Research on the Moral Responsibility of Face Recognition and Medical Diagnosis / 中国医学伦理学

The application of face recognition technology is gradually expanding to the medical field. It has been initially used in the medical diagnosis of endocrine diseases and genetic syndrome. This technology is expected to be used for the screening of genetic syndrome and endocrine diseases, shortening the delay period of disease diagnosis and helping the staging of endocrine diseases. However, this technology also has some moral risks, such as the risk of personal information security disclosure, the challenge to the future of mankind and the division of moral responsibility. This paper reflected on the dilemma of moral responsibility in the application of face recognition and medical diagnosis, and explored the two basic problems of "who is the subject of moral responsibility" and "the specific division of moral responsibility of different moral subjects" in face recognition and medical diagnosis. Finally, some suggestions on the moral responsibility in face recognition and medical diagnosis are put forward. The first is to determine the role of face recognition and medical diagnosis as an auxiliary category, and doctors are still the main medical subject; the second is to build the responsibility ethics mechanism and laws and regulations, the establishment of the responsibility system of face recognition and medical diagnosis is inseparable from the joint action of law and ethics.

Legal risks of face recognition technology in medical field / 中华医院管理杂志

Under the background of " Internet+ medical treatment" and the continuous deepening of face recognition technology research, the face recognition industry has continued to mature, and face recognition has been initially applied in medical fields such as hospital management, auxiliary medical care, and epidemic prevention and control. At the same time, face recognition technology brings problems including error risk, technical cracking risk, privacy risk, equality risk, abuse risk, and other issues in practice, which seriously threaten the personal and property rights and interests of the public. On the basis of summarizing the specific application direction of face recognition technology in hospitals, the authors sorted out the legal regulation of face recognition in China, and proposed that it should be based on technology research and development, strengthen the " gatekeeper" responsibility of medical institutions, improve legal system and recommendations for strengthening judicial leadership in order to improve the legal regulations of face recognition technology, reduce the risk of infringement by medical institutions in the application of face recognition technology, and protect the legitimate rights and interests of citizens.

Dados normativos de um conjunto de faces do Karolinska Directed Emotional Faces em uma amostra brasileira / Normative data of a faces set from Karolinska Directed Emotional Faces in a Brazilian sample / Datos normativos de un conjunto de caras de Karolinska Directed Emotional Faces en una muestra brasileña

O objetivo desta pesquisa foi obter dados normativos de um conjunto de faces do Karolinska Directed Emotional Faces (KDEF) em uma amostra brasileira. Para isso foi utilizada uma amostra não probabilística (por conveniência) de 100 participantes da cidade de João Pessoa-PB. Esses tinham idades entre 18 e 62 anos (M=21,6; DP=6,2), a maioria do sexo feminino (76%). Os resultados mostraram que os participantes obtiveram um percentual de acerto médio de 76,2%, de modo que expressões de Alegria (94.7%) e Surpresa (90.3%) foram as emoções mais facilmente identificáveis e Medo (40.65%) a mais difícil. Em relação às medidas de intensidade e valência, Nojo seguida de Surpresa obtiveram classificações mais intensas, e Alegria foi a única emoção com valência positiva alta. Esses achados foram bastante similares com àqueles relatados em pesquisas anteriores, fornecendo normas subjetivas de classificação mais adequadas às características da população brasileira.

The goal of this research was to obtain normative data of set of faces from Karolinska Directed Emotional Faces (KDEF) in a Brazilian sample. It was used a non-probabilistic sample (convenience sample) of 100 participants from the city of João Pessoa-PB. Age ranged from 18 to 62 years (M=21.6, SD=6.2), in which the majority was female (76%). Results showed that participants achieved a mean hit rate of 76.2%, and expressions of joy (94.7%) and surprise (90.3%) were the easiest emotions to identify and fear (40.65%) the most difficult. Regarding the measures of intensity and valence, disgust and surprise obtained more intense classifications, and joy was the only emotion with high positive valence. These findings were quite similar to those reported in previous research, providing subjective norms of classification suitable to the characteristics of the Brazilian population.

El objetivo de esta investigación fue obtener datos normativos de un conjunto de expresiones faciales del Karolinska Directed Emotional Faces (KDEF) en una muestra brasileña. Para ello, se utilizó una muestra no probabilística (por conveniencia) de 100 participantes de la ciudad de João Pessoa-PB. Ellos tenían edades entre 18 y 62 años (M=21,6, DP=6,2), la mayoría del sexo femenino (76%). Los resultados mostraron que los participantes obtuvieron un porcentaje de acierto promedio del 76,2%, de modo que las expresiones de Alegría (94.7%) y Sorpresa (90.3%) fueron las emociones más fáciles de identificar y Miedo (40.65%) la más difícil. Con relación a las medidas de intensidad y valencia, Asco seguida de Sorpresa obtuvieron clasificaciones más intensas, y Alegría fue la única emoción con valencia positiva alta. Estos hallazgos fueron muy similares con aquellos reportados en investigaciones anteriores, proporcionando normas subjetivas de clasificación más adecuadas a las características de la población brasileña.

Face holistic processing is modulated by high-motivation intensity:an event-related potential study / 中华行为医学与脑科学杂志

Objective To investigate the influence of high-motivated affect background on the holistic face processing and its neural mechanism.Methods Totally 26 subjects were instructed to perform composite-face process under the priming condition of three emotional pictures (food,neutral and fear).Results (1)Behavioral studies showed a significant interaction between emotion type and face combination(F(2,50) =5.37,P=0.008).Composite-face effect was observed in neutral emotional condition due to the fact that the recognition accuracy of the misaligned face ((90.21 ± 11.20)％) was significantly higher than that of the aligned face((84.83 ± 11.20) ％,t (25) =3.95,P=0.001).However,the composite-face effect disappeared under the condition of high-motivated emotion.That is,there was no significant difference in the recognition accuracy of the misaligned faces ((87.88± 10.28)％) and aligned faces ((85.12± 8.81)％,t(25)=1.93,P=0.065) under the condition of high-motivated negative emotion.Under the condition of high-motivated positive emotion,there was also no significant difference in the recognition accuracy between the misaligned faces ((87.92±11.00)％) and aligned faces((88.33±8.89)％,t(z5)=0.263,P=0.795).(2) According to the results of ERP,P3 components showed significant interaction between emotion type and face combination form(F(2,50) =3.23,P=0.048).Under the neutral emotional condition,the P3 amplitude of aligned faces ((3.46 ± 1.26) μV) was significantly higher than that of misaligned faces ((2.82 ± 1.43)μV,F (1,25)=5.28,P=0.03).In the case of high-motivated negative condition,there was no significant difference in P3amplitude between aligned faces ((3.21 ± 1.16) μV) and misaligned faces ((3.07 ± 1.22) μV,F (1,25) =0.53,P=0.476).The P3 amplitude of aligned faces((3.33±1.20) μV) and misaligned faces((3.42±1.40) μV) were also not significantly different under the condition of high-motivated positive affect (F(1,25)=0.23,P=0.638).Conclusion The study shows a high correlation between face processing and emotional processing.Additionally,the emotional motivation has a stronger influence on face processing.Regardless of the emotional valence,high-motivated intensity affect can cause local attention bias and reduce the holistic processing level of the face,which is mainly manifested in the later cognitive processing stage.

Serf-abnormality in patients with depression / 中华行为医学与脑科学杂志

Self-abnormality in patients with depression has been confirmed in a lot of research results.Negative self-referential processing of depression is associated with higher follow-up depressive symptoms and predicts the recurrence of major depressive episodes.Both self-processing and self-recognition are impaired in depression.The overgeneral autobiographical memory is also a risk factor for onset and maintenance of depression.This review examined self-abnormalities of depression from self-face recognition,self-referential processing and autobiographical memory and summarized the problems of current researches and the future research trends,and may provide insight to explore the self-abnormalities of depression patients comprehensively.

Application of human face recognition technology in genetic syndrome diagnosis / 中华实用儿科临床杂志

Genetic syndromes often involve craniofacial malformations,and certain syndromes are associated with a specific facial pattern such as Down syndrome.With the development of artificial intelligence in the medical field,face recognition technology has been successfully applied in the diagnosis of genetic syndrome,such as Down syndrome,Cornelia de Lange syndrome,22q11.2 deletion syndrome and Noonan syndrome.Some reports suggested that the detection rates of face recognition technology are higher than clinical specialists.In the future,face recognition technology is expected to be applied in the screening of and genetic syndromes,to help the diagnosis and to be applied in scientific research.

Time course of attention bias in face recognition of patients with depression / 中华行为医学与脑科学杂志

Objective@#To explore the characteristics of cognitive processing over time in recognition of different emotional faces in patients with depression.@*Methods@#Totally 29 patients with depression(patient group) and 30 normal control (control group)were selected.Subjects freely viewed positive, sad, threatening and neutral facial stimuli.Eye movement data were collected during the process of watching, and the differences of attention bias in time when the two groups of subjects recognized the four emotional faces were compared.The temporal variation trend of attention between the two groups was analyzed.@*Results@#There was no significant difference in the fixation duration between the patient group and the control group in 0-2 s((10.39±1.83)ms, (9.63±1.28)ms), 2-4 s((12.61±2.34)ms, (11.21±3.00)ms), 4-6 s ((12.26±3.11)ms, (10.47±4.06)ms)(all P>0.05). Within 6-8 s((12.64±3.42)ms, (8.90±4.02)ms), 8-10 s((12.55±4.19)ms, (9.36±4.02)ms), the fixation duration in the patient group was significantly longer than that in the control group (P<0.01). The fixation time of the positive faces in the patient group in 0-2 s ((12.07±2.65)ms, (14.50±3.54)ms), 2-4 s((10.69±3.26)ms, (14.66±5.25)ms), 4-6 s((10.11±4.24)ms, (15.43±8.02)ms), 6-8 s((9.39±5.12)ms, (16.88±9.60)ms), 8-10 s ((9.45±4.94)ms, (16.63±10.15)ms) was significantly lower than that in the control group (all P<0.01).@*Conclusion@#Patients with depression tend to pay attention to negative stimuli in the later stage of cognitive processing, while the attention to positive stimuli decreases in the early stage of cognitive processing.Cognitive processing of emotional stimuli in patients with depression is manifested as increased attention duration of negative stimuli and decreased attention duration of positive stimuli.

An Empirical Evaluation of the Local Texture Description Framework-Based Modified Local Directional Number Pattern with Various Classifiers for Face Recognition

ABSTRACT Texture is one of the chief characteristics of an image. In recent years, local texture descriptors have garnered attention among researchers in describing effective texture patterns to demarcate facial images. A feature descriptor titled Local Texture Description Framework-based Modified Local Directional Number pattern (LTDF_MLDN), capable of encoding texture patterns with pixels that lie at dissimilar regions, has been proposed recently to describe effective features for face images. However, the role of the descriptor can differ with different classifiers and distance metrics for diverse issues in face recognition. Hence, in this paper, an extensive evaluation of the LTDF_MLDN is carried out with an Extreme Learning Machine (ELM), a Support Vector Machine (SVM) and a Nearest Neighborhood Classifier (NNC) which uses Euclidian, Manhattan, Minkowski, G-statistics and chi-square dissimilarity metrics to illustrate differences in performance with respect to assorted issues in face recognition using six benchmark databases. Experimental results depict that the proposed descriptor is best suited with NNC for general case and expression variation, whereas, for the other facial variations ELM is found to produce better results.

An experimental study of facial emotional valence advantage in the process of recognition under stress in university students / 中华行为医学与脑科学杂志

Objective To investigate whether there is emotional memory advantage bias to negativc emotion facial expressions in stress.Methods A learning-recognition paradigm was used in this study,and 2 groups (stress/non stress)× 3 facial emotional valence (positive/neutral/negative) mixed designs were adopted to compare the reaction time deviations to different facial emotion expressions in stress participants and relaxing participants.Results It was shown by repeated measurement ANOVA that the main effect(E(2.59)=5.13,P=0.007) in facial emotional response time and the interaction effect (F(2.59) =6.02,P=0.003)between groups and facial emotional expression were found.The main effect in recognition of accuracy was observed(F(2.59)=14.47,P=0.000).Further analysis showed,for stress participants,that the reaction time to detect stimulus of positive facial expression ((645.84± 52.65) ms) was faster than negative ((721.46± 101.76) ms) and neutral faces ((692.99 ± 63.17) ms) (P＜0.05) and the recognition of the latter expressions was not significant.Interestingly,the accuracy to recognized of positive facial expression (0.68 ± 0.08) was the highest (P ＜ 0.05) for stress participants.Conclusion Positive faces advantage to facial expression is shown at the stage of face recognition in stress group.

Proposing the novelty classifier for face recognition

INTRODUCTION: Face recognition, one of the most explored themes in biometry, is used in a wide range of applications: access control, forensic detection, surveillance and monitoring systems, and robotic and human machine interactions. In this paper, a new classifier is proposed for face recognition: the novelty classifier. METHODS: The performance of a novelty classifier is compared with the performance of the nearest neighbor classifier. The ORL face image database was used. Three methods were employed for characteristic extraction: principal component analysis, bi-dimensional principal component analysis with dimension reduction in one dimension and bi-dimensional principal component analysis with dimension reduction in two directions. RESULTS: In identification mode, the best recognition rate with the leave-one-out strategy is equal to 100%. In the verification mode, the best recognition rate was also 100%. For the half-half strategy, the best recognition rate in the identification mode is equal to 98.5%, and in the verification mode, 88%. CONCLUSION: For face recognition, the novelty classifier performs comparable to the best results already published in the literature, which further confirms the novelty classifier as an important pattern recognition method in biometry.

O desenvolvimento do efeito da outra raça (EOR) em crianças: dos modelos de codificação de faces à emergência do EOR / The development of the other race effect (ORE) in children: from face coding mechanisms to the emergence of the ore

Em tarefas de reconhecimento facial, adultos e crianças apresentam dificuldade ao reconhecer faces de raças diferentes da sua. Esse efeito é conhecido como Efeito da Outra Raça (EOR) e tem sido consistentemente replicado em diversos estudos. O objetivo deste trabalho é apresentar uma revisão da literatura sobre alguns modelos teóricos que explicam sua emergência e desenvolvimento. Os modelos de codificação de faces baseado no Protótipo, e em Exemplares são a base para a explicação das diferenças na codificação entre faces da mesma raça e faces de outra raça. Uma revisão das recentes pesquisas sobre o EOR e como experiências com faces de outra raça, durante a primeira infância o influenciam, foi apresentada. Finalmente, pesquisas futuras relativas ao contexto cultural foram propostas, para melhor investigar o desenvolvimento do efeito da raça e dos modelos de codificação de faces.

In face recognition tasks, adults and children have difficulty recognizing faces from other races. This effect is known as the Other Race Effect (ORE) and has been consistently replicated in several studies. The aim of this paper is to present a review of some theoretical models that explain the ORE, its emergence and development. Face coding mechanisms based on Prototypes and Exemplars are the basis for the explanation of differences in coding between the same race and other race faces. It was presented a review of recent research on ORE and how the experiences with other race faces during early childhood influences the effect. Finally, further research on the cultural context has been proposed to better investigate the development of ORE and face coding mechanisms.

Prosopagnosia congênita: relato de caso / Congenital prosopagnosia: a case report

Prosopagnosia is a visual agnosia characterized by an inability to recognize previously known human faces and to learn new faces. The aim of this study was to present a forty-six year-old woman with congenital prosopagnosia, and to discuss the neural bases of perception and recognition of faces. The patients had a lifetime impairment in recognizing faces of family members, close friends, and even her own face in photos. She also had impairment in recognizing animals such as discriminating between cats and dogs. The patients basic visual skills showed impairment in identifying and recognizing the animal form perception on the coding subtest of the WAIS-R, recognizing overlapping pictures (Luria), and in identifying silhouettes depicting animals and objects (VOSP). Unconventional tests using pictures evidenced impairment in her capacity to identify famous faces, facial emotions and animals. Her face perception abilities were preserved, but recognition could not take place. Therefore, it appears that the agnosia in this case best fits the group of categories termed associative.

Prosopagnosia é uma agnosia visual caracterizada por uma incapacidade de reconhecer faces humanas vistas anteriormente e aprender outras. O objetivo é apresentar uma mulher de 46 anos com prosopagnosia congênita e discutir as bases neurais da percepção e do reconhecimento de faces. Ela nos procurou referindo apresentar desde a infância problemas no reconhecimento de faces de membros da família, amigos próximos e mesmo para sua própria imagem numa fotografia. Também diz apresentar prejuízo no reconhecimento de animais, como discriminar cães de gatos. Apresentou dificuldades em identificar e reconhecer animais desenhados; reconhecer figuras sobrepostas (Luria), incorrendo em paragnosias visuais e identificar silhuetas de animais (VOSP). Em testes não convencionais, usando figuras, evidenciou diminuição da capacidade em identificar faces famosas, expressões faciais e animais, mas não em estimar o sexo e a idade das pessoas. Concluindo, suas habilidades perceptuais para face estão preservadas, mas há um déficit de reconhecimento. Tudo indica que sua agnosia pertence ao grupo das associativas.

Event-related potentials in the implicit recognition of faces / 中华物理医学与康复杂志

Objective To use event-related potentials to identify whether face recognition is implicit, and to investigate specificity, race and inversion effects in face recognition. Methods Upright and inverted photographs of Eastern and Western human faces, upright dog faces and a mobile phone were presented in random order for 150 ms to 20 healthy subjects. The subject's attention was diverted away to search for the letter ”f ” in 6 small letters below the photos. Behavioral data were recorded synchronous with event-related potentials (ERPs). Results The P100,N170 and P170 potentials were enhanced and accelerated by faces compared with non-face images. The amplitude of P100 had the strongest correlation with the type of stimulus, showing the biggest amplitude with the mobile phone image (9.5 μV ), followed by the dog face (7.5 μV ). The weakest amplitude was evoked by the human face (6.9 μV). The amplitude of N170 responded most strongly to the human face, followed by the mobile phone (0.5 μV) and the dog face ( -1.9 μV). The P170 amplitude was largest with the human face and the latency was shortest (7.8 μV and 162.2 ms), followed by the dog face (5.2 μV, 174.4 ms) and the mobile phone (5.1 μV,173.5 ms). ERP components were not affected by the orientations or races of the faces in this situation. No difference was found in the late positive component evoked by the human faces and other images. Conclusions Implicit face recognition may depend on the specifics of the stimulus, although no race effect or inversion effect was detected using this procedure.

Faces prototípicas provenientes de amostras populacionais de uma região brasileira / Prototypical faces from samples of the population of a Brazilian region

O objetivo dessa pesquisa foi gerar, por meio de técnica de computação gráfica denominada morphing, faces prototípicas masculinas e femininas a partir de amostras de pessoas que se autodeclaram brancas, pardas e pretas na região de Ribeirão Preto (SP). A cada face prototípica foram utilizadas 64 imagens de faces de voluntários. As faces prototípicas e as faces que as compuseram tiveram a tipicidade de sua cor ou raça julgada por terceiros. Os julgamentos indicaram uma concordância entre a cor/raça-autodeclarada e a heteroatribuída o que indica que os protótipos são representativos de suas respectivas categorias de cor/raça. Discute-se como as faces prototípicas podem ser empregadas em pesquisas básicas e aplicadas em percepção e reconhecimento de faces.

The goal of this study was to generate male and female prototypical faces, by morphing technique, from faces of people who were self-declared as white, "mixed race" and black in Ribeirão Preto-SP. Each prototypical face was generated from 64 face images of volunteers. Another sample of volunteers judged the typicality of color/race of the prototypical faces and the faces that composed them. The results suggested a concordance between the self-declared color/race and the judgments, indicating that the prototypical faces are representative of their respective categories of color/race. Applications of the prototypical faces in studies of face perception and face recognition were discussed.

Neurobiology of Facial Perception and Facial Affect Recognition / 신경정신의학

Recognition of face is an essential cognitive processing to communicate information directly each other. The electrophysiological data have demonstrated that the negative event-related potential (ERP) recorded at occipitotemporal leads, the N170 potential, is a very sensitive neural correlate for faces. The fusiform gyrus, located on the ventromedial surface of temporal and occipital lobes, has been demonstrated to be a key brain region for face perception. Because the face contains the emotional informations, study about face recognition includes emotional processing as well as structural processing of face. So as a essential part of social cognition, the face recognition and facial affects preception have been paid attentions by many researchers. In this article, we reviewed some important research data about face recognition and face affect perception.

EPP: An ERP Late Positive Component Related to Facial Expression Processing / 航天医学与医学工程

Objective To examine ERP components related to different processings of emotional expression of fury, neutral and smile during face-map recognition. Method Twelve undergraduate students (7 male, 5 female), 19～24 years old, were served as subjects in the experiment. The visual stimuli consisted of six kinds of pictures: three kinds of facial pictures (neutral, smile, fury) and three kinds of non-facial ones. The subjects were asked to complete two tasks, i. e., face recognition (FR) and facial expression classification (FEC). During FR task, two runs of 210 stimuli (duration: 50 ms) with each (3 facial and 3 non-facial pictures) were randomly presented with equal probability (ISI: from 1 500 ms to 2 000 ms randomly), and subjects were asked to react to facial stimuli and non-facial stimuli by pressing the left button and right button respectively as quickly as possible. During FEC task, 210 stimuli (3 facial stimuli and 3 non-facial stimuli) were randomly presented with equal probability (duration: 50 ms, ISI: from 1 500 ms to 2 000 ms), and subjects were asked to react to different facial expressions by pressing 3 different buttons respectively as quickly as possible. Electroencephalogram (EEG) was recorded with NeuroScan-32 cap (10/20 system), in reference to bilateral mastoids. Result: 1) Compared with the ERP elicited by facial stimuli during FR task, the responses during FEC task evoked a P580 component clearly. 2) A positive-potential difference after 450 ms post-stimulation between ERP during FEC task and during FR task was observed, which was significant at central and parietal sites. According to the difference of task demand between two tasks, this wave was termed the expression processing positivity (EPP) to reflect the expression processing during face recognition. 3) The peak latency of furyEPP was the shortest and smile-EPP the longest. Conclusion The mode and site of the EPP are likeIv to represent brain function in late processinq of the expression.

Research of automatic human face recognition correlative with face pose / 医疗卫生装备

Engenfaces is a simple and effective face recognition method based on the relativity of the training set and testing set of face images.However,the correct recognition rate decreases by eigenfaces sharply when the face pose varies.Aiming at improving correct recognition rate in this case,a face recognition approach correlative with face poses is presented.The basic idea is toensure that testing face with particular pose should be matched in the face database whose poses are the same as that of the testing face.Experimental results show the new method gains higher correct recognition rates than traditional eigenfaces method.