Learning by doing with images and words

Objective Using olfactory task functional magnetic resonance imaging (fMRI) to investigate the difference in brain olfactory activation between patients with subjective cognitive decline and normal elderly subjects, and to explore the objective image markers for early identification and evaluation the progression from SCD to Alzheimer′s disease (AD). Methods Twenty patients with SCD (SCD group) and twenty matched normal elderly subjects (NC group) were recruited from the community from March 2017 to December 2018. A full neuropsychological scale tests battery, olfactory behavioral tests and olfactory task-fMRI were performed. The differences between olfactory behavior, neuropsychological scales, and task-fMRI brain activation between the two groups were tested. Further, brain regions, which had significantly different activations under task-fMRI, were used as seeds for resting state functional connectivity (FC) analysis. Finally, the correlations between brain activation and olfactory behavior along with clinical neuropsychological scale tests were examined. Results The results of this study showed SCD had a significant decrease in olfactory behavior (olfactory recognition ability) compared with NC (t=-3.042, P<0.01), and there was no statistically significant difference in olfactory threshold. Significant declines were also observed in the SCD self-rating scale (t=6.973, P<0.01), the immediate (t=-4.623, P<0.01) and delayed (t=-2.746, P<0.01) testing of Philadelphia word learning, while the remaining neuropsychological scales were normal. In the olfactory task-fMRI, activation of bilateral primary olfactory cortical regions was significantly reduced in SCD patients, including bilateral entorhinal cortex, amygdala, piriform cortex, anterior olfactory nucleus, and head of the hippocampus. The resting state functional connectivity with the primary olfactory cortex (POC) as the seed showed that the functional connectivity between the olfactory system and the default model network (DMN) of SCD patients was significantly weakened (AlphaSim correction with voxel level P<0.01 and cluster level P<0.05). The Beta value of the left POC was significantly positively correlated with the olfactory threshold and Montreal cognitive assessment (MoCA) (r=0.329, P=0.041; r=0.317, P=0.046). Partial correlation analysis indicated that there was a significantly positive correlation between the FC of the left POC with the right/left inferior frontal gyrus, the left frontal middle gyrus and the right inferior parietal, and the score of immediate Philadelphia word learning test(r=0.411, P=0.008; r=0.400, P=0.011; r=0.329, P=0.003; r=0.454, P=0.003). The FC between the left POC and the right inferior temporal gyrus was negatively correlated with the score of trail making test (TMT) B, and the FC between the left POC and the right inferior was negatively correlated with score of language fluency test (r=-0.317, P=0.047; r=-0.333, P=0.036). The FC between the right POC and the left inferior parietal was positively correlated with the score of immediate Philadelphia word learning (r=0.315, P=0.048), while the FC between right POC and left middle occipital gyrus was negatively corrected with Language Fluency Test (r=-0.403, P=0.01). Conclusion Olfactory function has been impaired in SCD patients with normal standard cognition and phychiatric rating scales, and the changes in the activation of the primary olfactory cortex, such as the entorhinal cortex, may be an early neural circuit damage biomarker for objective evaluation of SCD. Key words: Subjective cognitive decline; Olfactory perception; Magnetic resonance imaging

The purpose of this research is to investigate the correlation between technology usage, technology learning style and visual thinking for the Digital Natives, who are limited in this research to those who are borne after 1980. The researcher adopts purposive sampling to collect 237 valid samples in Tamkang University. The tools include “Technology Usage Questionnaire,” “Technology learning style scale,” and “Visual Thinking Test.” The gathered data are analyzed by several statistical procedures, such as descriptive statistics, t-test, one-way ANOVA, Pearson product-moment correlation and stepwise multiple regression analysis. The major findings are as followings: 1. The Digital Natives usually use digital technology products with the purpose to interact and entertain with the participants, and they regard themselves to have the digital learning style of “pictorial cognition” and “perceptive dependence.” 2. The Digital Natives can represent the images but only in a shallow way. Factually, they hold poorer ability to illustrate the ideas than their self-estimation. 3. The visual thinking ability of the Digital Natives is related to how they make use of digital technology. The more they tend to use the writing system, the better their visual thinking is. 4. Those Digital Natives who tend to use digital technology with images are marked more on technology learning style. On the contrary, those who tend to approach words, the less. 5. The technology learning style of the Digital Natives are related to visual thinking. The more actively they think via image, the better visual thinking ability they own. 6. The traits of digital learning style such as language expression ability, parallel thinking and pictorial cognition are predictors on the visual thinking. Based on the abovementioned results and discussions, the researcher could provide some recommendations and consultations in education and research in the future.

Objective To construct a real-time monitoring system based on computer vision for monitoring withdrawal speed of colonoscopy and to validate its feasibility and performance. Methods A total of 35 938 images and 63 videos of colonoscopy were collected in endoscopic database of Renmin Hospital of Wuhan University from May to October 2018. The images were divided into two datasets, one dataset included in vitro, in vivo and unqualified colonoscopy images, and another dataset included ileocecal and non-cecal area images. And then 3 594 and 2 000 images were selected respectively from the two datasets for testing the deep learning model, and the remaining images were used to train the model. Three colonoscopy videos were selected to evaluate the feasibility of real-time monitoring system, and 60 colonoscopy videos were used to evaluate its performance. Results The accuracy rate of the deep learning model for classification for in vitro, in vivo, and unqualified colonoscopy images was 90.79% (897/988), 99.92% (1 300/1 301), and 99.08% (1 293/1 305), respectively, and the overall accuracy rate was 97.11% (3 490/3 594). The accuracy rate of identifying ileocecal and non-cecal area was 96.70% (967/1 000) and 94.90% (949/1 000), respectively, and the overall accuracy rate was 95.80% (1 916/2 000). In terms of feasibility evaluation, 3 colonoscopy videos data showed a linear relationship between the retraction speed and the image processing interval, which indicated that the real-time monitoring system automatically monitored the retraction speed during the colonoscopy withdrawal process. In terms of performance evaluation, the real-time monitoring system correctly predicted entry time and withdrawal time of all 60 examinations, and the results showed that the withdrawal speed and withdrawal time was significantly negative-related (R=-0.661, P<0.001). The 95% confidence interval of withdrawal speed for the colonoscopy with withdrawal time of less than 5 min, 5-6 min, and more than 6 min was 43.90-49.74, 40.19-45.43, and 34.89-39.11 respectively. Therefore, 39.11 was set as the safe withdrawal speed and 45.43 as the alarm withdrawal speed. Conclusion The real-time monitoring system we constructed can be used to monitor real-time withdrawal speed of colonoscopy and improve the quality of endoscopy. Key words: Quality control; Artificial intelligence; Colonoscopy; Withdrawal time; Withdrawal speed

Objective To evaluate the application effect of teaching-picture archiving and communication system （T-PACS） in training image reading ability of eight-year-program students. Methods Sixty students were divided into two groups： control group （applying traditional teaching method） and experimental group （applying traditional teaching method and T-PACS） when doing clinical medical image reading ability training. Making multimedia teaching movie, clinical image reading teaching and establishing image database by T-PACS were used in experiment group. Mean-while, reading room was set up and classroom teaching and problem-oriented assignments were in-creased. Finally objective evaluation of image recognition ability and questionnaire were used to compare the teaching effect. Measurement data were expressed as x±s. Total scores of image reading were am-alyzed by t test. Full score rates were analyzed by X2 test. Results The results showed that the score of medical image reading ability was higher in experimental group （90.08±4.52） than in control group （86.92±4.68） （P=0.010）. There were differences in respiratory and orthopedic full score rate （P=0.006;P=0.035）. Most students thought that application of T-PACS could effectively improve the teaching and learning quality. Conclusions Application of T-PACS can improve the medical image reading ability of eight-year-program medical students and is worth further promoting. Key words: Teaching-picture archiving and communication system; Clinical medicine; Eight-year-program; Medical imaging teaching

This study aims to determine the effect of learning media and learning styles on learning outcomes in the cognitive domain of students in basic physiology subjects. This study used a quasi-experimental method using a 2X2 factorial design. The findings indicate that: (1) the learning outcomes of students who use model learning media are higher than those of students who use image learning media (2) the learning outcomes of students who have a kinesthetic learning style are higher than the learning outcomes of students who have visual learning styles. (3) there is an influence of the interaction between learning media and learning styles on student cognitive learning outcomes in Basic Physiology courses. (4) cognitive learning outcomes of students who have a kinesthetic learning style who learn using model learning media are higher than realm learning outcomes. cognitive students who have a kinesthetic learning style who learn using image learning media. (5) cognitive learning outcomes of students who have visual learning styles who learn using learning media models are lower than those of students who have visual learning styles who learn using image learning media. (6) the cognitive learning outcomes of students who have a kinesthetic learning style who learn using model learning media are higher than the cognitive learning outcomes of students who have a visual learning style who learn using model learning media. (7) the cognitive learning outcomes of students who have a kinesthetic learning style who learn using image learning media are lower than the cognitive learning outcomes of students who have a visual learning style who learn using image learning media.

Objective To analyze image feature of multislice spiral CT of intrapulmonary lymph nodes. Methods The image features and clinical data of 15 cases with intrapulmonary lymph nodes were analyzed retrospectively between January 2014 and June 2017. All cases diagnosed by surgery and pathology. Among 15 cases, there were 6 males and 9 females, aged 38-69 years, with an average of 55.6 years. Low dose CT scan and high resolution CT scan were performed in all cases. Three cases received enhanced CT scan at the same time. The location, number, size, image features and accompanying lesions of pulmonary lymph nodes were observed. Results In total, 19 intrapulmonary lymph nodes were discovered in 15 cases by multi-slice spiral CT, with a maximum diameter of 4-10 (5.53±1.50)mm. There were 14 intrapulmonary lymph nodes situated at the right lung, 1 in the right upper lobe, 5 in the middle and 8 in the lower lobe. Other 5 intrapulmonary lymph nodes were all located at the inferior lobe of left lung. There were 15 lymph nodes occurred at the outside of sub-pleural area or near the interlobular pleura. The vertical distance between lesions and adjacent pleura were from 3-16 mm, with an average of 9 mm. Other 4 lymph nodes were closely attached to the sub-pleura. The boundary of 17 lymph nodes was clear and sharp, and other 2 were rough. The density of 17 lymph nodes was uniform and other 2 were uneven, without calcification or cystoid variation. In axial images, 11 pulmonary lymph nodes were round like, 2 triangular and 6 irregular. In 12 lymph nodes, one or more uniform and compact linear shadows were linked to the lesions on the axis image or reconstructed image of coronal and sagittal plane. Three intrapulmonary lymph nodes were found in 3 cases by enhanced CT scan. Two lymph nodes showed no obvious enhancement and 1 showed progressive delayed enhancement. There were 4 cases accompanied with primary lung cancer at the same time. Conclusions The image features of intrapulmonary lymph nodes represent some common characteristics. Learning about these futures will contribute to clear diagnosis before surgery. Key words: Lung diseases; Intrapulmonary lymph nodes; Tomography, X-ray computer

Objective To investigate the effects of normobaric hyperoxia intervention on learning and memory abilities of valproic acid(VPA) autism model rats and the morphology of pyramidal cells in hippocampus CA1 area. Methods Animal model groups of autism were obtained in male offspring of the Wistar rats that received intraperitoneal injection of 600 mg/kg VPA at the 12.5 day after pregnancy. According to the eye opening time, behavior, weaning weight and the learning and memory abilities which were evaluated by the Y electricity maze test at the 28th day after birth, 40 male VPA autism model rats were randomly selected 20 only and divided into normobaric hyperoxia model group(group A, n=10), atmospheric air model group(group B, n=10). Normal control groups were obtained in male offspring of Wistar rats that received intraperitoneal injection of equivalent physiological saline at the same period pregnancy.(group C, n=10). Rats in group A were treated with oxygen for 1h per day and lasted 1 week; group B and C were treated with normal air. The learning and memory abilities of three groups were assessed at the 35th day after birth. The immunohistochemistry methods and image analysis were used to observe the pyramidal cells of autism model rats in hippocampal CA1 region. The effect of normobaric hyperoxia therapy on pyramidal cell of autism model rats in hippocampal CA1 region were evaluated by HE staining technique. Results The trying times of group A after treatment were less than those before treatment (31.15±0.99 vs 31.54±0.97, t=2.739, P=0.018). The memory times were more than those before treatment(3.00±0.58 vs 2.69±0.48, t=-2.309, P=0.040). The trying times of group A after treatment were less than those in group B after treatment(P=0.016). The memory times of group A were not different from that in group B after treatment(P=0.810). The morphology of pyramidal cells in hippocampal CA1 region showed that the pyramidal cells of the autism model rats had apoptosised. The number of apoptotic cells reduced and the number of normal form cells increased after the normobaric hyperoxia intervention compared with the autism model rats. Conclusion Normobaric hyperoxia intervention can improve the learning and memory abilities of the autism model rats.The apoptosis of the pyramidal neurons in hippocampal CA1 might be reduced after the normobaric hyperoxia intervention. Key words: Valproate; Normobaric hyperoxia; Autism; Learning and memory; Hippocampus

Objective To improve the effect of standardized residency training in radiology department by establishing WeChat learning platform, image teaching system and RIS-based report scoring and reporting feedback system, and adopting the process-based evaluation model, making multiple assessments at different time points and giving feedback. To explore whether the daily reports writing and doctor's scores can be used to evaluate the learning effect by analyzing the correlation between report score, the relearning time and the examination results. Methods The experimental control method was adopted. Twenty-seven trainees who participated in standardized training in Peking University Third Hospital were selected, including 12 trainees in 2017 as the experimental group and 15 trainees in 2016 as the control group. Two groups of trainees were similar in composition. In training, the experimental group was adopted the network learning platform teaching model, while the control group was adopted the traditional teaching model. After the training, results of X ray and CT were compared between the experimental group and control group. Results of different trainees in the experimental group were compared. Correlations between scores, relearn time and exams, one-stage mock examination were analyzed. Results The average scores of X ray: (85.7±6.9) in the experimental group and (80.0±7.7) in the control group, there was statistically significant difference in X ray scores (t=-1.988, P>0.05). The average scores of CT were (86.8±5.7) in the experimental group and were (81.1±7.5) in the control group, there was statistically significant difference in CT scores (t=-2.198, P<0.05). The experimental group was divided into two groups according to the source of trainees. There were differences of X ray examination, CT examination and one-stage mock examination between two groups (all P<0.05). Correlation analysis was made between the proportion of median score, superior doctor's score, median report relearning time and X ray and CT examination. The scores of X ray and CT examination were negatively correlated to the proportion of median scores, and relearning time, and were positively correlated to the scores of superior doctors. The relearning time after examination was negatively correlated to the results of one-stage mock examination. Conclusions The process evaluation model based on network learning platform improves the effect of standardized resident training and the ability to master professional knowledge. The correlations between scores, relearning time and examination results indicate that the daily report writing and the doctor's scores can be used to evaluate the learning effect. At the same time, the role of rating should be strengthened and the quality of report writing should be improved in the further. Key words: Networking; WeChat; Residents' standardized training; Process evaluation

Objective To evaluate the application value of an intelligent fundus assisted diagnosis system for detecting retinopathy of prematurity (ROP) based on deep learning. Methods A total of 38 895 fundus images for premature infants screening were collected from Renmin Hospital of Wuhan University Eye Center and were labeled by 10 licensed ophthalmologists.A deep learning network model was established to acquire automatic classification of disease stages and plus disease.The accuracy, sensitivity and specificity of the algorithm were calculated to evaluate the performance of the artificial intelligence system for ROP automatic diagnosis.This study protocol was approved by Ethic Committee of Renmin Hospital of Wuhan University (No.WDRY2019-K032). Written informed consent was obtained from the guardians of the children before entering the study cohort. Results The intelligent system achieved an accuracy of 0.931.Specifically, the accuracies in detecting demarcation line (stage Ⅰ) was 0.876, ridge (stage Ⅱ) was 0.942, ridge with extra retinal fibrovascular (stage Ⅲ) was 0.968, subtotal retinal detachment (stage Ⅳ) was 0.998, total retinal detachment (stage Ⅴ) was 0.999, vascular tortuosity and dilatation (plus disease) was 0.896, optic disc was 0.954, macular was 0.781, and laser scars were 0.974, respectively. Conclusions Deep learning algorithm can detect the stages and plus disease of ROP with excellent accuracy, and it provides the feasibility of applying the algorithm for ROP automated screening in clinical. Key words: Artificial intelligence; Deep learning; Retinopathy of prematurity; Assisted diagnosis

In order to realize the integration of theory and practice teaching in parasitology examination,the digital microscope system with the functions of sound,image,words,pictures and flexible interaction system were used in teaching.It shows that this system not only enhances the students' learning ability,but also improves the theoretical teachers' practical ability,experimental teachers' theoretical level,and improves significantly to the teaching quality and efficiency. Key words: Parasitology examination ; Integration of theory and practice teaching ; Interactive digital microscope system ;

Through the excellent experiment of cardiovascular system in pathology, the practice of the teaching reform was carried out such as combining flow model with real specimens, digital sections with light microscope slides, inserting the use of special staining experiments in combination with digital medical image analysis and the simulation of cardiovascular clinical pathological case discussions. Linking theory with experiment teaching and experiment teaching with clinical practice was focused on, which not only stimulated students＆#39; interest in learning and their exploring thinking and hands-on ability, but also promoted them to obtain good learning effect, thus improving the teaching quality. Key words: Pathology; Cardiovascular disease; Eexperimental course; Eexperimental methodology

Objective To investigate the radiation dose and image noise of abdominal CT with spectral imaging and conventional scan mode. Methods This was a retrospective study.The first part of this study was a phantom study.A standard quality assessment phantom was scanned with gemstone spectral imaging mode and conventional helical mode using 120 k Vp with different m As.A regression function for the m As between conventional scan mode and spectral imaging mode was obtained.According to the regression function,the m As on 120 k Vp corresponding to different gemstone spectral imaging protocol were calculated.The second part of this study was a clinical study.Twenty-two patients who underwent contrast enhanced abdominal CT scanning were included.The pre-contrast CT was scanned by a conventional mode with automatic exposure control technique.According to the m As of pre-contrast scan,corresponding spectral imaging parameters were selected for the portal venous phase of post-contrast scan.Five sets of images with 5 mm slice thickness were reconstructed,with 65 keV monochromatic image for spectral imaging and FBP,20%, 40% and 60% ASIR image for conventional imaging respectively.Image noise was measured on hepatic parenchyma and urinary bladder for each set of image.Radiation dose and image noise were compared between CT spectral imaging and conventional imaging with paired t-test. Results No significant differences were found for the CTDIvol[ (13.8±4.8) mGy versus (14.0±7.0) mGy] and DLP[ (691.1±274.2) mGy·cm versus (678.8±385.0) mGy·cm] between CT spectral imaging and conventional imaging.Monochromatic image (65 keV) by spectral imaging[ (7.6±1.4) HU for hepatic parenchyma and (9.1±1.3) HU for urinary bladder] had equal image noise with 40% ASIR[hepatic parenchyma: (7.8±0.7) HU]and 20% ASIR[urinary bladder: (9.2±1.2) HU]and lower image noise than FBP image[ (10.4±1.0) HU for hepatic parenchyma and (10.5±1.3) HU for urinary bladder]by conventional imaging. Conclusions The radiation dose of CT spectral imaging and conventional CT scan is equivalent with spectral imaging parameters appropriately selected.Monochromatic image (65 keV) by CT spectral imaging has lower image noise than FBP image,and same image noise with 40% ASIR (hepatic parenchyma) and 20% ASIR (urinary bladder) by conventional imaging. Key words: Tomography,X-ray computed; Radiation Dosage; Comparative study

By utilising web-based collaboration tools, institutions can engage users in the building of historical printed text resources created by mass digitisation projects. The paper presents the drivers for developing such tools and identifies the benefits that can be derived by both the user community and cultural heritage institutions. The perceived risks, such as errors introduced by the users or whether users will engage with resources in this way, will be set out. The paper will present the lessons that can be learnt from existing activities, such as the National Library of Australia’s newspaper website, which supports collaborative correction of Optical Character Recognition (OCR) output. The user collaboration tools being created by the IMPACT Project (Improving Access to Text, http://www.impact-project.eu), a large-scale integrating project funded by the European Commission as part of the Seventh Framework Programme (FP7), will be detailed. A primary aim of IMPACT is to develop tools that help improve OCR results for historical printed texts, specifically those works published before the industrial production of books in the middle of the 19th century. While technological improvements to image processing and OCR engine technology are key to improving access to historic text, engaging the user community also has an important role to play. Utilising the user community can aid in achieving the levels of accuracy currently found in born digital materials. Improving OCR results to this level is key to producing resources that support better resource discovery and enable greater performance when applying text mining and accessibility tools to the extracted text. The IMPACT project will specifically develop a tool that supports collaborative correction and validation of OCR results and to allow user involvement in building historical dictionaries that can be used to validate word recognition. The technologies use the characteristics of human perception as a basis for error detection.

An 18-year-old male patient developed numbness and fatigue in the limbs without any obvious causes, which aggravated to be unable to hold objects in hands, have difficulty in writing, standing and walking, and be easy to fall after 3 weeks. Laboratory tests showed his average erythrocyte volume 104 fl, homocysteine 89.1 μmol/L, vitamin B12 76 pmol/ml, and folic acid 14.9 nmol/L. MRI revealed image features of subacute combined degeneration of spinal cord. Nerve conduction velocity, action potential, and electromyography showed nerve damage signs. He was diagnosed with toxic myelopathy, subacute combined degeneration of spinal cord, and peripheral neuropathy. To find the cause of the disease, the doctor asked about the medical history repeatedly, and finally learned that the patient had taken nitrous oxide(N2O) for nearly 6 months(8-50 g/d). An IV injection of mecobalamine (0.5 mg, once daily), intramuscular injection of rat nerve growth factor (30 μg, once daily), as well as massage and acupuncture treatments were given. After 8 days, his average erythrocyte volume was 102 fl, homocysteine was 12.9 μmol/L, vitamin B12 was >1 500 pmol/ml, and folic acid was 14.1 nmol/L. After 3 weeks, numbness and fatigue of his upper limbs disappeared. After 3 months, the patient could walk up and down the hill and run. Key words: Nitrous oxide; Dosing, inhalation; Nervous system; Neurodegenerative diseases

Word Sense Disambiguation (WSD) is the process of resolving the meaning of a word unambiguously in a given natural language context. Within the scope of this thesis, it is the process of marking text with explicit sense labels. What constitutes a sense is a subject of great debate. An appealing perspective, aims to define senses in terms of their multilingual correspondences, an idea explored by several researchers, Dyvik (1998), Ide (1999), Resnik & Yarowsky (1999), and Chugur, Gonzalo & Verdejo (2002) but to date it has not been given any practical demonstration. This thesis is an empirical validation of these ideas of characterizing word meaning using cross-linguistic correspondences. The idea is that word meaning or word sense is quantifiable as much as it is uniquely translated in some language or set of languages. Consequently, we address the problem of WSD from a multilingual perspective; we expand the notion of context to encompass multilingual evidence. We devise a new approach to resolve word sense ambiguity in natural language, using a source of information that was never exploited on a large scale for WSD before. The core of the work presented builds on exploiting word correspondences across languages for sense distinction. In essence, it is a practical and functional implementation of a basic idea common to research interest in defining word meanings in cross-linguistic terms. We devise an algorithm, SALAAM for Sense Assignment Leveraging Alignment And Multilinguality, that empirically investigates the feasibility and the validity of utilizing translations for WSD. SALAAM is an unsupervised approach for word sense tagging of large amounts of text given a parallel corpus—texts in translation—and a sense inventory for one of the languages in the corpus. Using SALAAM, we obtain large amounts of sense annotated data in both languages of the parallel corpus, simultaneously. The quality of the tagging is rigorously evaluated for both languages of the corpora. The automatic unsupervised tagged data produced by SALAAM is further utilized to bootstrap a supervised learning WSD system, in essence, combining supervised and unsupervised approaches in an intelligent way to alleviate the resources acquisition bottleneck for supervised methods. Essentially, SALAAM is extended as an unsupervised approach for WSD within a learning framework; in many of the cases of the words disambiguated, SALAAM coupled with the machine learning system rivals the performance of a canonical supervised WSD system that relies on human tagged data for training. Realizing the fundamental role of similarity for SALAAM, we investigate different dimensions of semantic similarity as it applies to verbs since they are relatively more complex than nouns, which are the focus of the previous evaluations. We design a human judgment experiment to obtain human ratings on verbs' semantic similarity. The obtained human ratings are cast as a reference point for comparing different automated similarity measures that crucially rely on various sources of information. Finally, a cognitively salient model integrating human judgments in SALAAM is proposed as a means of improving its performance on sense disambiguation for verbs in particular and other word types in general.