Article
Computer Science, Artificial Intelligence
Omer Sumer, Patricia Goldberg, Sidney DMello, Peter Gerjets, Ulrich Trautwein, Enkelejda Kasneci
Summary: Student engagement is crucial for effective learning and teaching. This study focuses on analyzing student engagement in real classroom settings using computer vision methods. The researchers collected audiovisual recordings of secondary school classes and used deep embeddings for attentional and affective features to train engagement classifiers. The best performing classifiers achieved high AUC scores, with attention-based features being more effective than affective features. Additionally, personalization using just 60 seconds of person-specific data resulted in improved performance.
IEEE TRANSACTIONS ON AFFECTIVE COMPUTING
(2023)
Review
Computer Science, Information Systems
Muhammad Hameed Siddiqi, Khalil Khan, Rehan Ullah Khan, Amjad Alsirhani
Summary: Human face image analysis using machine learning is a crucial field in computer vision. This survey paper comprehensively reviews methods in both controlled and uncontrolled conditions, highlighting their strengths and weaknesses. It also compares the performance of previous methods on standard datasets and proposes future research directions.
Article
Engineering, Electrical & Electronic
Guoying Zhao, Xiaobai Li, Yante Li, Matti Pietikainen
Summary: Micro-expression (ME) is an involuntary, fleeting, and subtle facial expression that can provide essential clues to people's true feelings. In recent years, ME analysis, especially automatic ME analysis in computer vision, has gained much attention due to its practical importance. This survey provides a comprehensive review of ME development in the field of computer vision, discussing various computational ME analysis methods and future directions.
PROCEEDINGS OF THE IEEE
(2023)
Article
Computer Science, Theory & Methods
Manal Alamir, Manal Alghamdi
Summary: This article discusses recent advancements in generative adversarial networks (GANs), particularly in the medical field. It analyzes and summarizes different medical imaging modalities and the basic theory of GANs, determines evaluation metrics and training issues, classifies extension models of GANs, and illustrates applications of GANs in medical images such as cross-modality, augmentation, detection, classification, and reconstruction. Finally, it discusses future research directions.
ACM COMPUTING SURVEYS
(2023)
Article
Computer Science, Artificial Intelligence
Alagesan Bhuvaneswari Ahadit, Ravi Kumar Jatoth
Summary: Facial expressions are widely used to recognize human emotions. However, designing an automatic facial expression recognition (FER) model faces challenges such as strong intra-class correlation between different emotions. This paper proposes a multi-input hybrid FER model that combines deep and hand-engineered features, which achieves improved accuracy in distinguishing facial expression patterns.
MACHINE VISION AND APPLICATIONS
(2022)
Article
Engineering, Biomedical
Yan Zhuang, Mark M. McDonald, Chad M. Aldridge, Mohamed Abul Hassan, Omar Uribe, Daniel Arteaga, Andrew M. Southerland, Gustavo K. Rohde
Summary: The study proposed a framework for facial weakness detection, with experimental results showing that the algorithm performed well in accuracy, sensitivity, and specificity, and improved model transparency and interpretability. By implementing a proof-of-concept prototype, the feasibility of an inexpensive solution for facial weakness detection was demonstrated.
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING
(2021)
Article
Health Care Sciences & Services
Danning Wu, Shi Chen, Yuelun Zhang, Huabing Zhang, Qing Wang, Jianqiang Li, Yibo Fu, Shirui Wang, Hongbo Yang, Hanze Du, Huijuan Zhu, Hui Pan, Zhen Shen
Summary: This study summarized and quantitatively analyzed the performance of AI technology in diagnosing heterogeneous diseases based on facial features for the first time. The results showed that the facial recognition intensity (FRI) had a significant impact on diagnostic accuracy, and increasing training size and utilizing deep learning models could improve accuracy for diseases with low FRI.
JOURNAL OF PERSONALIZED MEDICINE
(2021)
Article
Chemistry, Analytical
Mukhriddin Mukhiddinov, Oybek Djuraev, Farkhod Akhmedov, Abdinabi Mukhamadiyev, Jinsoo Cho
Summary: Current AI systems for determining emotions rely on facial features, but low-light images and masks hinder accuracy. This study proposes a method that enhances low-light images and uses upper facial features with a convolutional neural network for emotion recognition. The approach utilizes the AffectNet dataset, achieving 69.3% accuracy.
Article
Computer Science, Artificial Intelligence
Efstratios Kakaletsis, Nikos Nikolaidis
Summary: In this paper, an active approach for face recognition is proposed, which utilizes both real and synthesized views to improve the performance of recognition. Experimental results verify the superiority of this method in three datasets.
MACHINE VISION AND APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Argyrios Zafeiriou, Athanasios Kallipolitis, Ilias Maglogiannis
Summary: With the increase in accuracy of AI systems, complexity has also risen. This study examines the effectiveness of ensembling deep convolutional neural networks to leverage explanations using saliency maps. The proposed methodology allows for the aggregation of saliency maps from multiple models, improving interpretability. The results suggest that combining information through this aggregation scheme enhances interpretability and offer useful insights for future work.
Review
Medicine, General & Internal
Sukhveer Singh Sandhu, Hamed Taheri Gorji, Pantea Tavakolian, Kouhyar Tavakolian, Alireza Akhbardeh, Fabiano Bini
Summary: This review focuses on the different applications of Federated Learning (FL) to medical imaging datasets. It provides a detailed description of FL architecture and models, compares the performance of FL models with traditional Machine Learning (ML) models, and discusses security benefits and privacy-preserving techniques. The review also highlights the challenges in deploying FL in medical imaging applications and provides recommendations for future directions.
Article
Computer Science, Artificial Intelligence
Bo Yang, Jianming Wu, Kazushi Ikeda, Gen Hattori, Masaru Sugano, Yusuke Iwasawa, Yutaka Matsuo
Summary: This paper proposes a method based on face parsing and vision Transformer to improve the accuracy of face-mask-aware facial expression recognition. The method re-trains a face parsing model and utilizes a cross attention mechanism-based FER classifier to handle both occluded and non-occluded facial regions effectively.
PATTERN RECOGNITION LETTERS
(2022)
Article
Telecommunications
Sumindar Kaur Saini, Niharika Thakur, Mamta Juneja
Summary: Radiomics is a rapidly emerging field in medical applications that analyzes large-scale radiological images associated with biology. It relies on feature extraction and is an extension of computer aided diagnosis (CAD). The high number of features used in radiomics makes it one of the best approaches for image analysis, as these features can be used in multiple modalities. This paper provides a detailed literature review on radiomics in oncology, highlighting the gaps and challenges faced in its future development.
WIRELESS PERSONAL COMMUNICATIONS
(2023)
Article
Ophthalmology
Ilke Bahceci Simsek, Can Sirolu
Summary: A computer vision algorithm was used to evaluate postoperative changes in patients who underwent upper eyelid blepharoplasty surgery, providing an objective and standardized method for assessment.
GRAEFES ARCHIVE FOR CLINICAL AND EXPERIMENTAL OPHTHALMOLOGY
(2021)
Article
Chemistry, Analytical
Julian Caba, Jesus Barba, Fernando Rincon, Jose Antonio de la Torre, Soledad Escolar, Juan Carlos Lopez
Summary: This research introduces a novel algorithm for masked face recognition using facial spectral features extracted from hyperspectral images. The algorithm utilizes computer vision techniques, specifically Histogram of Oriented Gradients, to extract spectral features from different regions of interests. Support Vector Machines with custom kernels, based on cosine similarity and Euclidean distance, are trained to classify unknown faces based on the distance of visible facial spectral features that are not occluded by a face mask or scarf.
Meeting Abstract
Orthopedics
A. Tiulpin, J. Thevenot, J. Hirvasniemi, J. Niinimaki, S. Saarakkala
OSTEOARTHRITIS AND CARTILAGE
(2018)
Article
Orthopedics
S. Kauppinen, S. S. Karhula, J. Thevenot, T. Ylitalo, L. Rieppo, I. Kestila, M. Haapea, I. Hadjab, M. A. Finnila, E. Quenneville, M. Garon, H. K. Gahunia, K. P. H. Pritzker, M. D. Buschmann, S. Saarakkala, H. J. Nieminen
OSTEOARTHRITIS AND CARTILAGE
(2019)
Article
Engineering, Biomedical
Jukka Hirvasniemi, Jaakko Niinimaeki, Jerome Thevenot, Simo Saarakkala
ANNALS OF BIOMEDICAL ENGINEERING
(2019)
Article
Veterinary Sciences
B. R. Shakya, A. Tiulpin, S. Saarakkala, S. Turunen, J. Thevenot
EQUINE VETERINARY JOURNAL
(2020)
Meeting Abstract
Orthopedics
A. Tiulpin, S. Klein, S. Bierma-Zeinstra, J. Thevenot, J. van Meurs, E. Oei, S. Saarakkala
OSTEOARTHRITIS AND CARTILAGE
(2019)
Article
Engineering, Biomedical
S. S. Karhula, M. A. J. Finnila, S. J. O. Rytky, D. M. Cooper, J. Thevenot, M. Valkealahti, K. P. H. Pritzker, M. Haapea, A. Joukainen, P. Lehenkari, H. Kroger, R. K. Korhonen, H. J. Nieminen, S. Saarakkala
ANNALS OF BIOMEDICAL ENGINEERING
(2020)
Article
Multidisciplinary Sciences
Aleksei Tiulpin, Stefan Klein, Sita M. A. Bierma-Zeinstra, Jerome Thevenot, Esa Rahtu, Joyce van Meurs, Edwin H. G. Oei, Simo Saarakkala
SCIENTIFIC REPORTS
(2019)
Article
Neurosciences
Tiina Petaisto, David Vicente, Kari A. Makela, Mikko A. Finnila, Ilkka Miinalainen, Jarkko Koivunen, Valerio Izzi, Mari Aikio, Sanna-Maria Karppinen, Raman Devarajan, Jerome Thevenot, Karl-Heinz Herzig, Ritva Heljasvaara, Taina Pihlajaniemi
JOURNAL OF PHYSIOLOGY-LONDON
(2020)
Article
Dentistry, Oral Surgery & Medicine
Riina Rytivaara, Ritva Napankangas, Tiina Kainulainen, Annina Sipola, Soili Kallio-Pulkkinen, Aune Raustia, Jerome Thevenot
Summary: Thermography could be a potential diagnostic tool for female TMD patients, as specific facial areas' thermal information could help differentiate TMD patients from non-TMD patients and quantify the pain associated with TMD.
CRANIO-THE JOURNAL OF CRANIOMANDIBULAR & SLEEP PRACTICE
(2021)
Article
Orthopedics
A. Peuna, J. Thevenot, S. Saarakkala, M. T. Nieminen, E. Lammentausta
Summary: This study introduced the application of local binary pattern and gray-level co-occurrence matrix texture analysis in osteoarthritis research, comparing the performance of different classification systems in discriminating OA patients from healthy controls. The results showed that LBP and GLCM were effective in distinguishing OA patients from controls, with classification models demonstrating high accuracy in OA assessment.
OSTEOARTHRITIS AND CARTILAGE
(2021)
Article
Multidisciplinary Sciences
Johanna A. Huhtakangas, Jere Huovinen, Sakari Laaksonen, Hanna-Marja Voipio, Olli Vuolteenaho, Mikko A. J. Finnila, Jerome Thevenot, Petri P. Lehenkari
Summary: This preliminary study suggests that calcipotriol may have an early disease-modifying effect in the rat ZIA model, reducing histological grade of synovitis a week after local injection, while dexamethasone did not show significant difference from the vehicle.
Article
Multidisciplinary Sciences
Mika T. Nevalainen, Olli Veikkola, Jerome Thevenot, Aleksei Tiulpin, Jukka Hirvasniemi, Jaakko Niinimaki, Simo S. Saarakkala
Summary: This study evaluated the acoustic emissions and kinematic instability of osteoarthritic knee joints, finding significant differences in parameters related to AE patterns, KI, BMI, and age between the OA and control groups in females. A predictive model for radiographic OA in females was constructed using selected AE signals, KI, age, and BMI, showing an improvement in accuracy compared to the reference model based on age and BMI. However, the predictive model did not show improvement in males, indicating that AE and KI provide complementary information for detecting radiographic knee OA in females.
SCIENTIFIC REPORTS
(2021)
Proceedings Paper
Engineering, Biomedical
V. K. O. Virtanen, J. Thevenot, A. Tiulpin, J. Hirvasniemi, J. Niinimaki, M. Nevalainen, S. Saarakkala
WORLD CONGRESS ON MEDICAL PHYSICS AND BIOMEDICAL ENGINEERING 2018, VOL 2
(2019)
Proceedings Paper
Engineering, Biomedical
Jerome Thevenot, Jukka Hirvasniemi, Mikko Finnila, Petri Lehenkari, Simo Saarakkala