Deep Learning Enhances Multiparametric Dynamic Volumetric Photoacoustic Computed Tomography In Vivo (DL-PACT).

Photoacoustic computed tomography (PACT) has become a premier preclinical and clinical imaging modality. Although PACT's image quality can be dramatically improved with a large number of ultrasound (US) transducer elements and associated multiplexed data acquisition systems, the associated high system cost and/or slow temporal resolution are significant problems. Here, a deep learning-based approach is demonstrated that qualitatively and quantitively diminishes the limited-view artifacts that reduce image quality and improves the slow temporal resolution. This deep learning-enhanced multiparametric dynamic volumetric PACT approach, called DL-PACT, requires only a clustered subset of many US transducer elements on the conventional multiparametric PACT. Using DL-PACT, high-quality static structural and dynamic contrast-enhanced whole-body images as well as dynamic functional brain images of live animals and humans are successfully acquired, all in a relatively fast and cost-effective manner. It is believed that the strategy can significantly advance the use of PACT technology for preclinical and clinical applications such as neurology, cardiology, pharmacology, endocrinology, and oncology.

High throughput screening of mesenchymal stem cell lines using deep learning.

Mesenchymal stem cells (MSCs) are increasingly used as regenerative therapies for patients in the preclinical and clinical phases of various diseases. However, the main limitations of such therapies include functional heterogeneity and the lack of appropriate quality control (QC) methods for functional screening of MSC lines; thus, clinical outcomes are inconsistent. Recently, machine learning (ML)-based methods, in conjunction with single-cell morphological profiling, have been proposed as alternatives to conventional in vitro/vivo assays that evaluate MSC functions. Such methods perform in silico analyses of MSC functions by training ML algorithms to find highly nonlinear connections between MSC functions and morphology. Although such approaches are promising, they are limited in that extensive, high-content single-cell imaging is required; moreover, manually identified morphological features cannot be generalized to other experimental settings. To address these limitations, we propose an end-to-end deep learning (DL) framework for functional screening of MSC lines using live-cell microscopic images of MSC populations. We quantitatively evaluate various convolutional neural network (CNN) models and demonstrate that our method accurately classifies in vitro MSC lines to high/low multilineage differentiating stress-enduring (MUSE) cells markers from multiple donors. A total of 6,120 cell images were obtained from 8 MSC lines, and they were classified into two groups according to MUSE cell markers analyzed by immunofluorescence staining and FACS. The optimized DenseNet121 model showed area under the curve (AUC) 0.975, accuracy 0.922, F1 0.922, sensitivity 0.905, specificity 0.942, positive predictive value 0.940, and negative predictive value 0.908. Therefore, our DL-based framework is a convenient high-throughput method that could serve as an effective QC strategy in future clinical biomanufacturing processes.

Ovarian tumor diagnosis using deep convolutional neural networks and a denoising convolutional autoencoder.

Discrimination of ovarian tumors is necessary for proper treatment. In this study, we developed a convolutional neural network model with a convolutional autoencoder (CNN-CAE) to classify ovarian tumors. A total of 1613 ultrasound images of ovaries with known pathological diagnoses were pre-processed and augmented for deep learning analysis. We designed a CNN-CAE model that removes the unnecessary information (e.g., calipers and annotations) from ultrasound images and classifies ovaries into five classes. We used fivefold cross-validation to evaluate the performance of the CNN-CAE model in terms of accuracy, sensitivity, specificity, and the area under the receiver operating characteristic curve (AUC). Gradient-weighted class activation mapping (Grad-CAM) was applied to visualize and verify the CNN-CAE model results qualitatively. In classifying normal versus ovarian tumors, the CNN-CAE model showed 97.2% accuracy, 97.2% sensitivity, and 0.9936 AUC with DenseNet121 CNN architecture. In distinguishing malignant ovarian tumors, the CNN-CAE model showed 90.12% accuracy, 86.67% sensitivity, and 0.9406 AUC with DenseNet161 CNN architecture. Grad-CAM showed that the CNN-CAE model recognizes valid texture and morphology features from the ultrasound images and classifies ovarian tumors from these features. CNN-CAE is a feasible diagnostic tool that is capable of robustly classifying ovarian tumors by eliminating marks on ultrasound images. CNN-CAE demonstrates an important application value in clinical conditions.

Bi-Modal Transfer Learning for Classifying Breast Cancers via Combined B-Mode and Ultrasound Strain Imaging.

Although accurate detection of breast cancer still poses significant challenges, deep learning (DL) can support more accurate image interpretation. In this study, we develop a highly robust DL model based on combined B-mode ultrasound (B-mode) and strain elastography ultrasound (SE) images for classifying benign and malignant breast tumors. This study retrospectively included 85 patients, including 42 with benign lesions and 43 with malignancies, all confirmed by biopsy. Two deep neural network models, AlexNet and ResNet, were separately trained on combined 205 B-mode and 205 SE images (80% for training and 20% for validation) from 67 patients with benign and malignant lesions. These two models were then configured to work as an ensemble using both image-wise and layer-wise and tested on a dataset of 56 images from the remaining 18 patients. The ensemble model captures the diverse features present in the B-mode and SE images and also combines semantic features from AlexNet and ResNet models to classify the benign from the malignant tumors. The experimental results demonstrate that the accuracy of the proposed ensemble model is 90%, which is better than the individual models and the model trained using B-mode or SE images alone. Moreover, some patients that were misclassified by the traditional methods were correctly classified by the proposed ensemble method. The proposed ensemble DL model will enable radiologists to achieve superior detection efficiency owing to enhance classification accuracy for breast cancers in ultrasound (US) images.

Gallbladder Polyp Classification in Ultrasound Images Using an Ensemble Convolutional Neural Network Model.

Differential diagnosis of true gallbladder polyps remains a challenging task. This study aimed to differentiate true polyps in ultrasound images using deep learning, especially gallbladder polyps less than 20 mm in size, where clinical distinction is necessary. A total of 501 patients with gallbladder polyp pathology confirmed through cholecystectomy were enrolled from two tertiary hospitals. Abdominal ultrasound images of gallbladder polyps from these patients were analyzed using an ensemble model combining three convolutional neural network (CNN) models and a 5-fold cross-validation. True polyp diagnosis with the ensemble model that learned only using ultrasonography images achieved an area under receiver operating characteristic curve (AUC) of 0.8960 and accuracy of 83.63%. After adding patient age and polyp size information, the diagnostic performance of the ensemble model improved, with a high specificity of 88.35%, AUC of 0.9082, and accuracy of 87.61%, outperforming the individual CNN models constituting the ensemble model. In the subgroup analysis, the ensemble model showed the best performance with AUC of 0.9131 for polyps larger than 10 mm. Our proposed ensemble model that combines three CNN models classifies gallbladder polyps of less than 20 mm in ultrasonography images with high accuracy and can be useful for avoiding unnecessary cholecystectomy with high specificity.

Convolutional neural network-based object detection model to identify gastrointestinal stromal tumors in endoscopic ultrasound images.

BACKGROUND AND AIM: We aimed to develop a convolutional neural network (CNN)-based object detection model for the discrimination of gastric subepithelial tumors, such as gastrointestinal stromal tumors (GISTs), and leiomyomas, in endoscopic ultrasound (EUS) images. METHODS: We used 376 images from 114 patients with histologically confirmed gastric GIST or leiomyoma to train the EUS-CNN. We constructed the EUS-CNN using an EfficientNet CNN model for feature extraction and a weighted bi-directional feature pyramid network for object detection. We assessed the performance of our EUS-CNN by calculating its accuracy, sensitivity, specificity, and area under receiver operating characteristic curve (AUC) using a validation set of 170 images from 54 patients. Four EUS experts and 15 EUS trainees were asked to judge the same validation dataset, and the diagnostic yields were compared between the EUS-CNN and human assessments. RESULTS: In the per-image analysis, the sensitivity, specificity, accuracy, and AUC of our EUS-CNN were 95.6%, 82.1%, 91.2%, and 0.9234, respectively. In the per-patient analysis, the sensitivity, specificity, accuracy, and AUC for our object detection model were 100.0%, 85.7%, 96.3%, and 0.9929, respectively. The EUS-CNN outperformed human assessment in terms of accuracy, sensitivity, and negative predictive value. CONCLUSIONS: We developed the EUS-CNN system, which demonstrated high diagnostic ability for gastric GIST prediction. This EUS-CNN system can be helpful not only for less-experienced endoscopists but also for experienced ones. Additional EUS image accumulation and prospective studies are required alongside validation in a large multicenter trial.

Keratinocytes-Derived Reactive Oxygen Species Play an Active Role to Induce Type 2 Inflammation of the Skin: A Pathogenic Role of Reactive Oxygen Species at the Early Phase of Atopic Dermatitis.

BACKGROUND: Atopic dermatitis (AD) is characterized by chronic, relapsing skin inflammation (eczema) with itchy sensation. Keratinocytes, which are located at the outermost part of our body, are supposed to play important roles at the early phase of type 2 inflammation including AD pathogenesis. OBJECTIVE: The purpose of this study was to evaluate whether keratinocytes-derived reactive oxygen species (ROS) could be produced by the allergens or non-allergens, and the keratinocytes-derived ROS could modulate a set of biomarkers for type 2 inflammation of the skin. METHODS: Normal human epidermal keratinocytes (NHEKs) were treated with an allergen of house dust mites (HDM) or a non-allergen of compound 48/80 (C48/80). Then, biomarkers for type 2 inflammation of the skin including those for neurogenic inflammation were checked by reverse transcriptase-polymerase chain reaction and western immunoblot experiments. RESULTS: HDM or C48/80 was found to upregulate expression levels of our tested biomarkers, including type 2 T helper-driving pathway (KLK5, PAR2, and NFκB), epithelial-cell-derived cytokines (thymic stromal lymphopoietin, interleukin [IL]-25, IL-33), and neurogenic inflammation (NGF, CGRP). The HDM- or C-48/80-induced expression levels of the biomarkers could be blocked by an antioxidant treatment with 5 mM N-acetyl-cysteine. In contrast, pro-oxidant treatment with 1 mM H2O2 could upregulate expression levels of the tested biomarkers in NHEKs. CONCLUSION: Our results reveal that keratinocytes-derived ROS, irrespective to their origins from allergens or non-allergens, have a potential to induce type 2 inflammation of AD skin.

HIGH CROSSOVER RATE1 encodes PROTEIN PHOSPHATASE X1 and restricts meiotic crossovers in Arabidopsis.

Meiotic crossovers are tightly restricted in most eukaryotes, despite an excess of initiating DNA double-strand breaks. The majority of plant crossovers are dependent on class I interfering repair, with a minority formed via the class II pathway. Class II repair is limited by anti-recombination pathways; however, similar pathways repressing class I crossovers have not been identified. Here, we performed a forward genetic screen in Arabidopsis using fluorescent crossover reporters to identify mutants with increased or decreased recombination frequency. We identified HIGH CROSSOVER RATE1 (HCR1) as repressing crossovers and encoding PROTEIN PHOSPHATASE X1. Genome-wide analysis showed that hcr1 crossovers are increased in the distal chromosome arms. MLH1 foci significantly increase in hcr1 and crossover interference decreases, demonstrating an effect on class I repair. Consistently, yeast two-hybrid and in planta assays show interaction between HCR1 and class I proteins, including HEI10, PTD, MSH5 and MLH1. We propose that HCR1 plays a major role in opposition to pro-recombination kinases to restrict crossovers in Arabidopsis.

Estimating the phase volume fraction of multi-phase steel via unsupervised deep learning.

Advanced high strength steel (AHSS) is a steel of multi-phase microstructure that is processed under several conditions to meet the current high-performance requirements from the industry. Deep neural network (DNN) has emerged as a promising tool in materials science for the task of estimating the phase volume fraction of these steels. Despite its advantages, one of its major drawbacks is its requirement of a sufficient amount of training data with correct labels to the network. This often comes as a challenge in many areas where obtaining data and labeling it is extremely labor-intensive. To overcome this challenge, an unsupervised way of learning DNN, which does not require any manual labeling, is proposed. Information maximizing generative adversarial network (InfoGAN) is used to learn the underlying probability distribution of each phase and generate realistic sample points with class labels. Then, the generated data is used for training an MLP classifier, which in turn predicts the labels for the original dataset. The result shows a mean relative error of 4.53% at most, while it can be as low as 0.73%, which implies the estimated phase fraction closely matches the true phase fraction. This presents the high feasibility of using the proposed methodology for fast and precise estimation of phase volume fraction in both industry and academia.

Convolutional Neural Network Classifies Pathological Voice Change in Laryngeal Cancer with High Accuracy.

Voice changes may be the earliest signs in laryngeal cancer. We investigated whether automated voice signal analysis can be used to distinguish patients with laryngeal cancer from healthy subjects. We extracted features using the software package for speech analysis in phonetics (PRAAT) and calculated the Mel-frequency cepstral coefficients (MFCCs) from voice samples of a vowel sound of /a:/. The proposed method was tested with six algorithms: support vector machine (SVM), extreme gradient boosting (XGBoost), light gradient boosted machine (LGBM), artificial neural network (ANN), one-dimensional convolutional neural network (1D-CNN) and two-dimensional convolutional neural network (2D-CNN). Their performances were evaluated in terms of accuracy, sensitivity, and specificity. The result was compared with human performance. A total of four volunteers, two of whom were trained laryngologists, rated the same files. The 1D-CNN showed the highest accuracy of 85% and sensitivity and sensitivity and specificity levels of 78% and 93%. The two laryngologists achieved accuracy of 69.9% but sensitivity levels of 44%. Automated analysis of voice signals could differentiate subjects with laryngeal cancer from those of healthy subjects with higher diagnostic properties than those performed by the four volunteers.

Recent Advances in the Application of Artificial Intelligence in Otorhinolaryngology-Head and Neck Surgery.

This study presents an up-to-date survey of the use of artificial intelligence (AI) in the field of otorhinolaryngology, considering opportunities, research challenges, and research directions. We searched PubMed, the Cochrane Central Register of Controlled Trials, Embase, and the Web of Science. We initially retrieved 458 articles. The exclusion of non-English publications and duplicates yielded a total of 90 remaining studies. These 90 studies were divided into those analyzing medical images, voice, medical devices, and clinical diagnoses and treatments. Most studies (42.2%, 38/90) used AI for image-based analysis, followed by clinical diagnoses and treatments (24 studies). Each of the remaining two subcategories included 14 studies. Machine learning and deep learning have been extensively applied in the field of otorhinolaryngology. However, the performance of AI models varies and research challenges remain.

The 308-nm excimer laser treatment does not increase the risk of skin cancer in patients with vitiligo: A population-based retrospective cohort study.

A 308-nm excimer laser (EL) has been widely used to treat patients with localized vitiligo. However, data are rare on the influence of EL treatment on the risks of skin cancer. To evaluate the skin cancer risks after long-term EL treatment, we performed a nationwide population-based retrospective cohort study using the Korean National Health Insurance Claims Database. A total of 5,052 patients with vitiligo were classified into three groups according to the EL treatment sessions between 2009 and 2016: no, 50-99, and 100 or more EL treatments after 2-year washout period (2007 and 2008). Using multivariable Cox proportional hazard models, we found that the risks of actinic keratosis, non-melanoma skin cancers, and melanoma did not significantly differ among the groups, respectively. In conclusion, EL treatment would not increase the risks of premalignant skin lesions and skin cancers in patients with vitiligo. Based on our results, EL is likely to be a safe treatment option for patients with localized vitiligo.

Development and validation of the Vitiligo Extent Score for a Target Area (VESTA) to assess the treatment response of a target lesion.

Since localized treatment for vitiligo is as essential as systemic treatment, a reliable instrument for target evaluation is needed besides those for whole body evaluation. We developed the Vitiligo Extent Score for a Target Area (VESTA) using reference images of both marginal and perifollicular repigmentation to measure the repigmentation rate (%) in a target lesion. In the validation study, a total of 65 dermatologists in 10 institutes evaluated 17 pairs of vitiligo images (pre- and post-treatment) using both a rough estimate and the VESTA. The VESTA (concordance correlation coefficient: 0.949, 95% confidence interval [CI] 0.942-0.955) was significantly more accurate than the rough estimate (0.896, 95% CI: 0.883-0.908). It was also associated with better inter-rater reliability over the rough estimate, albeit not significant. The VESTA can afford intuitive, convenient, and reliable assessment of the treatment response in a target area, and would be useful in clinical practice as well as retrospective studies.

Paraneoplastic Pyoderma Gangrenosum Associated with Rectal Adenocarcinoma.

Pyoderma gangrenosum (PG) is a rare chronic neutrophilic dermatosis characterized by painful necrotic ulceration. The most common diseases associated with PG are inflammatory bowel disease, certain rheumatologic and hematologic diseases, and malignancy. Here, we describe the case of a 60-year-old man who presented with pruritic and painful erythematous ulcerative macules and patches on both lower extremities, and was diagnosed with PG based on his clinical and histologic features. His PG became exacerbated despite standard therapy with a high-dose systemic steroid in combination with dapsone and cyclosporine. Systemic evaluation of underlying conditions revealed rectal adenocarcinoma at the rectosigmoid junction (T3N0M0), which was completely removed via Hartmann's procedure followed by adjuvant chemotherapy. Two months after anticancer therapy, his PG was completely healed with hypertrophic scarring. Herein, we present the first case of paraneoplastic PG caused by rectal adenocarcinoma in Korea.

Sensitivity and Usefulness of VE1 Immunohistochemical Staining in Acral Melanomas with BRAF Mutation.

BACKGROUND: Acral melanomas are known to have a low frequency of BRAF mutation, in contrary to higher KIT mutation. Recently, VE1 immunostaining was reported to have a good correlation with BRAF mutation status. OBJECTIVE: We aimed to evaluate the clinicopathological features of BRAF-mutated acral melanomas and validate the correlation of the VE1 immunohistochemical stains in those cases. METHODS: The clinical features (age, sex, anatomical site), and histopathological characteristics of 41 patients with acral melanoma were evaluated. We performed a next-generation sequencing to detect BRAF mutation status. We also determined the correlation of VE1 immunohistochemical staining with BRAF mutation status. RESULTS: Among 19 acral melanomas with BRAF mutation, common histopathological subtype was acral lentiginous melanoma (8/19, 42%) and nodular melanoma (8/19, 42%) and superficial spreading melanoma (3/19, 16%) followed. VE1 immunostaining results were positive in all 15 cases with BRAF V600E mutation (sensitivity 100%), and negative in 4 cases of BRAF non-V600E mutation. However, VE1 immunostaining was negative in all 22 patients with BRAF wild-type. CONCLUSION: VE1 immunostaining had a good correlation with BRAF V600E mutation status.

Citron Essential Oils Alleviate the Mediators Related to Rosacea Pathophysiology in Epidermal Keratinocytes.

BACKGROUND: Citron is well known for an abundance of antioxidative and anti-inflammatory ingredients such as vitamin C, polyphenol compounds, flavonoids, and limonoids. OBJECTIVE: In this study, we aimed to evaluate the effects of citron essential oils on rosacea mediators in activated keratinocytes in vitro. METHODS: Normal human epidermal keratinocytes (NHEKs) were stimulated with 1α, 25-dihydroxyvitamin D3 (VD3) and interleukin 33 (IL-33) with LL-37 to induce rosacea mediators such as kallikrein 5 (KLK5), cathelicidin, vascular endothelial growth factor (VEGF), and transient receptor potential vanilloid 1 (TRPV1). These mediators were analyzed by performing reverse-transcription polymerase chain reaction (PCR), quantitative real-time PCR, immunocytofluorescence and enzyme-linked immunosorbent assay after NHEKs were treated with citron seed and unripe citron essential oils. RESULTS: The messenger RNA (mRNA) and protein levels of KLK5 and LL-37 induced by VD3 were suppressed by citron seed and unripe citron essential oils. Furthermore, the mRNA and protein levels of VEGF and TRPV1 induced by IL-33 with LL-37 were also suppressed by citron essential oils. CONCLUSION: These results show that citron essential oils have suppressive effects on rosacea mediators in activated epidermal keratinocytes, which indicates that the citron essential oils may be valuable adjuvant therapeutic agents for rosacea.