Ufuk Bal
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
Ufuk Bal has a total of 25 co-authors. Here are the top co-authors by number of shared publications:
- Ahmet Çağdaş Seçkin
Computer Engineering Department, Aydın Adnan Menderes University, Aydın 09100, Türkiye
3 shared publications - Alkan Bal
Division of Pediatric Emergency Medicine, Department of Pediatrics, Manisa Celal Bayar University Faculty of Medicine, Manisa, Türkiye
3 shared publications Alkan Bal
Manisa Celal Bayar University Pediatrics Department, Manisa, Turkey
2 shared publications- Ebubekir Akkuş
Germencik Yamantürk Vocational School, Aydın Adnan Menderes University, 09010 Aydın, Türkiye
2 shared publications - Kubilay Muhammed Sunnetci
Department of Electrical and Electronics Engineering, Osmaniye Korkut Ata University, Osmaniye, Turkey
2 shared publications - Urs Utzinger
University of Arizona
2 shared publications
Latest Publications
- Open Access
- 24 September 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (19) , 9524
Explainable Cross-Dataset Violence Detection for Public Safety Using a Two-Stream I3D Network
Camera networks now cover almost every public space, but the footage they generate has long outpaced what any human team can watch in real time. That gap is exactly where automated violence detection earns its place. The hard part is ambiguity—a hug or a bit of play-fighting can look a lot like a real attack, and systems that get this wrong flood operators with false alarms. Since violence lives in motion as much as in appearance, we build GoogLeNet-based Inflated 3D (I3D) two-stream classifiers that read space and time together. Three variants are trained and validated on the Real Life Violence Situations (RLVS) dataset (2000 videos, 80/20 split), each differing in temporal depth and resolution. The strongest, I3D Network 3, reaches 98.75% in-domain validation accuracy, with an F1 score of 98.74% and a Matthews Correlation Coefficient of 97.51%, values that should be read as upper bounds because of the file-level split and the validation-based checkpoint selection. Numbers on the home dataset only tell part of the story, so we push every model onto the independent RWF-2000 set, untouched during training. Even under this strict zero-shot transfer, I3D Network 3 holds 79.75% accuracy, although it is seen that this single value hides a strongly asymmetric error profile. The recall of the violence class decreases to 65.70%, in other words 343 of 1000 violent clips are missed, while the recall of the peaceful class remains at 93.80%, and McNemar and paired t-tests confirm the gaps between networks are real, rather than noise. For this reason, this result is evaluated as partial generalization. The only published values obtained under the same RLVS to RWF-2000 zero-shot protocol are 68.76% and 74.68%, so the margin is real but narrow. To open the black box, 3D Grad-CAM produces separate spatial and motion explanations that show where and when the model sees violence. Finally, the best model is embedded in ViolenceNet, a lightweight interface that keeps a human in the loop. Considering the missed-incident rate measured under domain shift, ViolenceNet is positioned as an assisted-review and triage tool, rather than as an autonomous alarm system.
- 24 September 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (19) , 9524
Open AccessExplainable Cross-Dataset Violence Detection for Public Safety Using a Two-Stream I3D Network
Camera networks now cover almost every public space, but the footage they generate has long outpaced what any human team can watch in real time. That gap is exactly where automated violence detection earns its place. The hard part is ambiguity—a hug or a bit of play-fighting can look a lot like a real attack, and systems that get this wrong flood operators with false alarms. Since violence lives in motion as much as in appearance, we build GoogLeNet-based Inflated 3D (I3D) two-stream classifiers that read space and time together. Three variants are trained and validated on the Real Life Violence Situations (RLVS) dataset (2000 videos, 80/20 split), each differing in temporal depth and resolution. The strongest, I3D Network 3, reaches 98.75% in-domain validation accuracy, with an F1 score of 98.74% and a Matthews Correlation Coefficient of 97.51%, values that should be read as upper bounds because of the file-level split and the validation-based checkpoint selection. Numbers on the home dataset only tell part of the story, so we push every model onto the independent RWF-2000 set, untouched during training. Even under this strict zero-shot transfer, I3D Network 3 holds 79.75% accuracy, although it is seen that this single value hides a strongly asymmetric error profile. The recall of the violence class decreases to 65.70%, in other words 343 of 1000 violent clips are missed, while the recall of the peaceful class remains at 93.80%, and McNemar and paired t-tests confirm the gaps between networks are real, rather than noise. For this reason, this result is evaluated as partial generalization. The only published values obtained under the same RLVS to RWF-2000 zero-shot protocol are 68.76% and 74.68%, so the margin is real but narrow. To open the black box, 3D Grad-CAM produces separate spatial and motion explanations that show where and when the model sees violence. Finally, the best model is embedded in ViolenceNet, a lightweight interface that keeps a human in the loop. Considering the missed-incident rate measured under domain shift, ViolenceNet is positioned as an assisted-review and triage tool, rather than as an autonomous alarm system.
- 25 April 2026
- Published by Pamukkale University in Pamukkale University Journal of Engineering Sciences
- No. Advanced O
Uç bilişim ve kayan pencere analizi kullanarak yüz videosundan temassız hemoglobin tahmini
Bağlam— Hemoglobin, anemi taraması ve klinik karar destek sistemleri için önemli bir biyobelirteçtir. Bununla birlikte, standart ölçüm invazivdir ve saha ve afet senaryolarında hızlı, tekrarlanabilir değerlendirme için uygulanması zor olabilir. Bu durum, kaynak kısıtlı uç cihazlarda güvenilir bir şekilde çalışabilen temassız tahmin yaklaşımlarını motive etmektedir.Amaç— Bu çalışma, açık bir uç bilgi işlem bakış açısıyla yüz videolarından temassız periferik hemoglobin (SpHb) tahminini araştırmaktadır. Temel amaç, kayan pencere uzunluğunun, (i) tahmin doğruluğunu, (ii) ardışık çıktıların zamansal kararlılığını ve (iii) gömülü bir platformda güç/enerji maliyetini birlikte yöneten sistem düzeyinde bir tasarım parametresi olarak nasıl davrandığını nicelleştirmektir.Yöntem— Veri seti, her biri tek bir SpHb etiketiyle eşleştirilmiş 279 yüz videosu içermektedir. Her video beş bölüme ayrılmıştır ve 30, 60, 90 ve 120 karelik kayan pencere konfigürasyonları kullanılarak eğitim örnekleri oluşturulmuştur. Veri sızıntısını önlemek için, değerlendirme özne bağımsız bir bölme yöntemiyle yapılmıştır. Algoritmik olarak, 3D CNN temel modelleri hibrit uzamsal-zamansal mimarilerle (CNN–VAN–Transformer) karşılaştırılmıştır. Performans, regresyon metrikleri (RMSE/MAE) kullanılarak değerlendirilmiş ve farklı pencere uzunlukları altında çıktı kararlılığını tartışmak için temsili videolar üzerinde zaman serisi incelemesiyle desteklenmiştir.Sonuçlar—60 karelik segmentlerde, 3D CNN 0,4808'lik bir RMSE elde ederken, CNN–VAN–Transformer 0,7128'lik bir RMSE vermiştir. 90 karelik segmentlerde, CNN–VAN–Transformer 0,6946'lık bir RMSE ile en iyi doğruluğu sağlarken, 3D CNN 0,8894'lük bir RMSE elde etmiştir; V1-3DCNN 1,5828'e düşmüştür, bu da daha uzun bağlamın iyileşmeyi garanti etmediğini ve pencere-mimari etkileşiminin önemli olduğunu göstermektedir. Video düzeyindeki MAE, içeriğe ve hedef aralığına göre değişir: örneğin, en iyi MAE, HEM022 (0,350) ve HEM188 (0,150) için 120 karede, HEM086 (0,161) için 60 karede ve HEM028 (0,512) için 30 karede elde edilir; bu da konular/videolar arasında evrensel bir optimum pencere uzunluğunun olmadığını vurgular.Sonuç—ACS711 akım sensörü ve mikrodenetleyici tarafından tetiklenen kayıt kullanılarak Raspberry Pi 5 üzerinde yapılan kenar ölçümleri, pencere boyutuyla monoton bir enerji artışı göstermektedir: ortalama güç 2,481 W'tan (30 kare) 3,951 W'a (120 kare) ve ortalama enerji 1,029 J'den 5,047 J'ye yükselmektedir. Ölçülen maksimum enerji 7,14 J (HEM022, 120 kare), minimum ise 0,48 J'dir (HEM028, 30 kare); tepe güç 120 kare ayarında 6,80 W'a ulaşmaktadır.Bu bulgular, kayan pencere uzunluğunun sadece bir ayar düğmesi değil, pratik, konuşlandırılabilir SpHb tahmin sistemlerinde doğruluk, çıktı kararlılığı ve enerji bütçesini doğrudan dengeleyen birincil bir tasarım kararı olduğunu göstermektedir.
- 25 April 2026
- Published by Pamukkale University in Pamukkale University Journal of Engineering Sciences
- No. Advanced O
Uç bilişim ve kayan pencere analizi kullanarak yüz videosundan temassız hemoglobin tahmini
Bağlam— Hemoglobin, anemi taraması ve klinik karar destek sistemleri için önemli bir biyobelirteçtir. Bununla birlikte, standart ölçüm invazivdir ve saha ve afet senaryolarında hızlı, tekrarlanabilir değerlendirme için uygulanması zor olabilir. Bu durum, kaynak kısıtlı uç cihazlarda güvenilir bir şekilde çalışabilen temassız tahmin yaklaşımlarını motive etmektedir.Amaç— Bu çalışma, açık bir uç bilgi işlem bakış açısıyla yüz videolarından temassız periferik hemoglobin (SpHb) tahminini araştırmaktadır. Temel amaç, kayan pencere uzunluğunun, (i) tahmin doğruluğunu, (ii) ardışık çıktıların zamansal kararlılığını ve (iii) gömülü bir platformda güç/enerji maliyetini birlikte yöneten sistem düzeyinde bir tasarım parametresi olarak nasıl davrandığını nicelleştirmektir.Yöntem— Veri seti, her biri tek bir SpHb etiketiyle eşleştirilmiş 279 yüz videosu içermektedir. Her video beş bölüme ayrılmıştır ve 30, 60, 90 ve 120 karelik kayan pencere konfigürasyonları kullanılarak eğitim örnekleri oluşturulmuştur. Veri sızıntısını önlemek için, değerlendirme özne bağımsız bir bölme yöntemiyle yapılmıştır. Algoritmik olarak, 3D CNN temel modelleri hibrit uzamsal-zamansal mimarilerle (CNN–VAN–Transformer) karşılaştırılmıştır. Performans, regresyon metrikleri (RMSE/MAE) kullanılarak değerlendirilmiş ve farklı pencere uzunlukları altında çıktı kararlılığını tartışmak için temsili videolar üzerinde zaman serisi incelemesiyle desteklenmiştir.Sonuçlar—60 karelik segmentlerde, 3D CNN 0,4808'lik bir RMSE elde ederken, CNN–VAN–Transformer 0,7128'lik bir RMSE vermiştir. 90 karelik segmentlerde, CNN–VAN–Transformer 0,6946'lık bir RMSE ile en iyi doğruluğu sağlarken, 3D CNN 0,8894'lük bir RMSE elde etmiştir; V1-3DCNN 1,5828'e düşmüştür, bu da daha uzun bağlamın iyileşmeyi garanti etmediğini ve pencere-mimari etkileşiminin önemli olduğunu göstermektedir. Video düzeyindeki MAE, içeriğe ve hedef aralığına göre değişir: örneğin, en iyi MAE, HEM022 (0,350) ve HEM188 (0,150) için 120 karede, HEM086 (0,161) için 60 karede ve HEM028 (0,512) için 30 karede elde edilir; bu da konular/videolar arasında evrensel bir optimum pencere uzunluğunun olmadığını vurgular.Sonuç—ACS711 akım sensörü ve mikrodenetleyici tarafından tetiklenen kayıt kullanılarak Raspberry Pi 5 üzerinde yapılan kenar ölçümleri, pencere boyutuyla monoton bir enerji artışı göstermektedir: ortalama güç 2,481 W'tan (30 kare) 3,951 W'a (120 kare) ve ortalama enerji 1,029 J'den 5,047 J'ye yükselmektedir. Ölçülen maksimum enerji 7,14 J (HEM022, 120 kare), minimum ise 0,48 J'dir (HEM028, 30 kare); tepe güç 120 kare ayarında 6,80 W'a ulaşmaktadır.Bu bulgular, kayan pencere uzunluğunun sadece bir ayar düğmesi değil, pratik, konuşlandırılabilir SpHb tahmin sistemlerinde doğruluk, çıktı kararlılığı ve enerji bütçesini doğrudan dengeleyen birincil bir tasarım kararı olduğunu göstermektedir.
- Open Access
- 25 July 2025
- Published by MDPI AG in Biosensors
- Vol. 15 (8) , 485
Estimation of Total Hemoglobin (SpHb) from Facial Videos Using 3D Convolutional Neural Network-Based Regression
Hemoglobin plays a critical role in diagnosing various medical conditions, including infections, trauma, hemolytic disorders, and Mediterranean anemia, which is particularly prevalent in Mediterranean populations. Conventional measurement methods require blood sampling and laboratory analysis, which are often time-consuming and impractical during emergency situations with limited medical infrastructure. Although portable oximeters enable non-invasive hemoglobin estimation, they still require physical contact, posing limitations for individuals with circulatory or dermatological conditions. Additionally, reliance on disposable probes increases operational costs. This study presents a non-contact and automated approach for estimating total hemoglobin levels from facial video data using three-dimensional regression models. A dataset was compiled from 279 volunteers, with synchronized acquisition of facial video and hemoglobin values using a commercial pulse oximeter. After preprocessing, the dataset was divided into training, validation, and test subsets. Three 3D convolutional regression models, including 3D CNN, channel attention-enhanced 3D CNN, and residual 3D CNN, were trained, and the most successful model was implemented in a graphical interface. Among these, the residual model achieved the most favorable performance on the test set, yielding an RMSE of 1.06, an MAE of 0.85, and a Pearson correlation coefficient of 0.73. This study offers a novel contribution by enabling contactless hemoglobin estimation from facial video using 3D CNN-based regression techniques.
- 25 July 2025
- Published by MDPI AG in Biosensors
- Vol. 15 (8) , 485
Open AccessEstimation of Total Hemoglobin (SpHb) from Facial Videos Using 3D Convolutional Neural Network-Based Regression
Hemoglobin plays a critical role in diagnosing various medical conditions, including infections, trauma, hemolytic disorders, and Mediterranean anemia, which is particularly prevalent in Mediterranean populations. Conventional measurement methods require blood sampling and laboratory analysis, which are often time-consuming and impractical during emergency situations with limited medical infrastructure. Although portable oximeters enable non-invasive hemoglobin estimation, they still require physical contact, posing limitations for individuals with circulatory or dermatological conditions. Additionally, reliance on disposable probes increases operational costs. This study presents a non-contact and automated approach for estimating total hemoglobin levels from facial video data using three-dimensional regression models. A dataset was compiled from 279 volunteers, with synchronized acquisition of facial video and hemoglobin values using a commercial pulse oximeter. After preprocessing, the dataset was divided into training, validation, and test subsets. Three 3D convolutional regression models, including 3D CNN, channel attention-enhanced 3D CNN, and residual 3D CNN, were trained, and the most successful model was implemented in a graphical interface. Among these, the residual model achieved the most favorable performance on the test set, yielding an RMSE of 1.06, an MAE of 0.85, and a Pearson correlation coefficient of 0.73. This study offers a novel contribution by enabling contactless hemoglobin estimation from facial video using 3D CNN-based regression techniques.
- 22 November 2023
- Published by Springer Nature in Physical and Engineering Sciences in Medicine
- Vol. 47 (1) , 109-117
A deep learning feature extraction-based hybrid approach for detecting pediatric pneumonia in chest X-ray images
Pneumonia is a disease caused by bacteria, viruses, and fungi that settle in the alveolar sacs of the lungs and can lead to serious health complications in humans. Early detection of pneumonia is necessary for early treatment to manage and cure the disease. Recently, machine learning-based pneumonia detection methods have focused on pneumonia in adults. Machine learning relies on manual feature engineering, whereas deep learning can automatically detect and extract features from data. This study proposes a deep learning feature extraction-based hybrid approach that combines deep learning and machine learning to detect pediatric pneumonia, which is difficult to standardize. The proposed hybrid approach enhances the accuracy of detecting pediatric pneumonia and simplifies the approach by eliminating the requirement for advanced feature extraction. The experiments indicate that the hybrid approach using a Medium Neural Network based on AlexNet feature extraction achieved a 97.9% accuracy rate and 98.0% sensitivity rate. The results show that the proposed approach achieved higher accuracy rates than state-of-the-art approaches.
- 22 November 2023
- Published by Springer Nature in Physical and Engineering Sciences in Medicine
- Vol. 47 (1) , 109-117
A deep learning feature extraction-based hybrid approach for detecting pediatric pneumonia in chest X-ray images
Pneumonia is a disease caused by bacteria, viruses, and fungi that settle in the alveolar sacs of the lungs and can lead to serious health complications in humans. Early detection of pneumonia is necessary for early treatment to manage and cure the disease. Recently, machine learning-based pneumonia detection methods have focused on pneumonia in adults. Machine learning relies on manual feature engineering, whereas deep learning can automatically detect and extract features from data. This study proposes a deep learning feature extraction-based hybrid approach that combines deep learning and machine learning to detect pediatric pneumonia, which is difficult to standardize. The proposed hybrid approach enhances the accuracy of detecting pediatric pneumonia and simplifies the approach by eliminating the requirement for advanced feature extraction. The experiments indicate that the hybrid approach using a Medium Neural Network based on AlexNet feature extraction achieved a 97.9% accuracy rate and 98.0% sensitivity rate. The results show that the proposed approach achieved higher accuracy rates than state-of-the-art approaches.
- 9 November 2023
- Published by Springer Nature in Neural Computing and Applications
- Vol. 36 (4) , 1523-1537
Hyperparameter optimization of pre-trained convolutional neural networks using adolescent identity search algorithm
- 9 November 2023
- Published by Springer Nature in Neural Computing and Applications
- Vol. 36 (4) , 1523-1537
Hyperparameter optimization of pre-trained convolutional neural networks using adolescent identity search algorithm
- 4 March 2022
- Published by Taylor & Francis in Hemoglobin
- Vol. 46 (2) , 95-99
Noninvasive Hemoglobin Measurement Reduce Invasive Procedures in Thalassemia Patients
This study was conducted to investigate the agreement between laboratory hemoglobin (LabHb) measured in venous blood and noninvasive, spectrophotometric hemoglobin (SpHb) measurement and the usability of SpHb measurement in the transfusion decision-making in patients with thalassemia whose hemoglobin (Hb) was monitored by taking blood samples at frequent intervals and who were transfused. Cardiac pulse, oxygen saturation, Pleth variability index (PVI), and SpHb values were measured in patients who came to the hematology outpatient clinic for a control visit and whose Hb levels were planned to be measured. Venous blood samples were taken for LabHb measurement, which we accept as the gold standard. Cohen’s kappa value was calculated for the agreement between SpHb measurements and LabHb values. The relationship and predictability between both measurement methods were evaluated by Pearson correlation analysis, a modified Bland-Altman plot and the linear regression model. In the study conducted with a total of 110 children with thalassemia, a moderate level of agreement between the two measurement methods (kappa = 0.370, p < 0.0001) and a significantly high correlation between the two tests (r = 0.675) were found. The mean bias between the differences was found to be 0.3 g/dL (–1.27 to 1.86 g/dL). The sensitivity and the specificity of SpHb in identifying patients who needed transfusions (Hb <10.0 g/dL) were calculated as 92.2 and 57.1%, respectively. Our results suggest SpHb measurement may be used to screen anemia in hemodynamically stable hemoglobinopathy patients and even for transfusion decision-making with combination clinical findings.
- 4 March 2022
- Published by Taylor & Francis in Hemoglobin
- Vol. 46 (2) , 95-99
Noninvasive Hemoglobin Measurement Reduce Invasive Procedures in Thalassemia Patients
This study was conducted to investigate the agreement between laboratory hemoglobin (LabHb) measured in venous blood and noninvasive, spectrophotometric hemoglobin (SpHb) measurement and the usability of SpHb measurement in the transfusion decision-making in patients with thalassemia whose hemoglobin (Hb) was monitored by taking blood samples at frequent intervals and who were transfused. Cardiac pulse, oxygen saturation, Pleth variability index (PVI), and SpHb values were measured in patients who came to the hematology outpatient clinic for a control visit and whose Hb levels were planned to be measured. Venous blood samples were taken for LabHb measurement, which we accept as the gold standard. Cohen’s kappa value was calculated for the agreement between SpHb measurements and LabHb values. The relationship and predictability between both measurement methods were evaluated by Pearson correlation analysis, a modified Bland-Altman plot and the linear regression model. In the study conducted with a total of 110 children with thalassemia, a moderate level of agreement between the two measurement methods (kappa = 0.370, p < 0.0001) and a significantly high correlation between the two tests (r = 0.675) were found. The mean bias between the differences was found to be 0.3 g/dL (–1.27 to 1.86 g/dL). The sensitivity and the specificity of SpHb in identifying patients who needed transfusions (Hb <10.0 g/dL) were calculated as 92.2 and 57.1%, respectively. Our results suggest SpHb measurement may be used to screen anemia in hemodynamically stable hemoglobinopathy patients and even for transfusion decision-making with combination clinical findings.
- Open Access
- 27 June 2018
- Published by Mugla Sitki Kocman University in Mugla Journal of Science and Technology
- Vol. 4 (1) , 37-40
RETRACTED: CAPILLARY REFILL TIME MEASUREMENT DEVICE
Yükseköğretim Kurumları Bilimsel Araştırma ve Yayın Etiği Yönergesinin Madde 4/2-k uyarınca etik ihlal yapıldığına Dergi Kurulu olarak oy birliği ile karar verilen 10.22531/muglajsci.379126 DOI numaralı makale (Bal, U. (2018). “Capillary refill time measurement device”, Mugla Journal of Science and Technology, 4(1), 37-40.)” geri çekilmiştir.
- 27 June 2018
- Published by Mugla Sitki Kocman University in Mugla Journal of Science and Technology
- Vol. 4 (1) , 37-40
Open AccessRETRACTED: CAPILLARY REFILL TIME MEASUREMENT DEVICE
Yükseköğretim Kurumları Bilimsel Araştırma ve Yayın Etiği Yönergesinin Madde 4/2-k uyarınca etik ihlal yapıldığına Dergi Kurulu olarak oy birliği ile karar verilen 10.22531/muglajsci.379126 DOI numaralı makale (Bal, U. (2018). “Capillary refill time measurement device”, Mugla Journal of Science and Technology, 4(1), 37-40.)” geri çekilmiştir.
- Open Access
- 10 December 2014
- Published by Optica Publishing Group in Biomedical Optics Express
- Vol. 6 (1) , 86-97
Non-contact estimation of heart rate and oxygen saturation using ambient light
We propose a robust method for automated computation of heart rate (HR) from digital color video recordings of the human face. In order to extract photoplethysmographic signals, two orthogonal vectors of RGB color space are used. We used a dual tree complex wavelet transform based denoising algorithm to reduce artifacts (e.g. artificial lighting, movement, etc.). Most of the previous work on skin color based HR estimation performed experiments with healthy volunteers and focused to solve motion artifacts. In addition to healthy volunteers we performed experiments with child patients in pediatric intensive care units. In order to investigate the possible factors that affect the non-contact HR monitoring in a clinical environment, we studied the relation between hemoglobin levels and HR estimation errors. Low hemoglobin causes underestimation of HR. Nevertheless, we conclude that our method can provide acceptable accuracy to estimate mean HR of patients in a clinical environment, where the measurements can be performed remotely. In addition to mean heart rate estimation, we performed experiments to estimate oxygen saturation. We observed strong correlations between our SpO2 estimations and the commercial oximeter readings
- 10 December 2014
- Published by Optica Publishing Group in Biomedical Optics Express
- Vol. 6 (1) , 86-97
Open AccessNon-contact estimation of heart rate and oxygen saturation using ambient light
We propose a robust method for automated computation of heart rate (HR) from digital color video recordings of the human face. In order to extract photoplethysmographic signals, two orthogonal vectors of RGB color space are used. We used a dual tree complex wavelet transform based denoising algorithm to reduce artifacts (e.g. artificial lighting, movement, etc.). Most of the previous work on skin color based HR estimation performed experiments with healthy volunteers and focused to solve motion artifacts. In addition to healthy volunteers we performed experiments with child patients in pediatric intensive care units. In order to investigate the possible factors that affect the non-contact HR monitoring in a clinical environment, we studied the relation between hemoglobin levels and HR estimation errors. Low hemoglobin causes underestimation of HR. Nevertheless, we conclude that our method can provide acceptable accuracy to estimate mean HR of patients in a clinical environment, where the measurements can be performed remotely. In addition to mean heart rate estimation, we performed experiments to estimate oxygen saturation. We observed strong correlations between our SpO2 estimations and the commercial oximeter readings
- 1 October 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Non-contact heart rate estimation in pediatric intensive care units
In this study, we proposed a method for computation of heart rate (HR) from digital color video recordings of the human face. In order to extract photoplethysmographic signals, two orthogonal vectors of RGB (Red Green Blue) color space are used. We used a dual tree complex wavelet transform based denoising algorithm to reduce artifacts (e.g. artificial lighting, movement, etc.). Most of the previous work on skin color based HR estimation performed experiments with healthy volunteers and focused to solve motion artifacts. In addition to healthy volunteers we performed experiments with child patients in pediatric intensive care units. In order to investigate the possible factors that affect the non-contact HR monitoring in a clinical environment, we studied the relation between hemoglobin levels and HR estimation errors. Low hemoglobin causes underestimation of HR. Nevertheless, based on our results, it was concluded that our method could provide acceptable accuracy to estimate mean HR of patients in a clinical environment, especially given that the measurements can be performed remotely.
- 1 October 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Non-contact heart rate estimation in pediatric intensive care units
In this study, we proposed a method for computation of heart rate (HR) from digital color video recordings of the human face. In order to extract photoplethysmographic signals, two orthogonal vectors of RGB (Red Green Blue) color space are used. We used a dual tree complex wavelet transform based denoising algorithm to reduce artifacts (e.g. artificial lighting, movement, etc.). Most of the previous work on skin color based HR estimation performed experiments with healthy volunteers and focused to solve motion artifacts. In addition to healthy volunteers we performed experiments with child patients in pediatric intensive care units. In order to investigate the possible factors that affect the non-contact HR monitoring in a clinical environment, we studied the relation between hemoglobin levels and HR estimation errors. Low hemoglobin causes underestimation of HR. Nevertheless, based on our results, it was concluded that our method could provide acceptable accuracy to estimate mean HR of patients in a clinical environment, especially given that the measurements can be performed remotely.
- 2 July 2013
- Published by Springer Nature in Signal, Image and Video Processing
- Vol. 9 (4) , 787-799
A multiresolution approach for enhancement and denoising of microscopy images
- 2 July 2013
- Published by Springer Nature in Signal, Image and Video Processing
- Vol. 9 (4) , 787-799
A multiresolution approach for enhancement and denoising of microscopy images