Validation of Deep Convolutional Neural Network-based algorithm for detection of diabetic retinopathy – Artificial intelligence versus clinician for screening
“…During the training phase, up to 103,815 pictures were taken from both our reference population and the EyePACS database. The reason for using such a large number of retinographies extracted from different populations was to better adjust the algorithm and minimise the overfitting [23].…”
Background: The aim of the present study was to test our deep learning algorithm (DLA) by reading the retinographies. Methods: We tested our DLA built on convolutional neural networks in 14,186 retinographies from our population and 1200 images extracted from MESSIDOR. The retinal images were graded both by the DLA and independently by four retina specialists. Results of the DLA were compared according to accuracy (ACC), sensitivity (S), specificity (SP), positive predictive value (PPV), negative predictive value (NPV), and area under the receiver operating characteristic curve (AUC), distinguishing between identification of any type of DR (any DR) and referable DR (RDR). Results: The results of testing the DLA for identifying any DR in our population were: ACC = 99.75, S = 97.92, SP = 99.91, PPV = 98.92, NPV = 99.82, and AUC = 0.983. When detecting RDR, the results were: ACC = 99.66, S = 96.7, SP = 99.92, PPV = 99.07, NPV = 99.71, and AUC = 0.988. The results of testing the DLA for identifying any DR with MESSIDOR were: ACC = 94.79, S = 97.32, SP = 94.57, PPV = 60.93, NPV = 99.75, and AUC = 0.959. When detecting RDR, the results were: ACC = 98.78, S = 94.64, SP = 99.14, PPV = 90.54, NPV = 99.53, and AUC = 0.968. Conclusions: Our DLA performed well, both in detecting any DR and in classifying those eyes with RDR in a sample of retinographies of type 2 DM patients in our population and the MESSIDOR database.
“…During the training phase, up to 103,815 pictures were taken from both our reference population and the EyePACS database. The reason for using such a large number of retinographies extracted from different populations was to better adjust the algorithm and minimise the overfitting [23].…”
Background: The aim of the present study was to test our deep learning algorithm (DLA) by reading the retinographies. Methods: We tested our DLA built on convolutional neural networks in 14,186 retinographies from our population and 1200 images extracted from MESSIDOR. The retinal images were graded both by the DLA and independently by four retina specialists. Results of the DLA were compared according to accuracy (ACC), sensitivity (S), specificity (SP), positive predictive value (PPV), negative predictive value (NPV), and area under the receiver operating characteristic curve (AUC), distinguishing between identification of any type of DR (any DR) and referable DR (RDR). Results: The results of testing the DLA for identifying any DR in our population were: ACC = 99.75, S = 97.92, SP = 99.91, PPV = 98.92, NPV = 99.82, and AUC = 0.983. When detecting RDR, the results were: ACC = 99.66, S = 96.7, SP = 99.92, PPV = 99.07, NPV = 99.71, and AUC = 0.988. The results of testing the DLA for identifying any DR with MESSIDOR were: ACC = 94.79, S = 97.32, SP = 94.57, PPV = 60.93, NPV = 99.75, and AUC = 0.959. When detecting RDR, the results were: ACC = 98.78, S = 94.64, SP = 99.14, PPV = 90.54, NPV = 99.53, and AUC = 0.968. Conclusions: Our DLA performed well, both in detecting any DR and in classifying those eyes with RDR in a sample of retinographies of type 2 DM patients in our population and the MESSIDOR database.
“…e subthalamic nucleus-(STN-) deep brain stimulation (DBS) treatment is to improve the neurological function of the patient through high-frequency electrical stimulation. In previous studies, the efficacy of STN-DBS has been confirmed [8]. Magnetic resonance imaging (MRI) is a frequently used clinical examination method.…”
This study was to explore the effect of subthalamic nucleus- (STN-) deep brain stimulation (DBS) on the neuropsychiatric function of Parkinson’s disease (PD) patients using the magnetic resonance imaging (MRI) image analysis technology and the artificial intelligence (AI) algorithm. In this study, 40 PD patients admitted to our hospital from August 2018 to March 2020 were selected as the research objects, and they were divided into a control group and an observation group according to the random number table method, with 20 cases in each group. The patients in the control group were given oral treatment with levodopa tablets; and patients in the observation group were treated with STN-DBS + levodopa tablets. In patients, MRI examinations were performed before and after the treatment, and the image optimization processing algorithm under AI was adopted to process the images. The MRI imaging results of the two groups of patients were observed, analyzed, and compared before and after treatment; and the sports, cognition, and mental states of the two groups of patients were analyzed. It was believed that the MRI image before using the AI algorithm was blurry, and the image was clear after the noise reduction optimization process, which was convenient for observation. The data analysis revealed that the signal-to-noise ratio (SNR) after denoising (32.41) and structural similarity (SSIM) (0.79) had been improved. The results of the study suggested that the space occupation and bleeding symptoms of the two groups of patients were reduced after treatment, and those in the observation group were better than those of the control group; the incidences of dyskinesia and motor symptom fluctuations in the observation group were 5% and 0%, respectively, which were lower than those in the control group (35% and 25%, respectively). After treatment, the Unified Parkinson’s Disease Rating Scale (UPDRS) score of the two groups of patients decreased, and it was lower in the observation group than in the control group; and the Montreal Cognitive Assessment (MoCA) and Mini-Mental State Scale (MMSE) scores increased, and those in the observation group were higher in contrast to those in the control group (all
P
<
0.05
). STN-DBS was beneficial to improve the clinical symptoms of patients and delay the progress of the disease, and MRI based on AI algorithms can effectively observe the changes in the neuropsychiatric function of patients, which was conducive to further clinical diagnosis and treatment.
“…4 and 5 , it is clearly visible that a lot of research work on the New Fundus Algorithms seemed to converge into the sole use of convolutional neural networks (Pure CNN) since 2016 (Abràmoff et al 2016 ; Prentašić and Lončarić 2016 ; Gulshan et al 2016 ; Tan et al 2017a , b ; Xu et al 2017 ; Quellec et al 2017 ; Raju et al 2017 ; Ting et al 2017 ; Mansour 2018 ; Brown et al 2018 ; Gao et al 2018 ; Gonzalez-Gonzalo et al 2020 ; Sahlsten et al 2019 ; Liu et al 2019 ; Hemanth et al 2019 ; Sun 2019 ; Zhang et al 2019 ; Li et al 2019a ; Eftekhari et al 2019 ; Bellemo et al 2019 ; Pires et al 2019 ; Qummar et al 2019 ; Zeng et al 2019 ; Mateen et al 2020 ; Wu et al 2020 ; Shaban et al 2020 ; Pao et al 2020 ; Torre et al 2020 ; Shah et al 2020 ; Zago et al 2020 ; Qiao et al 2020 ; Srivastava and Purwar 2020 ; Shankar et al 2020a ; Samanta et al 2020 ; Xie et al 2020 ; Ayhan et al 2020 ). Particularly since 2019, a tremendous increase was observed in the annual number of entries among the New Fundus Algorithms which adopted CNN as their main model of development (Gao et al 2018 ; Gonzalez-Gonzalo et al 2020 ; Sahlsten et al 2019 ; Liu et al 2019 ; Hemanth et al 2019 ; Sun 2019 ; Zhang et al 2019 ; Li et al 2019a ; Eftekhari et al 2019 ; Bellemo et al 2019 ; Pires et al 2019 ; Qummar et al 2019 ; Zeng et al 2019 ; Mateen et al …”
Section: The Study Of the New Fundus Algorithms By Their Overall Mode...mentioning
confidence: 99%
“…Those articles which promotes pure CNN had also given elaboration on the method of preprocessing. This is then compared against those other algorithms who was not found to perform preprocessing [2016: (Abràmoff et al 2016 ), 2017: (Quellec et al 2017 ; Ting et al 2017 ), 2018: (Brown et al 2018 ), 2019: (Sun 2019 ; Bellemo et al 2019 ), 2020: (Shaban et al 2020 ; Torre et al 2020 ; Shah et al 2020 ; Qiao et al 2020 ; Shankar et al 2020a ; Xie et al 2020 ; Ayhan et al 2020 )], either the articles mention that themselves, or there is nowhere in their article which suggest any preprocessing had been done. Fig.…”
Section: The Study Of the New Fundus Algorithms By Their Overall Mode...mentioning
confidence: 99%
“…1.1) who were observed to perform standalone preprocessing in 2020 (Mateen et al 2020 ; Wu et al 2020 ; Pao et al 2020 ; Zago et al 2020 ; Srivastava and Purwar 2020 ; Samanta et al 2020 ), compared to the year 2019 (Gao et al 2018 ; Gonzalez-Gonzalo et al 2020 ; Sahlsten et al 2019 ; Liu et al 2019 ; Hemanth et al 2019 ; Zhang et al 2019 ; Li et al 2019a ; Eftekhari et al 2019 ; Pires et al 2019 ; Qummar et al 2019 ; Zeng et al 2019 ). Through the observation on the articles in 2020 (Mateen et al 2020 ; Wu et al 2020 ; Shaban et al 2020 ; Pao et al 2020 ; Torre et al 2020 ; Shah et al 2020 ; Zago et al 2020 ; Qiao et al 2020 ; Srivastava and Purwar 2020 ; Shankar et al 2020a ; Samanta et al 2020 ; Xie et al 2020 ; Ayhan et al 2020 ), it is concluded that preprocessing no longer needed due to the newly developed CNN architectures (since 2020 and beyond) which by itself is already capable of tracing even the tiniest difference among the colors or textures, without the needs to be provided with a manually enhanced image.…”
Section: The Study Of the New Fundus Algorithms By Their Overall Mode...mentioning
The exponential increase in the number of diabetics around the world has led to an equally large increase in the number of diabetic retinopathy (DR) cases which is one of the major complications caused by diabetes. Left unattended, DR worsens the vision and would lead to partial or complete blindness. As the number of diabetics continue to increase exponentially in the coming years, the number of qualified ophthalmologists need to increase in tandem in order to meet the demand for screening of the growing number of diabetic patients. This makes it pertinent to develop ways to automate the detection process of DR. A computer aided diagnosis system has the potential to significantly reduce the burden currently placed on the ophthalmologists. Hence, this review paper is presented with the aim of summarizing, classifying, and analyzing all the recent development on automated DR detection using fundus images from 2015 up to this date. Such work offers an unprecedentedly thorough review of all the recent works on DR, which will potentially increase the understanding of all the recent studies on automated DR detection, particularly on those that deploys machine learning algorithms. Firstly, in this paper, a comprehensive state-of-the-art review of the methods that have been introduced in the detection of DR is presented, with a focus on machine learning models such as convolutional neural networks (CNN) and artificial neural networks (ANN) and various hybrid models. Each AI will then be classified according to its type (e.g. CNN, ANN, SVM), its specific task(s) in performing DR detection. In particular, the models that deploy CNN will be further analyzed and classified according to some important properties of the respective CNN architectures of each model. A total of 150 research articles related to the aforementioned areas that were published in the recent 5 years have been utilized in this review to provide a comprehensive overview of the latest developments in the detection of DR.
Supplementary Information
The online version contains supplementary material available at 10.1007/s10462-022-10185-6.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
