scholarly journals Diagnosability of Keratoconus Using Deep Learning With Placido Disk-Based Corneal Topography

2021 ◽  
Vol 8 ◽  
Author(s):  
Kazutaka Kamiya ◽  
Yuji Ayatsuka ◽  
Yudai Kato ◽  
Nobuyuki Shoji ◽  
Yosai Mori ◽  
...  
Keyword(s):  
Deep Learning ◽  
Diagnostic Accuracy ◽  
Sensitivity And Specificity ◽  
Clinical Setting ◽  
Area Under The Curve ◽  
Daily Practice ◽  
Corneal Topography ◽  
Grade 3 ◽  
Healthy Eyes

Purpose: Placido disk-based corneal topography is still most commonly used in daily practice. This study was aimed to evaluate the diagnosability of keratoconus using deep learning of a color-coded map with Placido disk-based corneal topography.Methods: We retrospectively examined 179 keratoconic eyes [Grade 1 (54 eyes), 2 (52 eyes), 3 (23 eyes), and 4 (50 eyes), according to the Amsler-Krumeich classification], and 170 age-matched healthy eyes, with good quality images of corneal topography measured with a Placido disk corneal topographer (TMS-4TM, Tomey). Using deep learning of a color-coded map, we evaluated the diagnostic accuracy, sensitivity, and specificity, for keratoconus screening and staging tests, in these eyes.Results: Deep learning of color-coded maps exhibited an accuracy of 0.966 (sensitivity 0.988, specificity 0.944) in discriminating keratoconus from normal eyes. It also exhibited an accuracy of 0.785 (0.911 for Grade 1, 0.868 for Grade 2, 0.920 for Grade 3, and 0.905 for Grade 4) in classifying the stage. The area under the curve value was 0.997, 0.955, 0.899, 0.888, and 0.943 as Grade 0 (normal) to 4 grading tests, respectively.Conclusions: Deep learning using color-coded maps with conventional corneal topography effectively distinguishes between keratoconus and normal eyes and classifies the grade of the disease, indicating that this will become an aid for enhancing the diagnosis and staging ability of keratoconus in a clinical setting.

scholarly journals Multimodal Primary Open Angle Glaucoma Early Diagnosing Program Based on Clinical Process

2021 ◽  
Author(s):  
Jingyi Ma ◽  
Bin Lv ◽  
Yuanyuan Li ◽  
Pan Fan ◽  
Xu Zhao ◽  
...  
Keyword(s):  
Machine Learning ◽  
Deep Learning ◽  
Primary Open Angle Glaucoma ◽  
Open Angle Glaucoma ◽  
Area Under The Curve ◽  
Clinical Decision ◽  
Open Angle ◽  
Fiber Layer ◽  
Visual Field Examination ◽  
Healthy Eyes

Abstract Background: Glaucoma is one of the leading causes of blinding disease. Early detection can improve patients’ quality of vision. Effectively identifying primary open angle glaucoma (POAG) using structural and functional examination is critical. Computer aided diagnosis of glaucoma requires multimodal data to find an accurate model for early glaucoma diagnosis. Methods: This study collected 87 early POAG eyes, 85 suspected POAG eyes, and 129 healthy eyes from the ophthalmology department at Second Affiliated Hospital of Harbin Medical University. Retinal nerve fiber layer thickness (RNFLt), intraocular pressure (IOP) value, visual field examination parameters and age were obtained. A powerful deep learning network segmented the FP and extracted optic nerve head (ONH) features. Machine learning classifiers (MLCs) were adopted to get the final classification results and compared with the diagnosis results of glaucoma specialists and general non-glaucoma ophthalmologists. Result: The program diagnosing early POAG, suspected POAG, and healthy eyes made overall Area Under the Curve of 0.97. Dice of optic disc and optic cup segmentation is 0.9631, 0.8435 respectively. Accuracy of the program (0.9004) is higher than general ophthalmologists (0.8195). Specificity of the program (0.9635) is higher than glaucoma specialists (0.9366).Conclusions: The program delivers superior results in diagnosing early POAG. This study’s hybrid deep learning-machine learning framework can assist with clinical decision for early POAG effectively.

scholarly journals Comparison of the diagnostic accuracy of cone beam computed tomography and three intraoral radiographic systems in the diagnosis of carious lesions in vitro.

2020 ◽  
Vol 9 (6) ◽  
pp. 466-473
Author(s):  
Jorge A. Beltrán ◽  
◽  
Roberto A. León-Manco ◽  
Maria Eugenia Guerrero ◽  
◽  
...  
Keyword(s):  
Computed Tomography ◽  
Dental Caries ◽  
Diagnostic Accuracy ◽  
Sensitivity And Specificity ◽  
Cone Beam Computed Tomography ◽  
Area Under The Curve ◽  
Cone Beam ◽  
Occlusal Surfaces ◽  
Caries Lesions

Objective: The objective of the study was to compare the diagnostic accuracy of cone beam computed tomography and three intraoral radiographic systems in the detection of in vitro caries lesions. Material and Methods: One hundred teeth (46 molars and 54 premolars) were evaluated, including 176 proximal surfaces and 90 occlusal surfaces, with or without dental caries lesions. Digital images of all teeth were obtained using specific intraoral radiographs, VistaScan DürrDental®phosphor-plate radiography, XIOS XG Sirona® digital sensor radiography, and CBCT I-CATTM. Observers evaluated the images for the detection of caries lesions. The teeth were clinically sectioned and stereomicroscopy served as a validation tool. The relationship of sensitivity and specificity between all systems was determined through the ROC curve using Az values. Results: The values of the area under the curve (Az) selected for the CBCT I-CATTM system were 0.89 (0.84-0.93), for conventional radiography 0.71 (0.66-0.76), digital sensor radiography 0.74 (0.70-0.78) and digital radiography with phosphor-plates 0.73 (0.69-0.77). Statistically significant differences were found between the CBCT I-CATTM system and intraoral radiographic systems (p<0.01). The sensitivity and specificity values for the CBCT I-CATTM were 0.84 and 0.93 respectively. Conclusion: CBCT has a high sensitivity and specificity compared to intraoral radiographic systems for the diagnosis of dental caries lesions in vitro.

scholarly journals Validation of ICMR Neurocognitive Toolbox for Dementia in the Linguistically Diverse Context of India

2021 ◽  
Vol 12 ◽  
Author(s):  
Mansi Verma ◽  
Manjari Tripathi ◽  
Ashima Nehra ◽  
Avanthi Paplikar ◽  
Feba Varghese ◽  
...  
Keyword(s):  
Diagnostic Accuracy ◽  
Sensitivity And Specificity ◽  
Linguistic Diversity ◽  
Neuropsychological Test ◽  
Area Under The Curve ◽  
Cognitive Screening ◽  
Indian Languages ◽  
Middle Income ◽  
Significant Difference ◽  
Diagnosis Of Dementia

Objectives: The growing prevalence of dementia, especially in low- and middle-income countries (LMICs), has raised the need for a unified cognitive screening tool that can aid its early detection. The linguistically and educationally diverse population in India contributes to challenges in diagnosis. The present study aimed to assess the validity and diagnostic accuracy of the Indian Council of Medical Research-Neurocognitive Toolbox (ICMR-NCTB), a comprehensive neuropsychological test battery adapted in five languages, for the diagnosis of dementia.Methods: A multidisciplinary group of experts developed the ICMR-NCTB based on reviewing the existing tools and incorporation of culturally appropriate modifications. The finalized tests of the major cognitive domains of attention, executive functions, memory, language, and visuospatial skills were then adapted and translated into five Indian languages: Hindi, Bengali, Telugu, Kannada, and Malayalam. Three hundred fifty-four participants were recruited, including 222 controls and 132 dementia patients. The sensitivity and specificity of the adapted tests were established for the diagnosis of dementia.Results: A significant difference in the mean (median) performance scores between healthy controls and patients with dementia was observed on all tests of ICMR-NCTB. The area under the curve for majority of the tests included in the ICMR-NCTB ranged from 0.73 to 1.00, and the sensitivity and specificity of the ICMR-NCTB tests ranged from 70 to 100% and 70.7 to 100%, respectively, to identify dementia across all five languages.Conclusions: The ICMR-NCTB is a valid instrument to diagnose dementia across five Indian languages, with good diagnostic accuracy. The toolbox was effective in overcoming the challenge of linguistic diversity. The study has wide implications to address the problem of a high disease burden and low diagnostic rate of dementia in LMICs like India.

scholarly journals Alpha-Defensin versus Leukocyte Esterase in Periprosthetic Joint Infection: An Updated Meta-Analysis

2020 ◽  
Vol 2020 ◽  
pp. 1-14
Author(s):  
Zhizhuo Li ◽  
Qingyu Zhang ◽  
Lijun Shi ◽  
Fuqiang Gao ◽  
Wei Sun ◽  
...  
Keyword(s):  
Diagnostic Accuracy ◽  
Sensitivity And Specificity ◽  
Periprosthetic Joint Infection ◽  
Data Extraction ◽  
Meta Analysis ◽  
Joint Infection ◽  
Area Under The Curve ◽  
Cochrane Library ◽  
Systematic Research ◽  
High Diagnostic Accuracy

Periprosthetic joint infection (PJI) is a devastating complication after arthroplasty. Prompt establishment of an infection diagnosis is critical but can be very challenging at present. In order to evaluate the diagnostic accuracy of alpha-defensin or leukocyte esterase for PJI, we performed systematic research in PubMed, Embase, and Cochrane Library to retrieve relevant studies. Data extraction and quality assessment were performed by two reviewers independently. A total of thirty-one eligible studies were finally included in the quantitative analysis. The pooled sensitivity and specificity of alpha-defensin (21 studies) for the diagnosis of PJI were 0.89 (95% confidence interval (CI), 0.83 to 0.93) and 0.96 (95% CI, 0.95 to 0.97), respectively. The value of the pooled diagnostic odds ratios (DOR) of alpha-defensin for PJI was 209.14 (95% CI, 97.31 to 449.50), and the area under the curve (AUC) was 0.98 (95% CI, 0.96 to 0.99). The pooled sensitivity and specificity of leukocyte esterase (17 studies) for the diagnosis of PJI were 0.90 (95% CI, 0.84 to 0.95) and 0.96 (95% CI, 0.93 to 0.97), respectively. The value of the DOR of leukocyte esterase for PJI was 203.23 (95% CI, 96.14 to 429.61), and the AUC was 0.98 (95% CI, 0.96 to 0.99). Based on the results of our meta-analysis, we can conclude that alpha-defensin and leukocyte esterase are valuable synovial fluid markers for identifying PJI with comparable high diagnostic accuracy.

scholarly journals Pleural Fluid Soluble Triggering Receptor Expressed on Myeloid Cells-1 in Complicated and Uncomplicated Parapneumonic Pleural Effusions

2011 ◽  
Vol 2011 ◽  
pp. 1-6
Author(s):  
Adel Salah Bediwy ◽  
Mohamed Gamal A. Elkholy ◽  
Mohamed Mohamed Elbedewy ◽  
Mohamed A. Hasanein
Keyword(s):  
Pleural Effusion ◽  
Diagnostic Accuracy ◽  
Pleural Fluid ◽  
Sensitivity And Specificity ◽  
Myeloid Cells ◽  
Area Under The Curve ◽  
Pleural Effusions ◽  
Cutoff Value ◽  
High Concentrations ◽  
Early Diagnose

Background. Soluble triggering receptor expressed on myeoid cells-1 (sTREM-1) has recently been found to be high in infected pleural fluid (PF). Objectives. Diagnostic accuracy of PF sTREM-1 for differentiating uncomplicated parapneumonic effusions (UPPEs) from complicated parapneumonic effusions (CPPEs) was evaluated prospectively. Methods. Serum and PF sTREM-1 were measured for 68 patients with parapneumonic and transudative pleural effusion. Results. PF (but not serum) sTREM-1 concentrations were significantly higher in CPPE than in UPPE. Serum and PF sTREM-1 levels were higher in parapneumonic than in transudative groups. PF sTREM-1 had a sensitivity of 85.19% and a specificity of 83.33% at cutoff value of 250.5 pg/mL for differentiating CPPE and UPPE with area under the curve (AUC) of 0.9336. After excluding purulent CPPE cases, sensitivity and specificity became 90.48% and 83.33%, respectively (at the same cutoff value) with AUC of 0.9444. Conclusion. High concentrations of PF sTREM-1 (above 250.5 pg/mL) help to early diagnose and differentiate CPPE from UPPE.

scholarly journals Using Slit-Lamp Images for Deep Learning–Based Identification of Bacterial and Fungal Keratitis

2021 ◽  
Author(s):  
Ning Hung ◽  
Eugene Yu-Chuan Kang ◽  
Andy Guan-Yu Shih ◽  
Chi-Hung Lin ◽  
Ming‐Tse Kuo ◽  
...  
Keyword(s):  
Deep Learning ◽  
Diagnostic Accuracy ◽  
Learning Algorithm ◽  
Model Performance ◽  
Area Under The Curve ◽  
Fungal Keratitis ◽  
Classification Model ◽  
Mapping Technique ◽  
Slit Lamp ◽  
Deep Learning Algorithm

In this study, we aimed to develop a deep learning model for identifying bacterial keratitis (BK) and fungal keratitis (FK) by using slit-lamp images. We retrospectively collected slit-lamp images of patients with culture-proven microbial keratitis between January 1, 2010, and December 31, 2019, from two medical centers in Taiwan. We constructed a deep learning algorithm, consisting of a segmentation model for cropping cornea images and a classification model that applies convolutional neural networks to differentiate between FK and BK. The model performance was evaluated and presented as the area under the curve (AUC) of the receiver operating characteristic curves. A gradient-weighted class activation mapping technique was used to plot the heatmap of the model. By using 1330 images from 580 patients, the deep learning algorithm achieved an average diagnostic accuracy of 80.00%. The diagnostic accuracy for BK ranged from 79.59% to 95.91% and that for FK ranged from 26.31% to 63.15%. DenseNet169 showed the best model performance, with an AUC of 0.78 for both BK and FK. The heat maps revealed that the model was able to identify the corneal infiltrations. The model showed better diagnostic accuracy than the previously reported diagnostic performance of both general ophthalmologists and corneal specialists.

scholarly journals King-Devick Sensitivity and Specificity to Concussion in Collegiate Athletes

2021 ◽  
Author(s):  
Rachel K. Le ◽  
Justus D. Ortega ◽  
Sara P. D. Chrisman ◽  
Anthony P. Kontos ◽  
Thomas A. Buckley ◽  
...  
Keyword(s):  
Diagnostic Accuracy ◽  
Sensitivity And Specificity ◽  
Area Under The Curve ◽  
Collegiate Athletes ◽  
Return To Play ◽  
Cutoff Score ◽  
Confounding Factors ◽  
Post Injury ◽  
Cross Sectional ◽  
Cutoff Scores

Context: The King-Devick (K-D) is used to identify oculomotor impairment following concussion. However, the diagnostic accuracy of the K-D over time has not been evaluated. Objective: (a) Examine the sensitivity and specificity of the K-D test at 0–6 hours of injury, 24–48 hours, asymptomatic, return-to-play, and 6-months following concussion and (b) compare outcomes for differentiating athletes with a concussion from non-concussed across confounding factors (sex, age, contact level, school year, learning disorder, ADHD, concussion history, migraine history, administration mode). Design: Retrospective, cross-sectional design. Setting: Multisite institutions within the Concussion Assessment, Research, and Education (CARE) Consortium. Patients or Other Participants: 1239 total collegiate athletes without a concussion (age=20.31±1.18, male=52.2%) were compared to 320 athletes with a concussion (age=19.80±1.41, male=51.3%). Main Outcome Measure(s): We calculated K-D time difference (sec) by subtracting baseline from the most recent time. Receiver operator characteristics (ROC) and area under the curve (AUC) analyses were used to determine the diagnostic accuracy across timepoints. We identified cutoff scores and corresponding specificity at 80% and 70% sensitivity levels. We repeated ROC with AUC outcomes by confounding factors. Results: King-Devick predicted positive results at 0-6 hours (AUC=0.724, p&lt;0.001), 24-48 hours (AUC=0.701, p&lt;0.001), return-to-play (AUC=0.640, P&lt;0.001), and 6-months (AUC=0.615, P&lt;0.001), but not at asymptomatic (AUC=0.513, P=0.497). The 0–6 and 24–48-hour timepoints yielded an 80% sensitivity cutoff score of −2.6 and −3.2 seconds (faster) respectively, but 46% and 41% specificity. The K-D test had significantly better AUC when administered on an iPad (AUC=0.800, 95%CI:0.747,0.854) compared to the spiral card system (AUC=0.646, 95%CI:0.600,0.692; p&lt;0.001). Conclusions: The K-D test has the greatest diagnostic accuracy at 0–6 and 24–48 hours of concussion, but declines across subsequent post-injury timepoints. AUCs did not significantly differentiate between groups for confounding factors. Our negative cutoff scores indicate that practice effects contribute to improved performance, requiring athletes to outperform their baseline.

scholarly journals Transforming Retinal Photographs to Entropy Images in Deep Learning to Improve Automated Detection for Diabetic Retinopathy

2018 ◽  
Vol 2018 ◽  
pp. 1-6 ◽  
Author(s):  
Gen-Min Lin ◽  
Mei-Juan Chen ◽  
Chia-Hung Yeh ◽  
Yu-Yang Lin ◽  
Heng-Yu Kuo ◽  
...  
Keyword(s):  
Diabetic Retinopathy ◽  
Deep Learning ◽  
Sensitivity And Specificity ◽  
Block Size ◽  
Detection Accuracy ◽  
Data Sets ◽  
Feature Maps ◽  
Data Set ◽  
Fundus Photographs ◽  
Grade 3

Entropy images, representing the complexity of original fundus photographs, may strengthen the contrast between diabetic retinopathy (DR) lesions and unaffected areas. The aim of this study is to compare the detection performance for severe DR between original fundus photographs and entropy images by deep learning. A sample of 21,123 interpretable fundus photographs obtained from a publicly available data set was expanded to 33,000 images by rotating and flipping. All photographs were transformed into entropy images using block size 9 and downsized to a standard resolution of 100 × 100 pixels. The stages of DR are classified into 5 grades based on the International Clinical Diabetic Retinopathy Disease Severity Scale: Grade 0 (no DR), Grade 1 (mild nonproliferative DR), Grade 2 (moderate nonproliferative DR), Grade 3 (severe nonproliferative DR), and Grade 4 (proliferative DR). Of these 33,000 photographs, 30,000 images were randomly selected as the training set, and the remaining 3,000 images were used as the testing set. Both the original fundus photographs and the entropy images were used as the inputs of convolutional neural network (CNN), and the results of detecting referable DR (Grades 2–4) as the outputs from the two data sets were compared. The detection accuracy, sensitivity, and specificity of using the original fundus photographs data set were 81.80%, 68.36%, 89.87%, respectively, for the entropy images data set, and the figures significantly increased to 86.10%, 73.24%, and 93.81%, respectively (all p values <0.001). The entropy image quantifies the amount of information in the fundus photograph and efficiently accelerates the generating of feature maps in the CNN. The research results draw the conclusion that transformed entropy imaging of fundus photographs can increase the machinery detection accuracy, sensitivity, and specificity of referable DR for the deep learning-based system.

Is inmunotherapy a safe treatment in bladder cancer? A retrospective study in daily practice.

2019 ◽  
Vol 37 (15_suppl) ◽  
pp. e14083-e14083
Author(s):  
Natalia Vidal Casinello ◽  
Carlos Aguado De La Rosa ◽  
Javier Puente ◽  
Jose-Luis Gonzalez-Larriba
Keyword(s):  
Bladder Cancer ◽  
Clinical Setting ◽  
Side Effect ◽  
Checkpoint Inhibitors ◽  
Performance Status ◽  
Daily Practice ◽  
University Hospital ◽  
Common Side Effect ◽  
Grade 3 ◽  
Few Data

e14083 Background: Several checkpoint inhibitors antibodies have been approved to treat bladder cancer, and many trials are still ongoing. However, few data about their efficacy and safety outside of a clinical trial have been published. Here we report our experience in managing immune-related toxicity in our daily practice. Methods: We performed a retrospective analysis of 23 patients (pts) with metastatic bladder cancer treated with checkpoint inhibitors in Clínico San Carlos University Hospital from April 2016 to September 2018. The aim of this study is to evaluate the safety of checkpoint inhibitors in daily practice. Results: Median age was 75 years (range 53-92) and 75% were men. The performance status at the beginning of immunotherapy treatment was 0 in 4 pts (17.4%), 1 in 16 pts (69.6%) and 2 in 3 pts (13%). Thirteen pts (56.5%) received the treatment as first line, 6 (26.1%) as second line and 4 (17.4%) as third line and beyond. 18 pts (78.3%) received immunotherapy as monotherapy (15 pts Atezolizumab and 2 pts Pembrolizumab), 3 pts (13%) received an immunotherapy combination with Durvalumab plus Tremelimumab, and 2 pts (8.7%) received chemotherapy plus immunotherapy (Carboplatin plus Avelumab or Atezolizumab). Of the analyzed patients, 36% did not develop any immune related adverse event (irAE). However, 64% did present some side effect. The most common side effect was asthenia (12.5%), cutaneous (8.3%), endocrine (5.6%), gastrointestinal (4.2%) and hepatic (2.8%). Only16.6% of the events required some kind of treatment (9.7% required steroids). Three patients had grade 3 or 4 toxicity, 1 patient on Atezolizumab presented with grade 3 diarrhea and grade 3 hepatic toxicity, another patient on Atezolizumab presented two heart attacks after the two infusions and eventually died, lastly a patient on Durvalumab + Tremelimuab suffered grade 3 hypophysitis. Conclusions: Incidence and therapeutic management of irAEs occurring in our clinical setting closely resembles that reported in retrospective series even though they were mostly elderly patients and many of them were heavily pretreated. Increasing our experience and knowledge in the clinical setting will improve and optimize the use of these drugs.

scholarly journals The Clinical Application of the Deep Learning Technique for Predicting Trigger Origins in Patients With Paroxysmal Atrial Fibrillation With Catheter Ablation

2020 ◽  
Vol 13 (11) ◽  
Author(s):  
Chih-Min Liu ◽  
Shih-Lin Chang ◽  
Hung-Hsun Chen ◽  
Wei-Shiang Chen ◽  
Yenn-Jiang Lin ◽  
...  
Keyword(s):  
Computed Tomography ◽  
Atrial Fibrillation ◽  
Deep Learning ◽  
Catheter Ablation ◽  
Prediction Model ◽  
Pulmonary Vein ◽  
Sensitivity And Specificity ◽  
Paroxysmal Atrial Fibrillation ◽  
Area Under The Curve ◽  
Accuracy Of Prediction

Background: Non–pulmonary vein (NPV) trigger has been reported as an important predictor of recurrence post–atrial fibrillation ablation. Elimination of NPV triggers can reduce the recurrence of postablation atrial fibrillation. Deep learning was applied to preablation pulmonary vein computed tomography geometric slices to create a prediction model for NPV triggers in patients with paroxysmal atrial fibrillation. Methods: We retrospectively analyzed 521 patients with paroxysmal atrial fibrillation who underwent catheter ablation of paroxysmal atrial fibrillation. Among them, pulmonary vein computed tomography geometric slices from 358 patients with nonrecurrent atrial fibrillation (1–3 mm interspace per slice, 20–200 slices for each patient, ranging from the upper border of the left atrium to the bottom of the heart, for a total of 23 683 images of slices) were used in the deep learning process, the ResNet34 of the neural network, to create the prediction model of the NPV trigger. There were 298 (83.2%) patients with only pulmonary vein triggers and 60 (16.8%) patients with NPV triggers±pulmonary vein triggers. The patients were randomly assigned to either training, validation, or test groups, and their data were allocated according to those sets. The image datasets were split into training (n=17 340), validation (n=3491), and testing (n=2852) groups, which had completely independent sets of patients. Results: The accuracy of prediction in each pulmonary vein computed tomography image for NPV trigger was up to 82.4±2.0%. The sensitivity and specificity were 64.3±5.4% and 88.4±1.9%, respectively. For each patient, the accuracy of prediction for a NPV trigger was 88.6±2.3%. The sensitivity and specificity were 75.0±5.8% and 95.7±1.8%, respectively. The area under the curve for each image and patient were 0.82±0.01 and 0.88±0.07, respectively. Conclusions: The deep learning model using preablation pulmonary vein computed tomography can be applied to predict the trigger origins in patients with paroxysmal atrial fibrillation receiving catheter ablation. The application of this model may identify patients with a high risk of NPV trigger before ablation.

Sign in / Sign up

Export Citation Format

Share Document