, a sporadic yet potentially fatal complication during hemodialysis sessions for end-stage kidney disease patients, is diagnosed by a decrease of 20 mmHg in systolic blood pressure during hemodialysis or a drop of 10 mmHg in mean arterial pressure coupled with hypo-perfusion-related symptoms. In response to this, we have meticulously developed a Multi-modality-aware IDH Classifier (MMIDHC) using a natural language processing (NLP) model aimed at the classification of these records. Primarily, this model focuses on the classification of English-Chinese bilingual dialysis records concerning IDH. Composed of three distinct modules, the model's intricate processes are delineated in Figure 1. Free-text Pre-processing: To investigate the influence of long sentences and noise on classification, we implemented sentence segmentation. Additionally, we enhanced the contextual relationships and augmented the sample size by merging adjacent segments in both antegrade and retrograde fashions, as context ordered-shuffling. Free-text Feature Extraction: To expedite the process and avoid the need to rebuild NLP models, we adopted publicly available NLP models that support the Chinese language for feature extraction. In this context, we utilized the Multilingual Universal Sentence Encoder (mUSE) proposed by Yang et al. [3] and the Multilingual Sentence BERT (mSBERT) introduced by Reimers et al. [2]. The latter model builds upon Sentence BERT (SBERT) [1] and incorporates the Teacher-Student Model. Both these multilingual NLP models serve as encoders, parsing free-text, and converting natural language into feature vectors. The performance of these models, in terms of metric indicators, showed minimal differences at a fixed threshold. Nevertheless, the mSBERT encoder consistently outperformed mUSE in speed, approximately 20 times faster. Given the potential impact of encoder speed on the operation of the multimodule model, we selected mSBERT as the primary encoder. Classification Learning: We employed a Multilayer Perceptron (MLP) for classifying the extracted features. In order to improve the reliability of our classification, we incorporated the MLP's Intradialytic Hypotension (IDH) results with the blood pressure readings from the ongoing dialysis session into an ensemble model. This method facilitates more nuanced decision-making. For example, if the MLP detects IDH symptoms (Yes), but the blood pressure reading is normal (False), we classify the result as False owing to lack of fulfilling definition of IDH. However, if both the MLP and the blood pressure readings suggest a normal state, we categorize the dialysis record as Non-IDH. Our baseline model has achieved encouraging results across all metrics: Accuracy 0.951, Recall 0.955, Precision 0.913, and F1-score 0.932. For more information, please visit our website: https://github.com/IlikeBB/Far-Eastern-Memorial-Hospital-IDH_RecommendationSystem_Project.
- 3 September 2023
- Published by Association for Computing Machinery (ACM)
- p. 1
Open AccessMultimodality-aware Intra-dialytic Hypotension Classifier: A Bilingual NLP Approach to Classify Dialysis Records
Nursing records, rich in free-text data, encapsulate valuable insights into patients' physiological status and physicians' diagnoses, offering more holistic perspectives than mere tabular data. The scrutiny of these free-text descriptions often enables physicians to rapidly assimilate the patients' historical and current health states, thereby facilitating accurate diagnosis and treatment. One particularly pressing issue is the identification of intra-dialytic hypotension (IDH) symptoms within these records. IDH, a sporadic yet potentially fatal complication during hemodialysis sessions for end-stage kidney disease patients, is diagnosed by a decrease of 20 mmHg in systolic blood pressure during hemodialysis or a drop of 10 mmHg in mean arterial pressure coupled with hypo-perfusion-related symptoms. In response to this, we have meticulously developed a Multi-modality-aware IDH Classifier (MMIDHC) using a natural language processing (NLP) model aimed at the classification of these records. Primarily, this model focuses on the classification of English-Chinese bilingual dialysis records concerning IDH. Composed of three distinct modules, the model's intricate processes are delineated in Figure 1. Free-text Pre-processing: To investigate the influence of long sentences and noise on classification, we implemented sentence segmentation. Additionally, we enhanced the contextual relationships and augmented the sample size by merging adjacent segments in both antegrade and retrograde fashions, as context ordered-shuffling. Free-text Feature Extraction: To expedite the process and avoid the need to rebuild NLP models, we adopted publicly available NLP models that support the Chinese language for feature extraction. In this context, we utilized the Multilingual Universal Sentence Encoder (mUSE) proposed by Yang et al. [3] and the Multilingual Sentence BERT (mSBERT) introduced by Reimers et al. [2]. The latter model builds upon Sentence BERT (SBERT) [1] and incorporates the Teacher-Student Model. Both these multilingual NLP models serve as encoders, parsing free-text, and converting natural language into feature vectors. The performance of these models, in terms of metric indicators, showed minimal differences at a fixed threshold. Nevertheless, the mSBERT encoder consistently outperformed mUSE in speed, approximately 20 times faster. Given the potential impact of encoder speed on the operation of the multimodule model, we selected mSBERT as the primary encoder. Classification Learning: We employed a Multilayer Perceptron (MLP) for classifying the extracted features. In order to improve the reliability of our classification, we incorporated the MLP's Intradialytic Hypotension (IDH) results with the blood pressure readings from the ongoing dialysis session into an ensemble model. This method facilitates more nuanced decision-making. For example, if the MLP detects IDH symptoms (Yes), but the blood pressure reading is normal (False), we classify the result as False owing to lack of fulfilling definition of IDH. However, if both the MLP and the blood pressure readings suggest a normal state, we categorize the dialysis record as Non-IDH. Our baseline model has achieved encouraging results across all metrics: Accuracy 0.951, Recall 0.955, Precision 0.913, and F1-score 0.932. For more information, please visit our website: https://github.com/IlikeBB/Far-Eastern-Memorial-Hospital-IDH_RecommendationSystem_Project.
- Open Access
- 29 March 2023
- Published by MDPI AG in Diagnostics
- Vol. 13 (7) , 1289
Deep-Learning-Enabled Computer-Aided Diagnosis in the Classification of Pancreatic Cystic Lesions on Confocal Laser Endomicroscopy
Accurate classification of pancreatic cystic lesions (PCLs) is important to facilitate proper treatment and to improve patient outcomes. We utilized the convolutional neural network (CNN) of VGG19 to develop a computer-aided diagnosis (CAD) system in the classification of subtypes of PCLs in endoscopic ultrasound-guided needle-based confocal laser endomicroscopy (nCLE). From a retrospectively collected 22,424 nCLE video frames (50 videos) as the training/validation set and 11,047 nCLE video frames (18 videos) as the test set, we developed and compared the diagnostic performance of three CNNs with distinct methods of designating the region of interest. The diagnostic accuracy for subtypes of PCLs by CNNs with manual, maximal rectangular, and U-Net algorithm-designated ROIs was 100%, 38.9%, and 66.7% on a per-video basis and 88.99%, 73.94%, and 76.12% on a per-frame basis, respectively. Our per-frame analysis suggested differential levels of diagnostic accuracy among the five subtypes of PCLs, where non-mucinous PCLs (serous cystic neoplasm: 93.11%, cystic neuroendocrine tumor: 84.31%, and pseudocyst: 98%) had higher diagnostic accuracy than mucinous PCLs (intraductal papillary mucinous neoplasm: 84.43% and mucinous cystic neoplasm: 86.1%). Our CNN demonstrated superior specificity compared to the state-of-the-art for the classification of mucinous PCLs (IPMN and MCN), with high specificity (94.3% and 92.8%, respectively) but low sensitivity (46% and 45.2%, respectively). This suggests the complimentary role of CNN-enabled CAD systems, especially for clinically suspected mucinous PCLs.
- 29 March 2023
- Published by MDPI AG in Diagnostics
- Vol. 13 (7) , 1289
Open AccessDeep-Learning-Enabled Computer-Aided Diagnosis in the Classification of Pancreatic Cystic Lesions on Confocal Laser Endomicroscopy
Accurate classification of pancreatic cystic lesions (PCLs) is important to facilitate proper treatment and to improve patient outcomes. We utilized the convolutional neural network (CNN) of VGG19 to develop a computer-aided diagnosis (CAD) system in the classification of subtypes of PCLs in endoscopic ultrasound-guided needle-based confocal laser endomicroscopy (nCLE). From a retrospectively collected 22,424 nCLE video frames (50 videos) as the training/validation set and 11,047 nCLE video frames (18 videos) as the test set, we developed and compared the diagnostic performance of three CNNs with distinct methods of designating the region of interest. The diagnostic accuracy for subtypes of PCLs by CNNs with manual, maximal rectangular, and U-Net algorithm-designated ROIs was 100%, 38.9%, and 66.7% on a per-video basis and 88.99%, 73.94%, and 76.12% on a per-frame basis, respectively. Our per-frame analysis suggested differential levels of diagnostic accuracy among the five subtypes of PCLs, where non-mucinous PCLs (serous cystic neoplasm: 93.11%, cystic neuroendocrine tumor: 84.31%, and pseudocyst: 98%) had higher diagnostic accuracy than mucinous PCLs (intraductal papillary mucinous neoplasm: 84.43% and mucinous cystic neoplasm: 86.1%). Our CNN demonstrated superior specificity compared to the state-of-the-art for the classification of mucinous PCLs (IPMN and MCN), with high specificity (94.3% and 92.8%, respectively) but low sensitivity (46% and 45.2%, respectively). This suggests the complimentary role of CNN-enabled CAD systems, especially for clinically suspected mucinous PCLs.
- Open Access
- 30 August 2013
- Published by Springer Nature in Gut Pathogens
- Vol. 5 (1) , 26
Persistent gut barrier damage and commensal bacterial influx following eradication of Giardia infection in mice
Background: Recent studies ofGiardia lambliaoutbreaks have indicated that 40–80% of infected patients experience long-lasting functional gastrointestinal disorders after parasitic clearance. Our aim was to assess changes in the intestinal barrier and spatial distribution of commensal bacteria in the post-clearance phase ofGiardiainfection.Methods: Mice were orogastrically inoculated withG. lambliatrophozoites (strain GS/M) or pair-fed with saline and were sacrificed on post-infective (PI) days 7 (colonization phase) and 35 (post-clearance phase). Gut epithelial barrier function was assessed by Western blotting for occludin cleavage and luminal-to-serosal macromolecular permeability. Gut-associated, superficial adherent, and mucosal endocytosed bacteria were measured by agar culturing and were examined by fluorescencein situhybridization. Intracellular bacteria cultured from isolated mucosal cells were characterized by 16S rDNA sequencing. Neutrophil-specific esterase staining, a myeloperoxidase activity assay, and enzyme-linked immunosorbent assays for cytokine concentrations were used to verify intestinal tissue inflammation.Results: Tight junctional damage was detected in the intestinal mucosa ofGiardia-infected mice on PI days 7 and 35. Although intestinal bacterial overgrowth was evident only during parasite colonization (PI day 7), enhanced mucosal adherence and endocytosis of bacteria were observed on PI days 7 and 35. Multiple bacterial strains, includingBacillus,Lactobacillus,Staphylococcus, andPhenylobacterium, penetrated the gut mucosa in the post-infective phase. The mucosal influx of bacteria coincided with increases in neutrophil infiltration and myeloperoxidase activity on PI days 7 and 35. Elevated intestinal IFNγ, TNFα, and IL-1β levels also were detected on PI day 35.Conclusions: Giardia-infected mice showed persistent tight junctional damage and bacterial penetration, accompanied by mucosal inflammation, after parasite clearance. These novel findings suggest that the host’s unresolved immune reactions toward its own microbiota, due to an impaired epithelial barrier, may partly contribute to the development of post-infective gut disorders.
- 30 August 2013
- Published by Springer Nature in Gut Pathogens
- Vol. 5 (1) , 26
Open AccessPersistent gut barrier damage and commensal bacterial influx following eradication of Giardia infection in mice
Background: Recent studies ofGiardia lambliaoutbreaks have indicated that 40–80% of infected patients experience long-lasting functional gastrointestinal disorders after parasitic clearance. Our aim was to assess changes in the intestinal barrier and spatial distribution of commensal bacteria in the post-clearance phase ofGiardiainfection.Methods: Mice were orogastrically inoculated withG. lambliatrophozoites (strain GS/M) or pair-fed with saline and were sacrificed on post-infective (PI) days 7 (colonization phase) and 35 (post-clearance phase). Gut epithelial barrier function was assessed by Western blotting for occludin cleavage and luminal-to-serosal macromolecular permeability. Gut-associated, superficial adherent, and mucosal endocytosed bacteria were measured by agar culturing and were examined by fluorescencein situhybridization. Intracellular bacteria cultured from isolated mucosal cells were characterized by 16S rDNA sequencing. Neutrophil-specific esterase staining, a myeloperoxidase activity assay, and enzyme-linked immunosorbent assays for cytokine concentrations were used to verify intestinal tissue inflammation.Results: Tight junctional damage was detected in the intestinal mucosa ofGiardia-infected mice on PI days 7 and 35. Although intestinal bacterial overgrowth was evident only during parasite colonization (PI day 7), enhanced mucosal adherence and endocytosis of bacteria were observed on PI days 7 and 35. Multiple bacterial strains, includingBacillus,Lactobacillus,Staphylococcus, andPhenylobacterium, penetrated the gut mucosa in the post-infective phase. The mucosal influx of bacteria coincided with increases in neutrophil infiltration and myeloperoxidase activity on PI days 7 and 35. Elevated intestinal IFNγ, TNFα, and IL-1β levels also were detected on PI day 35.Conclusions: Giardia-infected mice showed persistent tight junctional damage and bacterial penetration, accompanied by mucosal inflammation, after parasite clearance. These novel findings suggest that the host’s unresolved immune reactions toward its own microbiota, due to an impaired epithelial barrier, may partly contribute to the development of post-infective gut disorders.
- Open Access
- 26 April 2013
- Published by Springer Nature in BMC Nephrology
- Vol. 14 (1) , 97-8
Risk factors of short-term mortality after acute nonvariceal upper gastrointestinal bleeding in patients on dialysis: a population-based study
Background Impaired kidney function is an established predictor of mortality after acute nonvariceal upper gastrointestinal bleeding (ANVUGIB); however, which factors are associated with mortality after ANVUGIB among patients undergoing dialysis is unknown. We examined the associations among demographic characteristics, dialysis-specific features, and comorbid conditions with short-term mortality after ANVUGIB among patients on dialysis. Methods Design: Retrospective cohort study. Setting: United States Renal Data System (USRDS), a nation-wide registry of patients with end-stage renal disease. Participants: All ANVUGIB episodes identified by validated algorithms in Medicare-covered patients between 2003 and 2007. Measurements: Demographic characteristics and comorbid conditions from 1 year of billing claims prior to each bleeding event. We used logistic regression extended with generalized estimating equations methods to model the associations among risk factors and 30-day mortality following ANVUGIB events. Results From 2003 to 2007, we identified 40,016 eligible patients with 50,497 episodes of ANVUGIB. Overall 30-day mortality was 10.7% (95% CI: 10.4-11.0). Older age, white race, longer dialysis vintage, peritoneal dialysis (vs. hemodialysis), and hospitalized (vs. outpatient) episodes were independently associated with a higher risk of 30-day mortality. Most but not all comorbid conditions were associated with death after ANVUGIB. The joint ability of all factors captured to discriminate mortality was modest (c=0.68). Conclusions We identified a profile of risk factors for 30-day mortality after ANVUGIB among patients on dialysis that was distinct from what had been reported in non-dialysis populations. Specifically, peritoneal dialysis and more years since initiation of dialysis were independently associated with short-term death after ANVUGIB.
- 26 April 2013
- Published by Springer Nature in BMC Nephrology
- Vol. 14 (1) , 97-8
Open AccessRisk factors of short-term mortality after acute nonvariceal upper gastrointestinal bleeding in patients on dialysis: a population-based study
Background Impaired kidney function is an established predictor of mortality after acute nonvariceal upper gastrointestinal bleeding (ANVUGIB); however, which factors are associated with mortality after ANVUGIB among patients undergoing dialysis is unknown. We examined the associations among demographic characteristics, dialysis-specific features, and comorbid conditions with short-term mortality after ANVUGIB among patients on dialysis. Methods Design: Retrospective cohort study. Setting: United States Renal Data System (USRDS), a nation-wide registry of patients with end-stage renal disease. Participants: All ANVUGIB episodes identified by validated algorithms in Medicare-covered patients between 2003 and 2007. Measurements: Demographic characteristics and comorbid conditions from 1 year of billing claims prior to each bleeding event. We used logistic regression extended with generalized estimating equations methods to model the associations among risk factors and 30-day mortality following ANVUGIB events. Results From 2003 to 2007, we identified 40,016 eligible patients with 50,497 episodes of ANVUGIB. Overall 30-day mortality was 10.7% (95% CI: 10.4-11.0). Older age, white race, longer dialysis vintage, peritoneal dialysis (vs. hemodialysis), and hospitalized (vs. outpatient) episodes were independently associated with a higher risk of 30-day mortality. Most but not all comorbid conditions were associated with death after ANVUGIB. The joint ability of all factors captured to discriminate mortality was modest (c=0.68). Conclusions We identified a profile of risk factors for 30-day mortality after ANVUGIB among patients on dialysis that was distinct from what had been reported in non-dialysis populations. Specifically, peritoneal dialysis and more years since initiation of dialysis were independently associated with short-term death after ANVUGIB.
- 6 August 2009
- Published by Wolters Kluwer Health in American Journal of Gastroenterology
- Vol. 104 (8) , 2121-2122
Differential Diagnosis of Crohn's Disease and Intestinal Tuberculous by Enzyme-Linked Immunospot Assay for Interferon-γ
The American Journal of Gastroenterology is published by Nature Publishing Group (NPG) on behalf of the American College of Gastroenterology (ACG). Ranked the #1 clinical journal covering gastroenterology and hepatology*, The American Journal of Gastroenterology (AJG) provides practical and professional support for clinicians dealing with the gastroenterological disorders seen most often in patients. Published with practicing clinicians in mind, the journal aims to be easily accessible, organizing its content by topic, both online and in print. www.amjgastro.com, *2007 Journal Citation Report (Thomson Reuters, 2008)
- 6 August 2009
- Published by Wolters Kluwer Health in American Journal of Gastroenterology
- Vol. 104 (8) , 2121-2122
Differential Diagnosis of Crohn's Disease and Intestinal Tuberculous by Enzyme-Linked Immunospot Assay for Interferon-γ
The American Journal of Gastroenterology is published by Nature Publishing Group (NPG) on behalf of the American College of Gastroenterology (ACG). Ranked the #1 clinical journal covering gastroenterology and hepatology*, The American Journal of Gastroenterology (AJG) provides practical and professional support for clinicians dealing with the gastroenterological disorders seen most often in patients. Published with practicing clinicians in mind, the journal aims to be easily accessible, organizing its content by topic, both online and in print. www.amjgastro.com, *2007 Journal Citation Report (Thomson Reuters, 2008)
- Open Access
- 15 October 2007
- Published by Oxford University Press (OUP) in Nephrology Dialysis Transplantation
- Vol. 23 (3) , 1005-1010
Peritoneal thickening is not inevitable in long-term peritoneal dialysis and is associated with peritoneal transport characteristics: a two-centre sonographic study
Background . The peritoneum is subject to alterations in the life-long course of peritoneal dialysis (PD). Studies of the parietal peritoneum by non-invasive ultrasonography in PD patients are limited. We hypothesize that a prolonged PD duration is associated with a thicker peritoneum on ultrasonography and alterations in Doppler indexes of mesenteric vessels. Methods . We recruited two groups of patients, 18 who had >7 years of PD and 18 who had Results . We found no significant difference between two groups of PD patients in peritoneal thickness and in Doppler indexes. However, our univariate and multivariate analysis indicated that peritoneal thickness is associated with peritoneal transport characteristics (dialysate/plasma creatinine) but not with age, duration of dialysis, body height, body weight or Doppler index. The peritoneum is significantly thicker in rapid transporters than in slow transporters (RUQ: 0.59 ± 0.40 mm versus 0.27 ± 0.29 mm, P = 0.01; LUQ: 0.60 ± 0.40 mm versus 0.27 ± 0.32 mm, P = 0.016; LQ: 1.07 ± 0.85 mm versus 0.48 ± 0.53 mm, P = 0.026). In addition, rapid transporters have a marginally lower Doppler resistive index of the superior mesenteric artery (0.87 ± 0.08 versus 0.90 ± 0.10, P = 0.028). Conclusions . Our data showed that peritoneal thickening is not inevitable in long-term PD patients. Sonographic thickness in the parietal peritoneum is associated with transport characteristics. Rapid transporters have a significantly thicker peritoneum. The Doppler index of mesenteric vessels had no association with PD duration or transport characteristics. Trans-abdominal ultrasonography is non-invasive and useful in evaluating peritoneal characteristics of PD patients.
- 15 October 2007
- Published by Oxford University Press (OUP) in Nephrology Dialysis Transplantation
- Vol. 23 (3) , 1005-1010
Open AccessPeritoneal thickening is not inevitable in long-term peritoneal dialysis and is associated with peritoneal transport characteristics: a two-centre sonographic study
Background . The peritoneum is subject to alterations in the life-long course of peritoneal dialysis (PD). Studies of the parietal peritoneum by non-invasive ultrasonography in PD patients are limited. We hypothesize that a prolonged PD duration is associated with a thicker peritoneum on ultrasonography and alterations in Doppler indexes of mesenteric vessels. Methods . We recruited two groups of patients, 18 who had >7 years of PD and 18 who had Results . We found no significant difference between two groups of PD patients in peritoneal thickness and in Doppler indexes. However, our univariate and multivariate analysis indicated that peritoneal thickness is associated with peritoneal transport characteristics (dialysate/plasma creatinine) but not with age, duration of dialysis, body height, body weight or Doppler index. The peritoneum is significantly thicker in rapid transporters than in slow transporters (RUQ: 0.59 ± 0.40 mm versus 0.27 ± 0.29 mm, P = 0.01; LUQ: 0.60 ± 0.40 mm versus 0.27 ± 0.32 mm, P = 0.016; LQ: 1.07 ± 0.85 mm versus 0.48 ± 0.53 mm, P = 0.026). In addition, rapid transporters have a marginally lower Doppler resistive index of the superior mesenteric artery (0.87 ± 0.08 versus 0.90 ± 0.10, P = 0.028). Conclusions . Our data showed that peritoneal thickening is not inevitable in long-term PD patients. Sonographic thickness in the parietal peritoneum is associated with transport characteristics. Rapid transporters have a significantly thicker peritoneum. The Doppler index of mesenteric vessels had no association with PD duration or transport characteristics. Trans-abdominal ultrasonography is non-invasive and useful in evaluating peritoneal characteristics of PD patients.
- Open Access
- 15 August 2006
- Published by Oxford University Press (OUP) in Clinical Infectious Diseases
- Vol. 43 (4) , 537-539
A Man with Jaundice and Abdominal Tumor
Diagnosis: Extrahepatic cholangiocarcinoma and liver flukes (Clonorchis sinensis).
- 15 August 2006
- Published by Oxford University Press (OUP) in Clinical Infectious Diseases
- Vol. 43 (4) , 537-539
Open AccessA Man with Jaundice and Abdominal Tumor
Diagnosis: Extrahepatic cholangiocarcinoma and liver flukes (Clonorchis sinensis).