Diabetic Retinopathy Diagnosis Based on RA-EfficientNet

Yi, Sanli; Yang, Xue-lian; Wang, Tianwei; She, Fu-Rong; Xiong, Xin; He, Jianfeng

doi:10.3390/app112211035

Cited by 28 publications

(15 citation statements)

References 21 publications

(27 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Yi et al [24] performed an EfficientNet-based DR classification task with several pre-processing steps to overcome the problem of imbalanced data on APTOS the network called RA-EfficientNet was proposed to extract additional valid image features. This network adds a residual attention component to EfficientNet to retrieve more features and address the issue of subtle changes among lesions.…”

Section: Related Workmentioning

confidence: 99%

LuNet-LightGBM: An Effective Hybrid Approach for Lesion Segmentation and DR Grading

Bapatla¹,

Harikiran²

2023

Computer Systems Science and Engineering

View full text Add to dashboard Cite

Diabetes problems can lead to an eye disease called Diabetic Retinopathy (DR), which permanently damages the blood vessels in the retina. If not treated early, DR becomes a significant reason for blindness. To identify the DR and determine the stages, medical tests are very labor-intensive, expensive, and timeconsuming. To address the issue, a hybrid deep and machine learning techniquebased autonomous diagnostic system is provided in this paper. Our proposal is based on lesion segmentation of the fundus images based on the LuNet network. Then a Refined Attention Pyramid Network (RAPNet) is used for extracting global and local features. To increase the performance of the classifier, the unique features are selected from the extracted feature set using Aquila Optimizer (AO) algorithm. Finally, the LightGBM model is applied to classify the input image based on the severity. Several investigations have been done to analyze the performance of the proposed framework on three publically available datasets (MESSIDOR, APTOS, and IDRiD) using several performance metrics such as accuracy, precision, recall, and f1-score. The proposed classifier achieves 99.29%, 99.35%, and 99.31% accuracy for these three datasets respectively. The outcomes of the experiments demonstrate that the suggested technique is effective for disease identification and reliable DR grading.

show abstract

Section: Related Workmentioning

confidence: 99%

LuNet-LightGBM: An Effective Hybrid Approach for Lesion Segmentation and DR Grading

Bapatla¹,

Harikiran²

2023

Computer Systems Science and Engineering

View full text Add to dashboard Cite

show abstract

“…In the second experiment, the same issue was examined, but the neighborhood during approximation was changed. Details of the experiment are provided below: → Experiment I-five groups with different lengths of word embedding vector (3,10,20, 50 and 100) with 4-neighborhood during approximation were tested, the parameter of window width was changed within {2,3,5}. → Experiment II-five groups with different lengths of word embedding vector (3,10,20, 50, and 100) with 5-neighborhood during approximation were tested, the parameter of window width was changed within {2,3,5}.…”

Section: Conducted Experimentsmentioning

confidence: 99%

“…Due to the fact that the main perceptual channel of humans is the vision system, in medicine, we mainly use image data. Therefore, we encounter a number of vision systems that are used in medicine: systems using medical image contrast enhancement [1], image processing systems of various modalities supporting the therapy of various diseases [2] or systems supporting the early detection of diseases using categorization algorithms as screening methods [3]. In medicine, we also independently deal with phenomena modeling systems using computer methods [4].…”

Section: Introductionmentioning

confidence: 99%

Determining the Intensity of Basic Emotions among People Suffering from Anorexia Nervosa Based on Free Statements about Their Body

et al. 2022

View full text Add to dashboard Cite

Objective: This study sought to address one of the challenges of psychiatry-computer aided diagnosis and therapy of anorexia nervosa. The goal of the paper is to present a method of determining the intensity of five emotions (happiness, sadness, anxiety, anger and disgust) in medical notes, which was then used to analyze the feelings of people suffering from anorexia nervosa. In total, 96 notes were researched (46 from people suffering from anorexia and 52 from healthy people). Method: The developed solution allows a comprehensive assessment of the intensity of five feelings (happiness, sadness, anxiety, anger and disgust) occurring in text notes. This method implements Nencki Affective Word List dictionary extension, in which the original version has a limited vocabulary. The method was tested on a group of patients suffering from anorexia nervosa and a control group (healthy people without an eating disorder). Of the analyzed medical, only 8% of the words are in the original dictionary. Results: As a result of the study, two emotional profiles were obtained: one pattern for a healthy person and one for a person suffering from anorexia nervosa. Comparing the average emotional intensity in profiles of a healthy person and person with a disorder, a higher value of happiness intensity is noticeable in the profile of a healthy person than in the profile of a person with an illness. The opposite situation occurs with other emotions (sadness, anxiety, disgust, anger); they reach higher values in the case of the profile of a person suffering from anorexia nervosa. Discussion: The presented method can be used when observing the patient’s progress during applied therapy. It allows us to state whether the chosen method has a positive effect on the mental state of the patient, and if his emotional profile is similar to the emotional profile of a healthy person. The method can also be used during first diagnosis visit.

show abstract

“…With the continuous development of computer vision, some computer-assisted retinal image research has also achieved great success [23]- [31]. For example, Chen et al [23] propose a general deep learning model for DR classification, which uses a 2-stage training method to solve the overfitting problem.…”

Section: A Retinal Image Classificationmentioning

confidence: 99%

“…Gangwar et al [30] tackle the challenge of automated diabetic retinopathy diagnosis and suggest an unique deep learning hybrid solution. Yi et al [31] suggest the network known as RA-EfficientNet, in which a residual attention (RA) block is added to EfficientNet in order to extract additional features and address the issue of minute changes between lesions.…”

Section: A Retinal Image Classificationmentioning

confidence: 99%

Graph Adversarial Transfer Learning for Diabetic Retinopathy Classification

Jing-bo

Wang

et al. 2022

IEEE Access

View full text Add to dashboard Cite

Diabetic retinopathy (DR) is an essential factor that has caused vision loss and even blindness in middle-aged and older adults. A system that can automatically perform DR diagnosis can help ophthalmologists save a lot of tedious work, such as DR grading or lesion detection. At the same time, patients can find their diseases earlier and perform the correct treatment. However, most of the existing methods require many DR annotations to train the model, and the DR data will vary to different degrees due to various shooting tools. The above problems lead to the inefficient use of existing data in the experiment, limiting actual deployment. To alleviate this problem, we propose a novel Graph Adversarial Transfer Learning (GATL) for DR diagnosis in a deep model through transfer learning, including intradomain alignment and inter-domain alignment. The proposed GATL enjoys several merits. First, our GATL adopts the self-supervised training to save the annotating cost in the target domain thus this domain adaptation method can significantly reduce annotation cost compared to the supervised approaches. Second, we introduce the graph neural network to extract potential features between unknown samples. Third, to enhance the robustness of the model, we use adversarial training to perform both inter-domain and intradomain alignment to further improve the model's classification accuracy. GATL achieved 94.3%, 97.5%, and 91.1% in accuracy, sensitivity, and specificity in the APTOS dataset and 92.7%, 95.7%, and 89.7% in the EyePACS dataset, respectively. Extensive experimental results on two challenging benchmarks, including APTOS 2019 and EyePACS, demonstrate that the proposed GATL performs favorably against baseline DR classification methods.

show abstract

Diabetic Retinopathy Diagnosis Based on RA-EfficientNet

Cited by 28 publications

References 21 publications

LuNet-LightGBM: An Effective Hybrid Approach for Lesion Segmentation and DR Grading

LuNet-LightGBM: An Effective Hybrid Approach for Lesion Segmentation and DR Grading

Determining the Intensity of Basic Emotions among People Suffering from Anorexia Nervosa Based on Free Statements about Their Body

Graph Adversarial Transfer Learning for Diabetic Retinopathy Classification

Contact Info

Product

Resources

About