On-Chip Annealing Using Embedded Micro-Heater for Highly Sensitive and Selective Gas Detection.

The demand for gas sensing systems that enable fast and precise gas recognition is growing rapidly. However, substantial challenges arise from the complex fabrication process of sensor arrays, time-consuming data transmission to an external processor, and high energy consumption in multi-stage data processing. In this study, a gas sensing system using on-chip annealing for fast and power-efficient gas detection is proposed. By utilizing a micro-heater embedded in the gas sensor, the sensing material of adjacent sensors in the same substrate can be easily varied without further fabrication steps. The response to oxidizing gas is constrained in metal oxide (MOX) sensing material with small grain sizes, as the depletion width of grain cannot extend beyond the grain size during the gas reaction. On the other hand, the response to reducing gases and humidity, which decrease the depletion width, is less affected by grain sizes. A readout circuit integrating a differential amplifier and dual FET-type gas sensors effectively emphasizes the response to oxidizing gases by canceling the response to reducing gases and humidity. The selective on-chip annealing method is applicable to various MOX sensing materials, demonstrating its potential for application in commercial fields due to its simplicity and expandability.

Exploring aryl hydrocarbon receptor expression and distribution in the tumor microenvironment, with a focus on immune cells, in various solid cancer types.

Introduction: Aryl hydrocarbon receptor (AhR) is a transcription factor that performs various functions upon ligand activation. Several studies have explored the role of AhR expression in tumor progression and immune surveillance. Nevertheless, investigations on the distribution of AhR expression, specifically in cancer or immune cells in the tumor microenvironment (TME), remain limited. Examining the AhR expression and distribution in the TME is crucial for gaining insights into the mechanism of action of AhR-targeting anticancer agents and their potential as biomarkers. Methods: Here, we used multiplexed immunohistochemistry (mIHC) and image cytometry to investigate the AhR expression and distribution in 513 patient samples, of which 292 are patients with one of five solid cancer types. Additionally, we analyzed the nuclear and cytosolic distribution of AhR expression. Results: Our findings reveal that AhR expression was primarily localized in cancer cells, followed by stromal T cells and macrophages. Furthermore, we observed a positive correlation between the nuclear and cytosolic expression of AhR, indicating that the expression of AhR as a biomarker is independent of its localization. Interestingly, the expression patterns of AhR were categorized into three clusters based on the cancer type, with high AhR expression levels being found in regulatory T cells (Tregs) in non-small cell lung cancer (NSCLC). Discussion: These findings are anticipated to serve as pivotal evidence for the design of clinical trials and the analysis of the anticancer mechanisms of AhR-targeting therapies.

Prediction of neovascular age-related macular degeneration recurrence using optical coherence tomography images with a deep neural network.

Neovascular age-related macular degeneration (nAMD) can result in blindness if left untreated, and patients often require repeated anti-vascular endothelial growth factor injections. Although, the treat-and-extend method is becoming popular to reduce vision loss attributed to recurrence, it may pose a risk of overtreatment. This study aimed to develop a deep learning model based on DenseNet201 to predict nAMD recurrence within 3 months after confirming dry-up 1 month following three loading injections in treatment-naïve patients. A dataset of 1076 spectral domain optical coherence tomography (OCT) images from 269 patients diagnosed with nAMD was used. The performance of the model was compared with that of 6 ophthalmologists, using 100 randomly selected samples. The DenseNet201-based model achieved 53.0% accuracy in predicting nAMD recurrence using a single pre-injection image and 60.2% accuracy after viewing all the images immediately after the 1st, 2nd, and 3rd injections. The model outperformed experienced ophthalmologists, with an average accuracy of 52.17% using a single pre-injection image and 53.3% after examining four images before and after three loading injections. In conclusion, the artificial intelligence model demonstrated a promising ability to predict nAMD recurrence using OCT images and outperformed experienced ophthalmologists. These findings suggest that deep learning models can assist in nAMD recurrence prediction, thus improving patient outcomes and optimizing treatment strategies.

CD81 and CD82 expressing tumor-infiltrating lymphocytes in the NSCLC tumor microenvironment play a crucial role in T-cell activation and cytokine production.

Introduction: To understand the immune system within the tumor microenvironment (TME) of non-small cell lung cancer (NSCLC), it is crucial to elucidate the characteristics of molecules associated with T cell activation. Methods: We conducted an in-depth analysis using single-cell RNA sequencing data obtained from tissue samples of 19 NSCLC patients. T cells were classified based on the Tumor Proportion Score (TPS) within the tumor region, and molecular markers associated with activation and exhaustion were analyzed in T cells from high TPS areas. Results: Notably, tetraspanins CD81 and CD82, belonging to the tetraspanin protein family, were found to be expressed in activated T cells, particularly in cytotoxic T cells. These tetraspanins showed strong correlations with activation and exhaustion markers. In vitro experiments confirmed increased expression of CD81 and CD82 in IL-2-stimulated T cells. T cells were categorized into CD81highCD82high and CD81lowCD82low groups based on their expression levels, with CD81highCD82high T cells exhibiting elevated activation markers such as CD25 and CD69 compared to CD81lowCD82low T cells. This trend was consistent across CD3+, CD8+, and CD4+ T cell subsets. Moreover, CD81highCD82high T cells, when stimulated with anti-CD3, demonstrated enhanced secretion of cytokines such as IFN-γ, TNF-α, and IL-2, along with an increase in the proportion of memory T cells. Bulk RNA sequencing results after sorting CD81highCD82high and CD81lowCD82low T cells consistently supported the roles of CD81 and CD82. Experiments with overexpressed CD81 and CD82 showed increased cytotoxicity against target cells. Discussion: These findings highlight the multifaceted roles of CD81 and CD82 in T cell activation, cytokine production, memory subset accumulation, and target cell cytolysis. Therefore, these findings suggest the potential of CD81 and CD82 as promising candidates for co-stimulatory molecules in immune therapeutic strategies for cancer treatment within the intricate TME.

Biogenic Salvia species synthesized silver nanoparticles with catalytic, sensing, antimicrobial, and antioxidant properties.

Salvia (Lamiaceae family) is used as a brain tonic to improve cognitive function. The species including S. plebeia and S. moorcroftiana are locally used to cure hepatitis, cough, tumours, hemorrhoids, diarrhoea, common cold, flu, and asthma. To the best of authors' knowledge, no previous study has been conducted on synthesis of S. plebeia and S. moorcroftiana silver nanoparticles (SPAgNPs and SMAgNPs). The study was aimed to synthesize AgNPs from the subject species aqueous and ethanol extracts, and assess catalytic potential in degradation of standard and extracted (from yums, candies, and snacks) dyes, nitrophenols, and antibiotics. The study also aimed at AgNPs as probe in sensing metalloids and heavy metal ions including Pb2+, Cu2+, Fe3+, Ni2+, and Zn2+. From the results, it was found that Salvia aqueous extract afforded stable AgNPs in 1:9 and 1:15 (quantity of aqueous extract and silver nitrate solution concentration) whereas ethanol extract yielded AgNPs in 1:10 (quantity of ethanol extract and silver nitrate solution concentration) reacted in sunlight. The size of SPAgNPs and SMAgNPs determined by scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were 21.7 nm and 19.9 nm, with spherical, cylindrical, and deep hollow morphology. The synthesized AgNPs demonstrated significant potential as catalyst in dyes; Congo red (85 %), methylene blue (75 %), Rhodamine B (<50 %), nitrophenols; ortho-nitrophenol (95-98 %) and para-nitrophenol (95-98 %), dyes extracted from food samples including yums, candies, and snacks. The antibiotics (amoxicillin, doxycycline, levofloxacin) degraded up to 80 %-95 % degradation. Furthermore, the synthesized AgNPs as probe in sensing of Pb2+, Cu2+, and Fe3+ in Kabul river water, due to agglomeration, caused a significant decrease and bathochromic shift of SPR band (430 nm) when analyzed after 30 min. The Pb2+ ions was comparatively more agglomerated and chelated. Thus, the practical applicability of AgNPs in Pb2+ sensing was significant. Based on the results of this research study, the synthesized AgNPs could provide promising efficiency in wastewater treatment containing organic dyes, antibiotics, and heavy metals.

Biomass- and Carbon Dioxide-Derived Polyurethane Networks for Thermal Interface Material Applications.

Recent environmental concerns have increased demand for renewable polymers and sustainable green resource usage, such as biomass-derived components and carbon dioxide (CO2). Herein, we present crosslinked polyurethanes (CPUs) fabricated from CO2- and biomass-derived monomers via a facile solvent-free ball milling process. Furan-containing bis(cyclic carbonate)s were synthesized through CO2 fixation and further transformed to tetraols, denoted FCTs, by aminolysis and utilized in CPU synthesis. Highly dispersed polyurethane-based hybrid composites (CPU-Ag) were also manufactured using a similar ball milling process. Due to the malleability of the CPU matrix, enabled by transcarbamoylation (dynamic covalent chemistry), CPU-based composites are expected to present very low interfacial thermal resistance between the heat sink and heat source. The characteristics of the dynamic covalent bond (i.e., urethane exchange reaction) were confirmed by the results of dynamic mechanical thermal analysis and stress relaxation analysis. Importantly, the high thermal conductivity of the CPU-based hybrid material was confirmed using laser flash analysis (up to 51.1 W/m·K). Our mechanochemical approach enables the facile preparation of sustainable polymers and hybrid composites for functional application.

High-Performing and Capacitive-Matched Triboelectric Implants Driven by Ultrasound.

In implantable bioelectronics, which aim for semipermanent use of devices, biosafe energy sources and packaging materials to protect devices are essential elements. However, research so far has been conducted in a direction where they cannot coexist. Here, the development of capacitance-matched triboelectric implants driven is reported by ultrasound under 500 mW cm-2 safe intensity and realize a battery-free, miniatured, and wireless neurostimulator with full titanium (Ti) packaging. The triboelectric implant with high dielectric composite, which has ultralow output impedance, can efficiently deliver sufficient power to generate the stimulation pulse without an energy-storing battery, despite ultrasound attenuation due to the Ti, and has the highest energy transmission efficiency among those reported so far. In vivo study using a rat model demonstrated that the proposed device system is an effective solution for relieving urinary symptoms. These achievements provide a significant step toward permanently implantable devices for controlling human organs and treating various diseases.

Developing and Evaluating an AI-Based Computer-Aided Diagnosis System for Retinal Disease: Diagnostic Study for Central Serous Chorioretinopathy.

BACKGROUND: Although previous research has made substantial progress in developing high-performance artificial intelligence (AI)-based computer-aided diagnosis (AI-CAD) systems in various medical domains, little attention has been paid to developing and evaluating AI-CAD system in ophthalmology, particularly for diagnosing retinal diseases using optical coherence tomography (OCT) images. OBJECTIVE: This diagnostic study aimed to determine the usefulness of a proposed AI-CAD system in assisting ophthalmologists with the diagnosis of central serous chorioretinopathy (CSC), which is known to be difficult to diagnose, using OCT images. METHODS: For the training and evaluation of the proposed deep learning model, 1693 OCT images were collected and annotated. The data set included 929 and 764 cases of acute and chronic CSC, respectively. In total, 66 ophthalmologists (2 groups: 36 retina and 30 nonretina specialists) participated in the observer performance test. To evaluate the deep learning algorithm used in the proposed AI-CAD system, the training, validation, and test sets were split in an 8:1:1 ratio. Further, 100 randomly sampled OCT images from the test set were used for the observer performance test, and the participants were instructed to select a CSC subtype for each of these images. Each image was provided under different conditions: (1) without AI assistance, (2) with AI assistance with a probability score, and (3) with AI assistance with a probability score and visual evidence heatmap. The sensitivity, specificity, and area under the receiver operating characteristic curve were used to measure the diagnostic performance of the model and ophthalmologists. RESULTS: The proposed system achieved a high detection performance (99% of the area under the curve) for CSC, outperforming the 66 ophthalmologists who participated in the observer performance test. In both groups, ophthalmologists with the support of AI assistance with a probability score and visual evidence heatmap achieved the highest mean diagnostic performance compared with that of those subjected to other conditions (without AI assistance or with AI assistance with a probability score). Nonretina specialists achieved expert-level diagnostic performance with the support of the proposed AI-CAD system. CONCLUSIONS: Our proposed AI-CAD system improved the diagnosis of CSC by ophthalmologists, which may support decision-making regarding retinal disease detection and alleviate the workload of ophthalmologists.

Efficient Hybrid Training Method for Neuromorphic Hardware Using Analog Nonvolatile Memory.

Neuromorphic hardware using nonvolatile analog synaptic devices provides promising advantages of reducing energy and time consumption for performing large-scale vector-matrix multiplication (VMM) operations. However, the reported training methods for neuromorphic hardware have appreciably shown reduced accuracy due to the nonideal nature of analog devices, and use conductance tuning protocols that require substantial cost for training. Here, we propose a novel hybrid training method that efficiently trains the neuromorphic hardware using nonvolatile analog memory cells, and experimentally demonstrate the high performance of the method using the fabricated hardware. Our training method does not rely on the conductance tuning protocol to reflect weight updates to analog synaptic devices, which significantly reduces online training costs. When the proposed method is applied, the accuracy of the hardware-based neural network approaches to that of the software-based neural network after only one-epoch training, even if the fabricated synaptic array is trained for only the first synaptic layer. Also, the proposed hybrid training method can be efficiently applied to low-power neuromorphic hardware, including various types of synaptic devices whose weight update characteristics are extremely nonlinear. This successful demonstration of the proposed method in the fabricated hardware shows that neuromorphic hardware using nonvolatile analog memory cells becomes a more promising platform for future artificial intelligence.

Fe-Catalyzed C-H Alkenylation of Dialkyl Anilines with Disulfonylethenes.

The oxidative alkenylation reaction of α-aminoalkyl C(sp3)-H bonds has been investigated with (E)-1,2-bis(sulfonyl)ethenes. The catalytic process of iron-polypyridyl complexes drives the single-electron oxidation of dialkyl anilines, resulting in the formation of α-aminoalkyl radical species. Subsequent cascades of radical addition and elimination reactions ensue, ultimately leading to the generation of sulfonylated allylic amine products. The utility of these products extends further, enabling the synthesis of multisubstituted heterocycles like pyrroles, pyrazines, and triazoles.

Polo-like Kinase 4: A Multifaceted Marker Linking Tumor Aggressiveness and Unfavorable Prognosis, and Insights into Therapeutic Strategies.

(1) Background: This study investigated whether polo-like kinase 4 (PLK4) is a suitable therapeutic target or biomarker for lung adenocarcinoma (LUAD). (2) Methods: We acquired LUAD data from The Cancer Genome Atlas (TCGA) database through the UCSC Xena data portal. Gene expression, clinical, survival, and mutation data from multiple samples were analyzed. Gene enrichment analysis, unsupervised clustering of PLK4-related pathways, and differential gene expression analyses were performed. Additionally, correlations, t-tests, survival analyses, and statistical analyses were performed. (3) Results: PLK4 expression was higher in LUAD tissues than in normal tissues and was associated with poor prognosis for both overall and progression-free survival in LUAD. PLK4 was highly correlated with cell-proliferation-related pathways using Gene Ontology (GO) biological process terms. PLK4 expression and pathways that were highly correlated with PLK4 expression levels were upregulated in patients with LUAD with the TP53 mutation. (4) Conclusions: PLK4 expression affects the survival of patients with LUAD and is a potential therapeutic target for LUAD with TP53 mutations.

Prediction of Visual Acuity in Pathologic Myopia with Myopic Choroidal Neovascularization Treated with Anti-Vascular Endothelial Growth Factor Using a Deep Neural Network Based on Optical Coherence Tomography Images.

Myopic choroidal neovascularization (mCNV) is a common cause of vision loss in patients with pathological myopia. However, predicting the visual prognosis of patients with mCNV remains challenging. This study aimed to develop an artificial intelligence (AI) model to predict visual acuity (VA) in patients with mCNV. This study included 279 patients with mCNV at baseline; patient data were collected, including optical coherence tomography (OCT) images, VA, and demographic information. Two models were developed: one comprising horizontal/vertical OCT images (H/V cuts) and the second comprising 25 volume scan images. The coefficient of determination (R2) and root mean square error (RMSE) were computed to evaluate the performance of the trained network. The models achieved high performance in predicting VA after 1 (R2 = 0.911, RMSE = 0.151), 2 (R2 = 0.894, RMSE = 0.254), and 3 (R2 = 0.891, RMSE = 0.227) years. Using multiple-volume scanning, OCT images enhanced the performance of the models relative to using only H/V cuts. This study proposes AI models to predict VA in patients with mCNV. The models achieved high performance by incorporating the baseline VA, OCT images, and post-injection data. This model could assist in predicting the visual prognosis and evaluating treatment outcomes in patients with mCNV undergoing intravitreal anti-vascular endothelial growth factor therapy.

Reconfigurable neuromorphic computing block through integration of flash synapse arrays and super-steep neurons.

Neuromorphic computing (NC) architecture inspired by biological nervous systems has been actively studied to overcome the limitations of conventional von Neumann architectures. In this work, we propose a reconfigurable NC block using a flash-type synapse array, emerging positive feedback (PF) neuron devices, and CMOS peripheral circuits, and integrate them on the same substrate to experimentally demonstrate the operations of the proposed NC block. Conductance modulation in the flash memory enables the NC block to be easily calibrated for output signals. In addition, the proposed NC block uses a reduced number of devices for analog-to-digital conversions due to the super-steep switching characteristics of the PF neuron device, substantially reducing the area overhead of NC block. Our NC block shows high energy efficiency (37.9 TOPS/W) with high accuracy for CIFAR-10 image classification (91.80%), outperforming prior works. This work shows the high engineering potential of integrating synapses and neurons in terms of system efficiency and high performance.

Fermented Whey Protein Supplementation Improves Muscular Strength, Muscle Parameters, and Physical Performance in Middle-Aged Korean Adults: An 8-Week Double Blind Randomized Controlled Trial.

The present study evaluated the effects of fermented whey protein using kimchi lactic acid bacteria Lactobacillus casei DK211 on skeletal muscle mass, muscle strength, and physical performance in healthy middle-aged males performing regular resistance exercises. Effective protein supplementation and regular exercise are two important factors for improving muscle health. Therefore, in this study, the effects of consuming fermented whey protein twice a day were investigated and compared with that of non-fermented supplementation. Forty-eight males (average age 44.8) were randomly assigned to two groups: Fermented whey protein supplementation (FWPS) and non-fermented whey protein concentration supplementation (WPCS) groups. Each group ingested 37 g of FWPS or WPCS twice a day for eight weeks. Body composition, muscle strength, and physical performance were assessed pre- and post-intervention. Independent t-tests or chi-square tests for the categorical variables were performed for analyzing the observations. FWPS was effective in promoting the physical performance in dynamic balance measurement and muscle health, indicated through the increment in grip strength (left), upper arm circumference, and flat leg circumference from the baseline. However, similar improvements were not observed in the WPCS group. These results imply that whey protein fermented by L. casei DK211 is an effective protein supplement for enhancing muscle health in males performing regular resistance exercises.

Iron Catalysis of C(sp3)-H Azidation Using a Heteroarene Radical Cation Strategy.

The FeIII(phen)3 catalysis of the benzylic C(sp3)-H azidation of indoles has been investigated. The Fe(III) complex can selectively oxidize indoles to form arene radical cations, which are transformed into benzylic C(sp3) radical intermediates. This strategy exhibits a difference in reactivity between N-heteroarenes and benzene, which is difficult to achieve via direct hydrogen abstraction approaches. Various biorelevant azide precursors were constructed, highlighting the utility of this mild first-row transition-metal catalyst system.

Detecting Macular Disease Based on Optical Coherence Tomography Using a Deep Convolutional Network.

Neovascular age-related macular degeneration (nAMD) and central serous chorioretinopathy (CSC) are two of the most common macular diseases. This study proposes a convolutional neural network (CNN)-based deep learning model for classifying the subtypes of nAMD (polypoidal choroidal vasculopathy, retinal angiomatous proliferation, and typical nAMD) and CSC (chronic CSC and acute CSC) and healthy individuals using single spectral-domain optical coherence tomography (SD-OCT) images. The proposed model was trained and tested using 6063 SD-OCT images from 521 patients and 47 healthy participants. We used three well-known CNN architectures (VGG-16, VGG-19, and ResNet) and two customized classification layers. Additionally, transfer learning and mix-up-based data augmentation were applied to improve robustness and accuracy. Our model demonstrated high accuracies of 99.7% and 91.1% in the nAMD and CSC classification and retinopathy (nAMD and CSC) subtype classification, including normal participants, respectively. Furthermore, we performed an external test to compare the classification accuracy with that of eight ophthalmologists, and our model showed the highest accuracy. The region determined to be important for classification by the model was confirmed using gradient-weighted class activation mapping. The model's clinical criteria were similar to that of the ophthalmologists.

Magnetofluoro-Immunosensing Platform Based on Binary Nanoparticle-Decorated Graphene for Detection of Cancer Cell-Derived Exosomes.

Multi-functionalized carbon nanomaterials have attracted interest owing to their excellent synergic properties, such as plasmon resonance energy transfer and surface-enhanced Raman scattering. Particularly, nanoparticle (NP)-decorated graphene (GRP) has been applied in various fields. In this study, silver NP (AgNP)- and magnetic iron oxide NP (IONP)-decorated GRP were prepared and utilized as biosensing platforms. In this case, AgNPs and GRP exhibit plasmonic properties, whereas IONPs exhibit magnetic properties; therefore, this hybrid nanomaterial could function as a magnetoplasmonic substrate for the magnetofluoro-immunosensing (MFI) system. Conversely, exosomes were recently considered high-potential biomarkers for the diagnosis of diseases. However, exosome diagnostic use requires complex isolation and purification methods. Nevertheless, we successfully detected a prostate-cancer-cell-derived exosome (PC-exosome) from non-purified exosomes in a culture media sample using Ag/IO-GRP and dye-tetraspanin antibodies (Ab). First, the anti-prostate-specific antigen was immobilized on the Ag/IO-GRP and it could isolate the PC-exosome from the sample via an external magnetic force. Dye-tetraspanin Ab was added to the sample to induce the sandwich structure. Based on the number of exosomes, the fluorescence intensity from the dye varied and the system exhibited highly sensitive and selective performance. Consequently, these hybrid materials exhibited excellent potential for biosensing platforms.

Carbon Nanotube and Its Derived Nanomaterials Based High Performance Biosensing Platform.

After the COVID-19 pandemic, the development of an accurate diagnosis and monitoring of diseases became a more important issue. In order to fabricate high-performance and sensitive biosensors, many researchers and scientists have used many kinds of nanomaterials such as metal nanoparticles (NPs), metal oxide NPs, quantum dots (QDs), and carbon nanomaterials including graphene and carbon nanotubes (CNTs). Among them, CNTs have been considered important biosensing channel candidates due to their excellent physical properties such as high electrical conductivity, strong mechanical properties, plasmonic properties, and so on. Thus, in this review, CNT-based biosensing systems are introduced and various sensing approaches such as electrochemical, optical, and electrical methods are reported. Moreover, such biosensing platforms showed excellent sensitivity and high selectivity against not only viruses but also virus DNA structures. So, based on the amazing potential of CNTs-based biosensing systems, healthcare and public health can be significantly improved.

Exosome-based drug delivery systems and their therapeutic applications.

In the past few decades, scientists have actively worked on developing effective drug delivery systems (DDSs) as means to control life-threatening diseases and challenging illnesses. In order to develop such DDSs, nanobiotechnological strategies have been introduced, and many nanomaterial-based DDS platforms have been proposed. Among these nanomaterials, DDSs based on exosomes and hybrids of exosomes have been focused upon and developed due to their low toxicity, high bioactivity, and biocompatibility. In this review, we describe the processes involved in drug loading into exosomes and the surface modification of exosomes with treatment agents. Furthermore, we describe the synthesis methods of hybrid exosomes with organic or inorganic nanoparticles. Moreover, we focus on the effective therapeutic applications of exosome-based DDSs against various diseases. In conclusion, we show that exosomes and hybrids of exosomes show excellent drug carrier potential and capacity.

Classifying neovascular age-related macular degeneration with a deep convolutional neural network based on optical coherence tomography images.

Neovascular age-related macular degeneration (nAMD) is among the main causes of visual impairment worldwide. We built a deep learning model to distinguish the subtypes of nAMD using spectral domain optical coherence tomography (SD-OCT) images. Data from SD-OCT images of nAMD (polypoidal choroidal vasculopathy, retinal angiomatous proliferation, and typical nAMD) and normal healthy patients were analyzed using a convolutional neural network (CNN). The model was trained and validated based on 4749 SD-OCT images from 347 patients and 50 healthy controls. To adopt an accurate and robust image classification architecture, we evaluated three well-known CNN structures (VGG-16, VGG-19, and ResNet) and two customized classification layers (fully connected layer with dropout vs. global average pooling). Following the test set performance, the model with the highest classification accuracy was used. Transfer learning and data augmentation were applied to improve the robustness and accuracy of the model. Our proposed model showed an accuracy of 87.4% on the test data (920 images), scoring higher than ten ophthalmologists, for the same data. Additionally, the part that our model judged to be important in classification was confirmed through Grad-CAM images, and consequently, it has a similar judgment criteria to that of ophthalmologists. Thus, we believe that our model can be used as an auxiliary tool in clinical practice.