OnDev-LCT: On-Device Lightweight Convolutional Transformers towards federated learning.

Federated learning (FL) has emerged as a promising approach to collaboratively train machine learning models across multiple edge devices while preserving privacy. The success of FL hinges on the efficiency of participating models and their ability to handle the unique challenges of distributed learning. While several variants of Vision Transformer (ViT) have shown great potential as alternatives to modern convolutional neural networks (CNNs) for centralized training, the unprecedented size and higher computational demands hinder their deployment on resource-constrained edge devices, challenging their widespread application in FL. Since client devices in FL typically have limited computing resources and communication bandwidth, models intended for such devices must strike a balance between model size, computational efficiency, and the ability to adapt to the diverse and non-IID data distributions encountered in FL. To address these challenges, we propose OnDev-LCT: Lightweight Convolutional Transformers for On-Device vision tasks with limited training data and resources. Our models incorporate image-specific inductive biases through the LCT tokenizer by leveraging efficient depthwise separable convolutions in residual linear bottleneck blocks to extract local features, while the multi-head self-attention (MHSA) mechanism in the LCT encoder implicitly facilitates capturing global representations of images. Extensive experiments on benchmark image datasets indicate that our models outperform existing lightweight vision models while having fewer parameters and lower computational demands, making them suitable for FL scenarios with data heterogeneity and communication bottlenecks.

A Hybrid Multimodal Emotion Recognition Framework for UX Evaluation Using Generalized Mixture Functions.

Multimodal emotion recognition has gained much traction in the field of affective computing, human-computer interaction (HCI), artificial intelligence (AI), and user experience (UX). There is growing demand to automate analysis of user emotion towards HCI, AI, and UX evaluation applications for providing affective services. Emotions are increasingly being used, obtained through the videos, audio, text or physiological signals. This has led to process emotions from multiple modalities, usually combined through ensemble-based systems with static weights. Due to numerous limitations like missing modality data, inter-class variations, and intra-class similarities, an effective weighting scheme is thus required to improve the aforementioned discrimination between modalities. This article takes into account the importance of difference between multiple modalities and assigns dynamic weights to them by adapting a more efficient combination process with the application of generalized mixture (GM) functions. Therefore, we present a hybrid multimodal emotion recognition (H-MMER) framework using multi-view learning approach for unimodal emotion recognition and introducing multimodal feature fusion level, and decision level fusion using GM functions. In an experimental study, we evaluated the ability of our proposed framework to model a set of four different emotional states (Happiness, Neutral, Sadness, and Anger) and found that most of them can be modeled well with significantly high accuracy using GM functions. The experiment shows that the proposed framework can model emotional states with an average accuracy of 98.19% and indicates significant gain in terms of performance in contrast to traditional approaches. The overall evaluation results indicate that we can identify emotional states with high accuracy and increase the robustness of an emotion classification system required for UX measurement.

Robust Perturbation for Visual Explanation: Cross-Checking Mask Optimization to Avoid Class Distortion.

Along with the outstanding performance of the deep neural networks (DNNs), considerable research efforts have been devoted to finding ways to understand the decision of DNNs structures. In the computer vision domain, visualizing the attribution map is one of the most intuitive and understandable ways to achieve human-level interpretation. Among them, perturbation-based visualization can explain the "black box" property of the given network by optimizing perturbation masks that alter the network prediction of the target class the most. However, existing perturbation methods could make unexpected changes to network predictions after applying a perturbation mask to the input image, resulting in a loss of robustness and fidelity of the perturbation mechanisms. In this paper, we define class distortion as the unexpected changes of the network prediction during the perturbation process. To handle that, we propose a novel visual interpretation framework, Robust Perturbation, which shows robustness against the unexpected class distortion during the mask optimization. With a new cross-checking mask optimization strategy, our proposed framework perturbs the target prediction of the network while upholding the non-target predictions, providing more reliable and accurate visual explanations. We evaluate our framework on three different public datasets through extensive experiments. Furthermore, we propose a new metric for class distortion evaluation. In both quantitative and qualitative experiments, tackling the class distortion problem turns out to enhance the quality and fidelity of the visual explanation in comparison with the existing perturbation-based methods.

Integrating Text-Mining and Balanced Scorecard Techniques to Investigate the Association between CEO Message of Homepage Words and Financial Status: Emphasis on Hospitals.

(1) Background: The Chief Executive Officer's (CEO's) message on a hospital's homepage on the Internet contains various components, such as the hospital's future vision, promises to customers, availability of upgraded services and public activities. This statement usually includes non-financial information as well as financial information about the corporate entity owning/operating the hospital. In addition, it provides useful information about not only the company's goals and vision, but also firm performance targets and strategies for the future. This study aims to investigate associations between the CEO's message and the financial status of the institution. We used the balanced scorecard framework to analyze what content on the hospital's homepage is related to the hospital's various financial ratios. (2) Methods: We adopted a text-mining method to extract significantly repeated keywords from the CEO's message on the hospital's website. Then, we classified these keywords using a balanced scorecard approach. To examine the relationship between keywords in the CEO's message and the hospital's financial ratios, a t-test was conducted for the difference in the term frequency divided by inverse document frequency (TF-IDF) mean of the home page contents and its relationship with the views of the balanced scorecard framework. (3) Results: According to our empirical results on 65 samples collected from local hospitals, there are some significant relationships between the qualitative content of the hospital's homepage and the quantitative financial ratios that indicate profitability, activity, leverage, liquidity, and accumulating reserves for proper business purposes. (4) Conclusions: The introduction section of a homepage is the part most accessible to customers, containing the aims and ideals of the hospital and reflecting the institution's values and visions. In addition, in the coverage of financial status, the organization can either emphasize financial strength or focus on other areas to divert attention from any weakness shown in the financial information. This study reminds us of the importance of the hospital website's disclosure, and what can be inferred from the financial status of the hospital. It also highlights the need for reconciliation and harmony between the quantitative data, financial statements, and qualitative data in the CEO's message. (5) Implications: To the best of our knowledge, this paper is the first research attempting to investigate the relationship between text on the hospital's homepage and the hospital's financial ratios using text-mining techniques and the balanced scorecard framework. Hospitals play a crucial role in a country's welfare and healthcare industry. Nevertheless, in many countries, hospital organizations tend to remain a source of critical fiscal deficits due to ineffective and sloppy management. We expect that the result of this paper can provide hospital managers with useful information to address that situation.

Guidelines for the Management of Unilateral Vocal Fold Paralysis From the Korean Society of Laryngology, Phoniatrics and Logopedics.

The Korean Society of Laryngology, Phoniatrics and Logopedics appointed a task force to establish clinical practice guidelines for the management of unilateral vocal fold paralysis (UVFP). These guidelines cover a comprehensive range of management-related factors, including the diagnosis and treatment of UVFP, and provide in-depth information based on current, up-to-date knowledge. Detailed evidence profiles are provided for each recommendation. The CORE databases, including OVID Medline, Embase, the Cochrane Library, and KoreaMed, were searched to identify all relevant papers, using a predefined search strategy. When insufficient evidence existed, expert opinions and Delphi questionnaires were used to fill the evidence gap. The committee developed 16 evidence-based recommendations in six categories: initial evaluation (R1-4), spontaneous recovery (R5), medical treatment (R6), surgical treatment (R7-14), voice therapy (R15), and aspiration prevention (R16). The goal of these guidelines is to assist general otolaryngologists and speech-language pathologists who are primarily responsible for treating patients with UVFP. These guidelines are also intended to facilitate understanding of the condition among other health-care providers, including primary care physicians, nurses, and policy-makers.

Clinical Application of Two-Dimensional Scanning Digital Kymography in Discrimination of Diplophonia.

Purpose The purpose of this study was to investigate the characteristics of diplophonia using an auditory perception and multimodal simultaneous examination, which included sound waveform analysis, electroglottography (EGG), digital kymography (DKG), and 2-dimensional scanning digital kymography (2D DKG). Additionally, we compared the diagnostic accuracy of each method using a binary classifier in confusion matrix and convenience of discrimination, based on the time required for interpretation. Method One normophonic male, 12 patients with diplophonia, and 12 dysphonia patients without diplophonia were enrolled. A multimodal simultaneous evaluation was used to analyze the vibration pattern of diplophonia. Sensitivity, specificity, accuracy, area under the curve, and interpretation time were used to compare the various diagnostic methods. Discrimination was determined by 3 raters. Results There are 3 types of asymmetric vibratory patterns in diplophonia. The types are based on the oscillators vibrating at different frequencies: asymmetry of the left and right cords (6 subjects with unilateral palsy and 1 subject with vocal polyps), asymmetry of anterior and posterior cords (2 subjects with vocal polyps), and asymmetry of true and false cords (3 subjects with muscle tension dysphonia). All evaluation methods were useful as diagnostic tools, with all areas under the curve > .70. The diagnostic accuracy was highest with DKG (95.83%), followed by 2D DKG (83.33%), EGG (81.94%), auditory-perceptual evaluation (80.56%), and sound waveform (77.78%). The interpretation time was the shortest for auditory-perceptual evaluation (6.07 ± 1.34 s), followed by 2D DKG (10.04 ± 3.00 s), EGG (12.49 ± 2.76 s), and DKG (13.53 ± 2.60 s). Conclusions Auditory-perceptual judgment was the easiest and fastest method for experienced raters, but its diagnostic accuracy was lower than that of DKG or 2D DKG. The diagnostic accuracy of DKG was the highest, but 2D DKG allowed rapid interpretation and showed relatively high diagnostic accuracy, except in cases with space-occupying lesions. Supplemental Material https://doi.org/10.23641/asha.9911786.

Vacuolar H+-ATPase Subunit V0C Regulates Aerobic Glycolysis of Esophageal Cancer Cells via PKM2 Signaling.

The vacuolar H+-adenosine triphosphatase (ATPase) subunit V0C (ATP6V0C), a proton-conducting, pore-forming subunit of vacuolar ATPase, maintains pH homeostasis and induces organelle acidification. The intracellular and extracellular pH of cancer cells affects their growth; however, the role of ATP6V0C in highly invasive esophageal cancer cells (ECCs) remains unclear. In this study, we examined the role of ATP6V0C in glucose metabolism in ECCs. The ATP6V0C depletion attenuated ECC proliferation, invasion, and suppressed glucose metabolism, as indicated by reduced glucose uptake and decreased lactate and adenosine triphosphate (ATP) production in cells. Consistent with this, expression of glycolytic enzyme and the extracellular acidification rate (ECAR) were also decreased by ATP6V0C knockdown. Mechanistically, ATP6V0C interacted with pyruvate kinase isoform M2 (PKM2), a key regulator of glycolysis in ECCs. The ATP6V0C depletion reduced PKM2 phosphorylation at tyrosine residue 105 (Tyr105), leading to inhibition of nuclear translocation of PKM2. In addition, ATP6V0C was recruited at hypoxia response element (HRE) sites in the lactate dehydrogenase A (LDHA) gene for glycolysis. Thus, our data suggest that ATP6V0C enhances aerobic glycolysis and motility in ECCs.

Analysis of phototoxin taste closely correlates nucleophilicity to type 1 phototoxicity.

Pigments often inflict tissue-damaging and proaging toxicity on light illumination by generating free radicals and reactive oxygen species (ROS). However, the molecular mechanism by which organisms sense phototoxic pigments is unknown. Here, we discover that Transient Receptor Potential Ankyrin 1-A isoform [TRPA1(A)], previously shown to serve as a receptor for free radicals and ROS induced by photochemical reactions, enables Drosophila melanogaster to aphotically sense phototoxic pigments for feeding deterrence. Thus, TRPA1(A) detects both cause (phototoxins) and effect (free radicals and ROS) of photochemical reactions. A group of pigment molecules not only activates TRPA1(A) in darkness but also generates free radicals on light illumination. Such aphotic detection of phototoxins harboring the type 1 (radical-generating) photochemical potential requires the nucleophile-sensing ability of TRPA1. In addition, agTRPA1(A) from malaria-transmitting mosquitoes Anopheles gambiae heterologously produces larger current responses to phototoxins than Drosophila TRPA1(A), similar to their disparate nucleophile responsiveness. Along with TRPA1(A)-stimulating capabilities, type 1 phototoxins exhibit relatively strong photo-absorbance and low energy gaps between the highest occupied molecular orbital and the lowest unoccupied molecular orbital. However, TRPA1(A) activation is more highly concordant to type 1 phototoxicity than are those photochemical parameters. Collectively, nucleophile sensitivity of TRPA1(A) allows flies to taste potential phototoxins for feeding deterrence, preventing postingestive photo-injury. Conversely, pigments need to bear high nucleophilicity (electron-donating propensity) to act as type 1 phototoxins, which is consistent with the fact that transferring photoexcited electrons from phototoxins to other molecules causes free radicals. Thus, identification of a sensory mechanism in Drosophila reveals a property fundamental to type 1 phototoxins.

Clinical Practicability of a Newly Developed Real-time Digital Kymographic System.

BACKGROUND: A digital kymogram shows real images of vocal fold vibration. However, DKG is difficult to use in clinical practice because the recorded image cannot be seen instantaneously after examination, as considerable encoding time is required to visualize a digital kymogram. In addition, frame-by frame analysis should be implemented to evaluate high-speed videoendoscopy data, but is time- and labor-intensive. PURPOSE: The purpose of the study was to validate the clinical practicability of a real-time multislice digital kymographic system developed by the authors. We analyzed the promptness and accuracy of the examination before and after intracordal injections in patients with unilateral vocal fold paralysis. METHODS: To assess the clinical applicability of this system, six patients with unilateral vocal fold paralysis were selected. Real-time DKG was performed before and immediately after intracordal injection. We observed changes in the digital kymogram after the intracordal injection. RESULTS: Using this system, 10 scanning lines and up to five vertical pixel row could be obtained in real time, and the maximum acquisition time for the DKG image was 10 seconds. A digital kymogram of the patients could be instantaneously acquired, and whether the intracordal injection was appropriate or not. CONCLUSION: This article is the first validation study after the development of the real-time multislice digital kymographic system. Our system may be a promising tool in clinical practice for immediate assessment of the vibratory patterns of the vocal cords. More research is necessary for further clinical validation.

Efficacy of Two-dimensional Scanning Digital Kymography in Evaluation of Atrophic Vocal Folds.

OBJECTIVE: The purpose of this study was to evaluate the clinical feasibility and diagnostic accuracy of two-dimensional scanning digital kymography (2D DKG) in patients with vocal cord atrophy before and after treatment. MATERIALS AND METHODS: We analyzed the characteristics of vocal fold vibration in five patients with unilateral vocal fold paralysis and five patients with presbyphonia. In patients with vocal cord paralysis, the status before and after intracordal injection was compared. Furthermore, in patients with presbyphonia, we compared the status before and after voice therapy (Seong-Tae Kim's laryngeal calibration technique). Quantitative parameters such as amplitude and phase symmetry indices, jitter, shimmer, noise-to-harmonic ratio, and maximum phonation time and qualitative parameters such as Voice Handicap Index, glottal gap, amplitude, and phase difference were used to evaluate the pre- and post-treatment status. RESULTS: In cases of vocal cord paralysis, vibratory changes of the vocal folds before and after intracordal injection could be identified immediately using 2D DKG. In overcorrection cases, all of the measured parameters were poor except for improvement of the glottal gap. In addition, 2D DKG showed appropriately the changes in vocal cord vibration before and after voice therapy in patients with presbyphonia. CONCLUSION: Two-dimensional DKG may be a useful diagnostic tool in evaluation of the vibratory characteristics of entire vocal cords. In addition, it may also play a role in providing a decision for treatment modalities.

Visually interpretable deep network for diagnosis of breast masses on mammograms.

Recently, deep learning technology has achieved various successes in medical image analysis studies including computer-aided diagnosis (CADx). However, current CADx approaches based on deep learning have a limitation in interpreting diagnostic decisions. The limited interpretability is a major challenge for practical use of current deep learning approaches. In this paper, a novel visually interpretable deep network framework is proposed to provide diagnostic decisions with visual interpretation. The proposed method is motivated by the fact that the radiologists characterize breast masses according to the breast imaging reporting and data system (BIRADS). The proposed deep network framework consists of a BIRADS guided diagnosis network and a BIRADS critic network. A 2D map, named BIRADS guide map, is generated in the inference process of the deep network. The visual features extracted from the breast masses could be refined by the BIRADS guide map, which helps the deep network to focus on more informative areas. The BIRADS critic network makes the BIRADS guide map to be relevant to the characterization of masses in terms of BIRADS description. To verify the proposed method, comparative experiments have been conducted on public mammogram database. On the independent test set (170 malignant masses and 170 benign masses), the proposed method was found to have significantly higher performance compared to the deep network approach without using the BIRADS guide map (p < 0.05). Moreover, the visualization was conducted to show the location where the deep network exploited more information. This study demonstrated that the proposed visually interpretable CADx framework could be a promising approach for visually interpreting the diagnostic decision of the deep network.

Spinopelvic Orientation on Radiographs in Various Body Postures: Upright Standing, Chair Sitting, Japanese Style Kneel Sitting, and Korean Style Cross-Legged Sitting.

BACKGROUND: Several previous studies reported on the impact of upright standing and chair sitting on the sagittal spinopelvic alignment. However, there are no studies on the impact of the two Asian (Korean and Japanese) style floor-sitting positions on the sagittal spinopelvic alignment. The purpose of this study was to evaluate the impact of four different body postures (standing, chair sitting, kneel sitting, and cross-legged sitting) on the sagittal spinopelvic alignment. METHODS: Sixteen selected healthy volunteers (10 males and six females) were subjects of this pilot study. In all subjects, radiographs were taken in comfortable standing and sitting positions. All spinal curvatures including lumbar lordotic angle (LLA), sacral slope (SS), pelvic tilt (PT), and pelvic incidence (PI) were measured on the radiographs. RESULTS: In standing position, the average LLA, SS, PT, and PI were 37.1°, 35.3°, 15.7°, and 51.0°, respectively. In chair sitting, the average LLA, SS, PT, and PI were 17.9°, 20.3°, 28.2°, and 49.5°, respectively. In kneel sitting (Japanese style), the average LLA, SS, PT, and PI were 31.8°, 38.3°, 14.2°, and 52.5°, respectively. In cross-legged sitting (Korean style), the average LLA, SS, PT, and PI were 9.8°, 13.4°, 38.3°, and 51.7°, respectively. LLA in standing (37.1°) and kneel sitting (31.8°) were very similar. Remarkable reduction in LLA was observed in Korean-style cross-legged sitting (9.8°), and LLA in chair sitting (17.9°) was about half of that in standing. SS was similar in standing (35.3°) and kneel sitting (38.3°), and it was reduced remarkably in cross-legged sitting (13.4°). PT was largest in cross-legged sitting (38.3°), and it was similar between standing (15.7°) and kneel sitting (14.2°). PIs were similar in all positions. CONCLUSIONS: The kneel sitting position did not show significant differences with the standing position when assessed using four parameters related to the sagittal spinopelvic alignment, whereas chair sitting and cross-legged sitting positions significantly altered the spinopelvic alignment compared to the standing position.

OCT4 directly regulates stemness and extracellular matrix-related genes in human germ cell tumours.

Germ cell tumours (GCTs) are one of the most threatening malignancies in young men and women. Although several reports have suggested the importance of OCT4 in human GCTs, its role has not been clearly investigated on a molecular level. In this study, we revealed GCT-specific direct transcriptional target genes of OCT4. Conditional knockdown of OCT4 in GCT cell lines reduced cell proliferation by affecting both cell cycle and death. Knockdown of OCT4 also reduced stemness of GCTs, as assessed by the expression of other stemness factors, alkaline phosphatase staining, and tumour sphere formation ability. Analysis of whole mRNA expression patterns among GCT cells harbouring endogenous, depleted, and rescued OCT4 revealed 1133 OCT4 target genes in GCT. Combined analysis of both the chromatin binding signature of OCT4 and the genes whose expression levels were changed by OCT4 revealed 258 direct target genes of OCT4 in GCTs. In a similar way, 594 direct target genes in normal embryonic stem cells (ESCs) were identified. Among these two sets of OCT4 direct target genes, 38 genes were common between GCTs and ESCs, most of which were related to regulation of pluripotency, and 220 genes were specific to GCTs, most of which were related to focal adhesion and extracellular matrix organisation. These results provide a molecular basis for how OCT4 regulates GCT stemness and will aid our understanding of the role of OCT4 in other cancers.

Abundance of C-terminal binding protein isoform is a prerequisite for exit from pluripotency in mouse embryonic stem cells.

Unlike lower organisms, mammals have 2 C-terminal binding protein (Ctbp) isoforms, Ctbp1 and Ctbp2. Ctbp2 is revealed as a key factor involved in determining cell fate decisions by regulating the epigenetic state in active embryonic stem cell (ESC) genes. However, the molecular mechanism underlying how Ctbp1 and Ctbp2 have different roles remains elusive. Here we demonstrate that Ctbp isoform abundance is important for mouse embryonic ESCs (mESCs) to exit from pluripotency. Temporal expression patterns of Ctbp isoforms were quite different; Ctbp2 is more highly expressed in mESCs and decreases during differentiation, while Ctbp1 is constantly expressed at a lower level. Ctbp2 knockdown, but not Ctbp1 knockdown, in mESCs resulted in impaired exit from pluripotency. Interestingly, Ctbp1 and Ctbp2 overexpression in Ctbp2-knockdown mESCs leads to exiting from pluripotency in a manner similar to that of wild-type mESCs. Quantification of Ctbp1 and Ctbp2 revealed that differentiation ability correlates with abundance of Ctbp isoform in undifferentiated mESCs, suggesting that a sufficient amount of Ctbp isoform is a prerequisite for exiting from pluripotency. The results support the contention that 2 redundant Ctbp isoforms regulate elaborate differentiation via temporally distinctive regulatory patterns in mESCs.-Suh, M. Y., Kim, T. W., Lee, H.-T., Shin, J., Kim, J.-H., Jang, H., Kim, H. J., Kim, S.-T., Cho, E.-J., Youn, H.-D. Abundance of C-terminal binding protein isoform is a prerequisite for exit from pluripotency in mouse embryonic stem cells.

Prevalence of Diffuse Idiopathic Skeletal Hyperostosis Diagnosed by Whole Spine Computed Tomography: A Preliminary Study.

BACKGROUND: Diffuse idiopathic skeletal hyperostosis (DISH) is characterized by ossification of the enthesis. The diagnosis has been mainly based on the chest or whole spine lateral plain film. Recently, chest or thoracolumbar computed tomography (CT) has been reported to be more reliable for the diagnosis of DISH. The purposes of this study were to investigate the prevalence and location of DISH and evaluate the prevalence of comorbidities, such as ossification of the posterior longitudinal ligament (OPLL) and ossification of the ligamentum flavum (OLF), using whole spine CT. METHODS: Whole spine CT scans of patients over 16 years of age who were examined at Cheju Halla General Hospital between February 2011 and December 2016 were reviewed for this study. The diagnosis of DISH was made according to the modified Resnick criteria. The prevalence of DISH in each age decade and its location were evaluated. Also, the prevalence of OPLL and OLF in DISH patients was investigated. RESULTS: The overall incidence of DISH was 24.4% (40 of 164 cases). There was no case of DISH in patients in their 40s and younger. The percentile incidences of DISH in patients in their fifth, sixth, seventh, eighth, and ninth decades were 20.0% (4 of 20 cases), 32.3% (10 of 31 cases), 40.0% (10 of 25 cases), 34.5% (10 of 29 cases), and 27.3% (6 of 22 cases), respectively. A strong positive correlation between the age decade and the incidence of DISH was noted (r = 0.853, p = 0.007). DISH patients had higher incidences of OLF (22.5%) and OPLL (37.5%). The most common location of DISH was the middle thoracic spine (90.0%) followed by the lower thoracic spine (87.5%). There was one case of DISH involving only the cervical spine. CONCLUSIONS: The incidence of DISH diagnosed by CT was higher than we expected. Whole spine CT can be a valuable modality to evaluate the location of DISH in the cervical and lumbar spine and the comorbidity rates of OLF and OPLL.

The Meniscofemoral Ligament Mimicking a Lateral Meniscus Tear.

A 58-year-old male patient who had developed left knee pain with a history of trauma was referred to our hospital. Physical examination and further examination by magnetic resonance imaging revealed results that closely resemble a horizontal tear of the lateral meniscus and a tear of the medial meniscus. Arthroscopically, we found a cord-like structure originating from the posterior 1/3 portion of the lateral meniscus and passing obliquely toward the medial femoral condyle in front of the posterior cruciate ligament without a tear of the lateral meniscus. In this report, we describe a rare case of anterior meniscofemoral ligament that was clearly seen on arthroscopy and mimicked a meniscal tear, which is also known as a pseudo-tear of the meniscus.

Latent feature representation with depth directional long-term recurrent learning for breast masses in digital breast tomosynthesis.

Characterization of masses in computer-aided detection systems for digital breast tomosynthesis (DBT) is an important step to reduce false positive (FP) rates. To effectively differentiate masses from FPs in DBT, discriminative mass feature representation is required. In this paper, we propose a new latent feature representation boosted by depth directional long-term recurrent learning for characterizing malignant masses. The proposed network is designed to encode mass characteristics in two parts. First, 2D spatial image characteristics of DBT slices are encoded as a slice feature representation by convolutional neural network (CNN). Then, depth directional characteristics of masses among the slice feature representations are encoded by the proposed depth directional long-term recurrent learning. In addition, to further improve the class discriminability of latent feature representation, we have devised three objective functions aiming to (a) minimize classification error, (b) minimize intra-class variation within the same class, and (c) preserve feature representation consistency in a central slice. Experimental results have demonstrated that the proposed latent feature representation achieves a higher level of classification performance in terms of receiver operating characteristic (ROC) curves and the area under the ROC curve values compared to performance with feature representation learned by conventional CNN and hand-crafted features.

Migration and invasion of drug-resistant lung adenocarcinoma cells are dependent on mitochondrial activity.

A small proportion of cancer cells have stem-cell-like properties, are resistant to standard therapy and are associated with a poor prognosis. The metabolism of such drug-resistant cells differs from that of nearby non-resistant cells. In this study, the metabolism of drug-resistant lung adenocarcinoma cells was investigated. The expression of genes associated with oxidative phosphorylation in the mitochondrial membrane was negatively correlated with the prognosis of lung adenocarcinoma. Because the mitochondrial membrane potential (MMP) reflects the functional status of mitochondria and metastasis is the principal cause of death due to cancer, the relationship between MMP and metastasis was evaluated. Cells with a higher MMP exhibited greater migration and invasion than those with a lower MMP. Cells that survived treatment with cisplatin, a standard chemotherapeutic drug for lung adenocarcinoma, exhibited increased MMP and enhanced migration and invasion compared with parental cells. Consistent with these findings, inhibition of mitochondrial activity significantly impeded the migration and invasion of cisplatin-resistant cells. RNA-sequencing analysis indicated that the expression of mitochondrial complex genes was upregulated in cisplatin-resistant cells. These results suggested that drug-resistant cells have a greater MMP and that inhibition of mitochondrial activity could be used to prevent metastasis of drug-resistant lung adenocarcinoma cells.

Inpatient treatment effect and Minnesota Multiphasic Personality Inventory characteristics of motor vehicle collision injuries in a traditional korean medicine hospital: Retrospective chart review.

OBJECTIVES: To examine the changes in pain, disability, and quality of life in motor vehicle collision injury (MVCI) patients after treatment with traditional Korean medicine (TKM), and to investigate the psychological characteristics of these patients. METHODS: Forty-one patients with MVCI were treated with TKM including acupuncture, pharmacopuncture, moxibustion, cupping, herbal medication, chuna manual therapy, and physical therapy. Numeric Rating Scale (NRS), Medical Outcomes Study 36-Item Short Form Health Survey (SF-36), Neck Disability Index (NDI), Oswestry Disability Index (ODI), and Lysholm Knee Scoring Scale were assessed at admission and discharge. The Minnesota Multiphasic Personality Inventory (MMPI) was assessed at admission. RESULTS: After treatment, NRS scores for headache, cervical pain, and lumbar pain were significantly decreased (P<0.05); NDI, ODI scores were significantly decreased, and Lysholm score was signifificantly increased (P<0.05). The following SF-36 scores were signifificantly increased: physical and mental component summary, bodily pain, role-physical, role-emotional, social functioning, and mental health scores (all P<0.05). MMPI identifified 3-1 profifile conversion V shape. CONCLUSIONS: Treatment of MVCI with TKM provided effective management of complex symptoms such as pain, disability, and loss of quality of life. A comprehensive plan must be implemented for treatment and research in cases of MVCIs owing to the correlation between physical symptoms and psychological profifiles.