A novel lncRNA associated with the prognosis of patients with colorectal cancer resists apoptosis through the LYN/BCL-2 pathway.

PURPOSE: We found a novel lncRNA named lncAC138150.2 related to the overall survival and staging of patients with colorectal cancer (CRC) by bioinformatic analysis using data from the Cancer Genome Atlas (TCGA), and the study aimed to elucidate the function of lncAC138150.2 and underlying mechanisms. METHODS: Target molecules were knocked down by transfection with antisense oligonucleotides (ASOs), siRNAs, or lentiviruses and overexpressed by transfection with plasmids. The function of lncAC138150.2 was determined using histological, cytological, and molecular biology methods. The underlying mechanism of lncAC138150.2 function was investigated using RNA-seq, bioinformatics analysis, and molecular biology methods. RESULTS: The expression of lncAC138150.2 was increased in colorectal tissues compared with paired normal tissues. The lncAC138150.2 knockdown increased apoptosis but did not change the cell proliferation, cell cycle distribution, or cell migration ability of CRC cells, while lncAC138150.2 overexpression decreased CRC apoptosis. lncAC138150.2 was mainly located in the cell nucleus, and each lncAC138150.2 transcript knockdown increased CRC apoptosis. BCL-2 pathway was significantly altered in apoptosis induced by lncAC138150.2 knockdown, which was alleviated by BAX knockdown. The expression of LYN was significantly decreased with lncAC138150.2 knockdown, LYN knockdown increased CRC apoptosis, and its overexpression completely alleviated CRC apoptosis induced by lncAC138150.2 knockdown. CONCLUSION: lncAC138150.2 significantly inhibited CRC apoptosis and affected the prognosis of patients with CRC, through the LYN/BCL-2 pathway.

Effects of electromagnetic pulses, exosomes inhibition and their coaction on A549 cells.

Mounting literature indicates that electromagnetic pulses (EMP) is the promising modality to treat cancers with advantages such as noninvasiveness and few side-effects, but its appropriate parameters and underlying mechanisms such as its influence on tumor-derived exosomes (TDEs) are largely unknown. This study aimed to elucidate effects of EMP, exosome inhibition and their coaction on A549 lung adenocarcinoma cells. A549 cells were randomly divided into control group, GW4869 group treated by 20 µM GW4869, vehicle group treated by dimethyl sulfoxide, EMP group treated by EMP exposure, and EMPG group treated by EMP exposure combined with 20 µM GW4869. After EMP exposure, cell proliferation was determined by CCK8 assay, cell cycle and apoptosis was detected by flow cytometry, and cell migration was determined by transwell assay. The results showed that EMP or exosomes inhibition did not affect cell proliferation, cell cycle, apoptosis and cell migration (p > 0.05), but cell migration in EMPG group was significantly decreased compared with vehicle group (p < 0.05). We concluded that under the experimental condition, EMP or GW4869 alone had no effects on behaviors of A549 cells, but their coaction could effectively inhibit the migration of A549 cells.

Quantitative proteomics reveals effects of environmental radiofrequency electromagnetic fields on embryonic neural stem cells.

The effects of environmental radiofrequency electromagnetic fields (RF-EMF) on embryonic neural stem cells have not been determined, particularly at the proteomic level. This study aims to elucidate the effects of environmental levels of RF-EMF radiation on embryonic neural stem cells. Neuroectodermal stem cells (NE-4C cells) were randomly divided into a sham group and an RF group, which were sham-exposed and continuously exposed to a 1950 MHz RF-EMF at 2 W/kg for 48 h. After exposure, cell proliferation was determined by a Cell Counting Kit-8 (CCK8) assay, the cell cycle distribution and apoptosis were measured by flow cytometry, protein abundance was detected by liquid chromatography-tandem mass spectrometry (LC-MS/MS), and mRNA expression was evaluated by quantitative reverse transcription polymerase chain reaction (qRT-PCR). We did not detect differences in cell proliferation, cell cycle distribution, and apoptosis between the two groups. However, we detected differences in the abundance of 23 proteins between the two groups, and some of these differences were consistent with alterations in transcript levels determined by qRT-PCR (P < 0.05). A bioinformatics analysis indicated that the differentially regulated proteins were mainly enriched in 'localization' in the cellular process category; however, no significant pathway alterations in NE-4C cells were detected. We conclude that under the experimental conditions, low-level RF-EMF exposure was not neurotoxic but could induce minor changes in the abundance of some proteins involved in neurodevelopment or brain function.

Effects of pulsed electromagnetic fields on tumor cell viability: a meta-analysis of in vitro randomized controlled experiments.

Malignant tumor treatment remains a big challenge till now, and expanding literature indicated that pulsed electromagnetic fields (PEMF) is promising in tumor treatment with the advantage of safety and being economical, but it is still controversial on whether PEMF could affect the tumor cell viability. Therefore, we conducted the meta-analysis to evaluate effects of PEMF on tumor cell viability. The PubMed, EMBASE, Web of Science, and Cochrane Library databases were searched for studies published up to February 2021. Studies on the direct effects of PEMF on tumor cell viability, determined using colorimetric analysis, were included. Two authors extracted the data and completed the quality assessment. A meta-analysis was performed to calculate the absorbance values and 95% confidence intervals (CIs) using random-effects models. Seven studies, including 32 randomized controlled experiments, were analyzed. Compared with the control group, tumor cell viability in the PEMF exposure group was obviously lower (SMD, -0.67; 95% CI: -1.12 to -0.22). The subgroup meta-analysis results showed that PEMF significantly reduced epithelial cancer cell viability (SMD, -0.58; 95% CI: -0.92 to -0.23) but had no influence on stromal tumor cell viability (SMD, -0.93; 95% CI: -0.21 to 0.15). Our study demonstrated that PEMF could inhibit tumor cell proliferation to some extent, but the risk of bias and high heterogeneity (I2 > 75%) weakened the strength of the conclusions drawn from the analysis.

Cathodal tDCS exerts neuroprotective effect in rat brain after acute ischemic stroke.

BACKGROUND: Transcranial direct current stimulation (tDCS) is a non-invasive brain modulation technique that has been proved to exert beneficial effects in the acute phase of stroke. To explore the underlying mechanism, we investigated the neuroprotective effects of cathodal tDCS on brain injury caused by middle cerebral artery occlusion (MCAO). RESULTS: We established the MCAO model and sham MCAO model with an epicranial electrode implanted adult male Sprague-Dawley rats, and then they were randomly divided into four groups (MCAO + tDCS, MCAO + sham tDCS (Sham), Control + tDCS and Control + Sham group). In this study, the severity degree of neurological deficit, the morphology of brain damage, the apoptosis, the level of neuron-specific enolase and inflammatory factors, the activation of glial cells was detected. The results showed that cathodal tDCS significantly improved the level of neurological deficit and the brain morphology, reduced the brain damage area and apoptotic index, and increased the number of Nissl body in MCAO rats, compared with MCAO + Sham group. Meanwhile, the high level of NSE, inflammatory factors, Caspase 3 and Bax/Bcl2 ratio in MCAO rats was reduced by cathodal tDCS. Additionally, cathodal tDCS inhibited the activation of astrocyte and microglia induced by MCAO. No difference was found in two Control groups. CONCLUSION: Our results suggested that cathodal tDCS could accelerate the recovery of neurologic deficit and brain damage caused by MCAO. The inhibition of neuroinflammation and apoptosis resulted from cathodal tDCS may be involved in the neuroprotective process.

A safety study of 500 µA cathodal transcranial direct current stimulation in rat.

BACKGROUND: Transcranial direct current stimulation (tDCS) is a noninvasive neural control technology that has become a research hotspot. To facilitate further research of tDCS, the biosafety of 500 µA cathodal tDCS, a controversial parameter in rats was evaluated. RESULTS: 24 animals were randomly divided into two groups: a cathodal tDCS group (tDCS, n = 12) and control group (control, n = 12). Animals in the tDCS group received 5 consecutive days of cathodal tDCS (500 µA, 15 min, once per day) followed by a tDCS-free interval of 2 days and 5 additional days of stimulation, totally two treatments of tDCS for a total of 10 days. Computational 3D rat model was adopted to calculate the current density distributions in brain during tDCS treatment. Essential brain functions including motor function and learning and memory ability were evaluated. Additionally, to estimate the neurotoxicity of tDCS, the brain morphology, neurotransmitter levels and cerebral temperature were investigated. Our results showed that the current density inside the brain was less than 20 A/m2 during tDCS treatment in computational model. tDCS did not affect motor functions and learning and memory ability after tDCS treatment. In addition, no significant differences were found for the tDCS group in hematology, serum biochemical markers or the morphology of major organs. Moreover, tDCS treatment had no effect on the brain morphology, neural structures, neurotransmitter levels or cerebral temperature. CONCLUSION: 500 µA cathodal tDCS as performed in the present study was safe for rodents.

Effects of 1.8 GHz radiofrequency field on microstructure and bone metabolism of femur in mice.

To investigate the effects of 1.8 GHz radiofrequency (RF) field on bone microstructure and metabolism of femur in mice, C57BL/6 mice (male, age 4 weeks) were whole-body exposed or sham exposed to 1.8 GHz RF field. Specific absorption rates of whole body and bone were approximately 2.70 and 1.14 W/kg (6 h/day for 28 days). After exposure, microstructure and morphology of femur were observed by microcomputed tomography (micro-CT), Hematoxylin and Eosin (HE) and Masson staining. Subsequently, bone parameters were calculated directly from the reconstructed images, including structure model index, bone mineral density, trabecular bone volume/total volume, connectivity density, trabecular number, trabecular thickness, and trabecular separation. Biomarkers that reflect bone metabolism, such as serum total alkaline phosphatase (ALP), bone-specific alkaline phosphatase (BALP), and tartrate-resistant acid phosphatase 5b (TRACP-5b), were determined by biochemical assay methods. Micro-CT and histology results showed that there was no significant change in bone microstructure and the above parameters in RF group, compared with sham group. The activity of serum ALP and BALP increased 29.47% and 16.82%, respectively, in RF group, compared with sham group (P < 0.05). In addition, there were no significant differences in the activity of serum TRACP-5b between RF group and sham group. In brief, under present experimental conditions, we did not find support for an effect of 1.8 GHz RF field on bone microstructure; however, it might promote metabolic function of osteoblasts in mice. Bioelectromagnetics. 39:386-393, 2018. © 2018 Wiley Periodicals, Inc.

Effects of electromagnetic pulse exposure on gelatinase of blood-brain barrier in vitro.

The biological effects of electromagnetic pulse (EMP) on the brain have been focused on for years. It was reported that gelatinase played an important role in maintaining brain function through regulating permeability in the blood-brain barrier (BBB). To investigate the effects of EMP on gelatinase of BBB, an in vitro BBB model was established using primary cultured rat brain microvascular endothelial cells (BMVEC), astrocytes and half-contact culture of these cells in a transwell chamber. Cultured supernatant and cells were collected at different time points after exposure to EMP (peak intensity 400 kV/m, rise time 10 ns, pulse width 350 ns, 0.5 pps and 200 pulses). Protein levels of cellular gelatinase MMP-2 and MMP-9, and endogenous inhibitor TIMP-1 and TIMP-2 were detected by Western blot. The activity of gelatinase in culture supernatant was detected by gelatin zymography. It was found that compared with the sham-exposed group, the protein level of MMP-2 was significantly increased at 6 h (p < 0.05), and the protein level of its endogenous inhibitor TIMP-2 did not change after EMP exposure. In addition, the protein levels of MMP-9 and its endogenous inhibitor TIMP-1 did not change after EMP exposure. Gelatin zymography results showed that the activity of MMP-2 in the inner pool and the outer pool of the transwell chamber was significantly increased at 6 h after EMP exposure compared with that of the sham group. These results suggested that EMP exposure could affect the expression and activity of MMP-2 in the BBB model.

Oxidative changes and apoptosis induced by 1800-MHz electromagnetic radiation in NIH/3T3 cells.

To investigate the potential adverse effects of mobile phone radiation, we studied reactive oxygen species (ROS), DNA damage and apoptosis in mouse embryonic fibroblasts (NIH/3T3) after intermittent exposure (5 min on/10 min off, for various durations from 0.5 to 8 h) to an 1800-MHz GSM-talk mode electromagnetic radiation (EMR) at an average specific absorption rate of 2 W/kg. A 2',7'-dichlorofluorescin diacetate fluorescence probe was used to detect intracellular ROS levels, immunofluorescence was used to detect γH2AX foci as a marker for DNA damage, and flow cytometry was used to measure apoptosis. Our results showed a significant increase in intracellular ROS levels after EMR exposure and it reached the highest level at an exposure time of 1 h (p < 0.05) followed by a slight decrease when the exposure continued for as long as 8 h. No significant effect on the number of γH2AX was detected after EMR exposure. The percentage of late-apoptotic cells in the EMR-exposed group was significantly higher than that in the sham-exposed groups (p < 0.05). These results indicate that an 1800-MHz EMR enhances ROS formation and promotes apoptosis in NIH/3T3 cells.

Stellate Ganglion Block Therapy for Complex Regional Pain Syndrome: A Systematic Review and Meta-Analysis.

BACKGROUND: Sympathetic ganglion block (SGB) technique is becoming increasingly prevalent in the treatment of complex regional pain syndromes (CRPS). Given the varied reported effectiveness of these techniques and the heterogeneity of treatment regimens, there is an urgent need for consistent and high-quality evidence on the efficacy and safety of such procedures. OBJECTIVES: This study aimed to compare the efficacy of SGB therapy for CRPS-related pain. STUDY DESIGN: A meta-analysis of randomized controlled trials (RCTs). METHODS: PubMed, EMBASE, Web of Science, CINAHL, US National Institutes of Health Clinical Trials Registry, Google Scholar, and Cochrane Library Databases were systematically searched between January 1967 and April 2023. A meta-analysis of the included RCTs on SGB was conducted to evaluate the effectiveness and risk of bias (ROBs) of SGB. RESULTS: After screening 8523 records, 12 RCTs were included in this meta-analysis. Compared with controls, the visual analog pain score decreased by a weighted mean difference (WMD) of -6.24 mm (95% CI, -11.45, -1.03; P = 0.019) in the random-effects model, and the numerical scale score was reduced by a WMD of -1.17 mm (95% CI, -2.42, 0.08; P = 0.067) in the fixed-effects model, indicating a pain relief. The methodological quality of the included RCTs was high, with an average PEDro score of 7.0 (range: 5-9). LIMITATIONS: The number of included trials was limited. CONCLUSIONS: SGB therapy can reduce pain intensity in patients with CRPS with few adverse events. However, owing to the relatively high heterogeneity of the included RCTs, a larger sample of high-quality RCTs is needed to further confirm this conclusion.

[Effect of long-term power frequency electromagnetic field exposure on proliferation and apoptosis of SRA01/04 cells].

OBJECTIVE: To investigate the effect of long-term power frequency electromagnetic field (50 Hz) exposure on the proliferation and apoptosis of human lens epithelial cells (SRA01/04 cells). METHODS: SRA01/04 cells in the exponential growth phase were exposed or sham-exposed to power frequency electromagnetic field (50 Hz, 2.3 mT) for 2 hours per day, 5 days every week. After 11 weeks of exposure, the cells were collected; the cell morphology was observed under a microscope, the cell viability was measured by MTT assay, the cell cycle and apoptosis were examined by flow cytometry, and the protein expression levels of cyclin D and proliferating cell nuclear antigen (PCNA) were determined by western blot. RESULTS: Compared with the sham-exposed SRA01/04 cells, most exposed cells became rounded and more stereoscopic, and heterochromatin gathered near the nuclear membrane in some exposed cells. The MTT assay showed that the viability of exposed cells was significantly increased compared with that of the sham-exposed cells (P < 0.05). Long-term power frequency electromagnetic field exposure led to significantly increased number of cells in S phase (P < 0.05), and the proliferation index was significantly higher in the exposed cells than in the sham-exposed cells (P < 0.05). There was no significant difference in apoptotic rate between the exposed cells and sham-exposed cells (P > 0.05). The exposed cells had significantly higher protein expression levels of cyclin D and PCNA than the sham-exposed cells (P < 0.05). CONCLUSION: Long-term power frequency electromagnetic field exposure can promote cellular proliferation and change cell cycle in SRA01/04 cells, but it has no marked effect on the apoptosis of SRA01/04 cells.

Regulatory role of exosomes in colorectal cancer progression and potential as biomarkers.

Colorectal cancer (CRC) remains an enormous challenge to human health worldwide. Unfortunately, the mechanism underlying CRC progression is not well understood. Mounting evidence has confirmed that exosomes play a vital role in CRC progression, which has attracted extensive attention among researchers. In addition to acting as messengers between CRC cells, exosomes also participate in the CRC immunomodulatory process and reshape immune function. As stable message carriers and liquid biopsy option under development, exosomes are promising biomarkers in the diagnosis or treatment of CRC. In this review we have described and analyzed the biogenesis and release of exosomes and current research on the role of exosomes in immune regulation and metastasis of CRC. Moreover, we have discussed candidate exosomal molecules as potential biomarkers to diagnose CRC, predict CRC progression, or determine CRC chemoresistance, and described the significance of exosomes in the immunotherapy of CRC. This review provides insight to further understand the role of exosomes in CRC progression and identify valuable biomarkers that facilitate the clinical management of CRC patients.

Single cell molecular alterations reveal target cells and pathways of conditioned fear memory.

OBJECTIVES: Recent evidence indicates that hippocampus is important for conditioned fear memory (CFM). Though few studies consider the roles of various cell types' contribution to such a process, as well as the accompanying transcriptome changes during this process. The purpose of this study was to explore the transcriptional regulatory genes and the targeted cells that are altered by CFM reconsolidation. METHODS: A fear conditioning experiment was established on adult male C57 mice, after day 3 tone-cued CFM reconsolidation test, hippocampus cells were dissociated. Using single cell RNA sequencing (scRNA-seq) technique, alterations of transcriptional genes expression were detected and cell cluster analysis were performed and compared with those in sham group. RESULTS: Seven non-neuronal and eight neuronal cell clusters (including four known neurons and four newly identified neuronal subtypes) has been explored. Among them, CA subtype 1 has characteristic gene markers of Ttr and Ptgds, which is speculated to be the outcome of acute stress and promotes the production of CFM. The results of KEGG pathway enrichment indicate the differences in the expression of certain molecular protein functional subunits in long-term potentiation (LTP) pathway between two types of neurons (DG and CA1) and astrocytes, thus providing a new transcriptional perspective for the role of hippocampus in the CFM reconsolidation. More importantly, the correlation between the reconsolidation of CFM and neurodegenerative diseases-linked genes is substantiated by the results from cell-cell interactions and KEGG pathway enrichment. Further analysis shows that the reconsolidation of CFM inhibits the risk-factor genes App and ApoE in Alzheimer's Disease (AD) and activates the protective gene Lrp1. CONCLUSIONS: This study reports the transcriptional genes expression changes of hippocampal cells driven by CFM, which confirm the involvement of LTP pathway and suggest the possibility of CFM-like behavior in preventing AD. However, the current research is limited to normal C57 mice, and further studies on AD model mice are needed to prove this preliminary conclusion.

Hierarchical deep learning models using transfer learning for disease detection and classification based on small number of medical images.

Deep learning is being employed in disease detection and classification based on medical images for clinical decision making. It typically requires large amounts of labelled data; however, the sample size of such medical image datasets is generally small. This study proposes a novel training framework for building deep learning models of disease detection and classification with small datasets. Our approach is based on a hierarchical classification method where the healthy/disease information from the first model is effectively utilized to build subsequent models for classifying the disease into its sub-types via a transfer learning method. To improve accuracy, multiple input datasets were used, and a stacking ensembled method was employed for final classification. To demonstrate the method's performance, a labelled dataset extracted from volumetric ophthalmic optical coherence tomography data for 156 healthy and 798 glaucoma eyes was used, in which glaucoma eyes were further labelled into four sub-types. The average weighted accuracy and Cohen's kappa for three randomized test datasets were 0.839 and 0.809, respectively. Our approach outperformed the flat classification method by 9.7% using smaller training datasets. The results suggest that the framework can perform accurate classification with a small number of medical images.

Artificial intelligence for classifying uncertain images by humans in determining choroidal vascular running pattern and comparisons with automated classification between artificial intelligence.

PURPOSE: Abnormalities of the running pattern of choroidal vessel have been reported in eyes with pachychoroid diseases. However, it is difficult for clinicians to judge the running pattern with high reproducibility. Thus, the purpose of this study was to compare the degree of concordance of the running pattern of the choroidal vessels between that determined by artificial intelligence (AI) to that determined by experienced clinicians. METHODS: The running pattern of the choroidal vessels in en face images of Haller's layer of 413 normal and pachychoroid diseased eyes was classified as symmetrical or asymmetrical by human raters and by three supervised machine learning models; the support vector machine (SVM), Xception, and random forest models. The data from the human raters were used as the supervised data. The accuracy rates of the human raters and the certainty of AI's answers were compared using confidence scores (CSs). RESULTS: The choroidal vascular running pattern could be determined by each AI model with an area under the curve better than 0.94. The random forest method was able to discriminate with the highest accuracy among the three AIs. In the CS analyses, the percentage of certainty was highest (66.4%) and that of uncertainty was lowest (6.1%) in the agreement group. On the other hand, the rate of uncertainty was highest (27.3%) in the disagreement group. CONCLUSION: AI algorithm can automatically classify with ambiguous criteria the presence or absence of a symmetrical blood vessel running pattern of the choroid. The classification was as good as that of supervised humans in accuracy and reproducibility.

tDCS Accelerates the Rehabilitation of MCAO-Induced Motor Function Deficits via Neurogenesis Modulated by the Notch1 Signaling Pathway.

Background. Ischemic stroke carries a high mortality rate and is a leading cause of severe neurological disability. However, the efficacy of current therapeutic options remains limited. Objective. We aimed to investigate the treatment efficacy of transcranial direct current stimulation (tDCS) in motor function rehabilitation after ischemic stroke and explore the underlying mechanisms. Methods. Male Sprague-Dawley rats with epicranial electrodes were used to establish pathogenetic model through temporary right middle cerebral artery occlusion (MCAO). Subsequently, animals were randomly divided into 4 groups: MCAO + tDCS/sham tDCS, Control + tDCS/sham tDCS. Animals in the groups with tDCS underwent 10 days of cathodal tDCS totally (500 µA, 15 minutes, once a day). During and after tDCS treatment, the motor functions of the animals, ischemic damage area, proliferation and differentiation of neural stem cells (NSCs), and distribution, and protein expression of Notch1 signaling molecules were detected. Results. The rehabilitation of MCAO-induced motor function deficits was dramatically accelerated by tDCS treatment. NSC proliferation in the subventricular zone (SVZ) was significantly increased after MCAO surgery, and tDCS treatment promoted this process. Additionally, NSCs probably migrated from the SVZ to the ischemic striatum and then differentiated into neurons and oligodendrocytes after MCAO surgery, both of which processes were accelerated by tDCS treatment. Finally, tDCS treatment inhibited the activation of Notch1 signaling in NSCs in the ischemic striatum, which may be involved in NSC differentiation in the MCAO model. Conclusion. Our results suggest that tDCS may exert therapeutic efficacy after ischemic stroke in a regenerative medical perspective.

Structural Characterization of Glaucoma Patients with Low Ocular Blood Flow.

Purpose: There is an unclear relationship between ocular blood flow (OBF) and the structural characteristics of the optic nerve head (ONH) in glaucoma, a multifactorial disease. This study used laser speckle flowgraphy (LSFG) to identify low-OBF glaucoma patients and investigated the ONH in these patients. Materials and Methods: In 533 eyes with glaucoma, we determined confounding factors for LSFG-measured OBF (tissue-area mean blur rate: MT) and corrected MT with a linear mixed-effects model (LMM). Structural ONH data (from fundus stereo photography), OCT data, and clinical characteristics were then compared in patients with corrected MT in the upper and lower quartiles using the LMM. Results: Single regression showed significant correlations between MT and age, spherical equivalent (SE), central corneal thickness (CCT), and a weighted count of retinal ganglion cells (wRGC), but not axial length or systemic blood pressure. Gender also significantly influenced MT; MT was corrected for these correlated factors and also glaucoma type with the LMM. The lower-quartile MT group had a significantly larger cup area and cup-disc area ratio and lower temporal quadrant circumpapillary retinal nerve fiber layer thickness (cpRNFLT) and macular ganglion cell complex (GCC) than the upper-quartile group. Conclusions: Low-OBF glaucoma patients were characterized by a larger cup-disc ratio, and higher susceptibility to damage in the temporal disc and the macular area.

Retinal Thickness Analysis in High Myopia based on Medial Axis Transforms.

This paper presents a retinal thickness analysis method from 3D images acquired by optical coherence tomography (OCT). Given OCT images with segmented boundaries of retinal layers, medial axes of the layers are computed by medial axis transforms (MAT), and thickness is evaluated based on Euclidean distance fields. Since the MAT computes the closest points on the boundary of the layer, it can compute more correct thickness than conventional methods that evaluate Y-axis-aligned thickness. Experimental results show that our method can detect thin-parts around distorted regions, or a clue of high myopia. This is useful for early diagnosis of high myopia and other eye diseases.

Deep Learning Classification Models Built with Two-step Transfer Learning for Age Related Macular Degeneration Diagnosis.

The objective of this study was to build deep learning models with optical coherence tomography (OCT) images to classify normal and age related macular degeneration (AMD), AMD with fluid, and AMD without any fluid. In this study, 185 normal OCT images from 49 normal subjects, 535 OCT images of AMD with fluid, and 514 OCT mages of AMD without fluid from 120 AMD eyes as training data, while 49 normal images from 25 normal eyes, 188 AMD OCT images with fluid and 154 AMD images without any fluid from 77 AMD eyes as test data, were enrolled. Data augmentation was applied to increase the number of images to build deep learning models. Totally, two classification models were built in two steps. In the first step, a VGG16 model pre-trained on ImageNet dataset was transfer learned to classify normal and AMD, including AMD with fluid and/or without any fluid. Then, in the second step, the fine-tuned model in the first step was transfer learned again to distinguish the images of AMD with fluid from the ones without any fluid. With the first model, normal and AMD OCT images were classified with 0.999 area under receiver operating characteristic curve (AUC), and 99.2% accuracy. With the second model, AMD with the presence of any fluid, and AMD without fluid were classified with 0.992 AUC, and 95.1% accuracy. Compared with a transfer learned VGG16 model pre-trained on ImageNet dataset, to classify the three categories directly, higher classification performance was achieved with our notable approach. Conclusively, two classification models for AMD clinical practice were built with high classification performance, and these models should help improve the early diagnosis and treatment for AMD.

Glaucoma Diagnosis with Machine Learning Based on Optical Coherence Tomography and Color Fundus Images.

This study aimed to develop a machine learning-based algorithm for glaucoma diagnosis in patients with open-angle glaucoma, based on three-dimensional optical coherence tomography (OCT) data and color fundus images. In this study, 208 glaucomatous and 149 healthy eyes were enrolled, and color fundus images and volumetric OCT data from the optic disc and macular area of these eyes were captured with a spectral-domain OCT (3D OCT-2000, Topcon). Thickness and deviation maps were created with a segmentation algorithm. Transfer learning of convolutional neural network (CNN) was used with the following types of input images: (1) fundus image of optic disc in grayscale format, (2) disc retinal nerve fiber layer (RNFL) thickness map, (3) macular ganglion cell complex (GCC) thickness map, (4) disc RNFL deviation map, and (5) macular GCC deviation map. Data augmentation and dropout were performed to train the CNN. For combining the results from each CNN model, a random forest (RF) was trained to classify the disc fundus images of healthy and glaucomatous eyes using feature vector representation of each input image, removing the second fully connected layer. The area under receiver operating characteristic curve (AUC) of a 10-fold cross validation (CV) was used to evaluate the models. The 10-fold CV AUCs of the CNNs were 0.940 for color fundus images, 0.942 for RNFL thickness maps, 0.944 for macular GCC thickness maps, 0.949 for disc RNFL deviation maps, and 0.952 for macular GCC deviation maps. The RF combining the five separate CNN models improved the 10-fold CV AUC to 0.963. Therefore, the machine learning system described here can accurately differentiate between healthy and glaucomatous subjects based on their extracted images from OCT data and color fundus images. This system should help to improve the diagnostic accuracy in glaucoma.