Weakly supervised detection of central serous chorioretinopathy based on local binary patterns and discrete wavelet transform

Optimized deep CNN for detection and classification of diabetic retinopathy and diabetic macular edema

Diabetic Retinopathy (DR) and Diabetic Macular Edema (DME) are vision related complications prominently found in diabetic patients. The early identification of DR/DME grades facilitates the devising of an appropriate treatment plan, which ultimately prevents the probability of visual impairment in more than 90% of diabetic patients. Thereby, an automatic DR/DME grade detection approach is proposed in this work by utilizing image processing. In this work, the retinal fundus image provided as input is pre-processed using Discrete Wavelet Transform (DWT) with the aim of enhancing its visual quality. The precise detection of DR/DME is supported further with the application of suitable Artificial Neural Network (ANN) based segmentation technique. The segmented images are subsequently subjected to feature extraction using Adaptive Gabor Filter (AGF) and the feature selection using Random Forest (RF) technique. The former has excellent retinal vein recognition capability, while the latter has exceptional generalization capability. The RF approach also assists with the improvement of classification accuracy of Deep Convolutional Neural Network (CNN) classifier. Moreover, Chicken Swarm Algorithm (CSA) is used for further enhancing the classifier performance by optimizing the weights of both convolution and fully connected layer. The entire approach is validated for its accuracy in determination of grades of DR/DME using MATLAB software. The proposed DR/DME grade detection approach displays an excellent accuracy of 97.91%.

Application and visualization study of an intelligence-assisted classification model for common eye diseases using B-mode ultrasound images

Aim

Conventional approaches to diagnosing common eye diseases using B-mode ultrasonography are labor-intensive and time-consuming, must requiring expert intervention for accuracy. This study aims to address these challenges by proposing an intelligence-assisted analysis five-classification model for diagnosing common eye diseases using B-mode ultrasound images.

Methods

This research utilizes 2064 B-mode ultrasound images of the eye to train a novel model integrating artificial intelligence technology.

Results

The ConvNeXt-L model achieved outstanding performance with an accuracy rate of 84.3% and a Kappa value of 80.3%. Across five classifications (no obvious abnormality, vitreous opacity, posterior vitreous detachment, retinal detachment, and choroidal detachment), the model demonstrated sensitivity values of 93.2%, 67.6%, 86.1%, 89.4%, and 81.4%, respectively, and specificity values ranging from 94.6% to 98.1%. F1 scores ranged from 71% to 92%, while AUC values ranged from 89.7% to 97.8%.

Conclusion

Among various models compared, the ConvNeXt-L model exhibited superior performance. It effectively categorizes and visualizes pathological changes, providing essential assisted information for ophthalmologists and enhancing diagnostic accuracy and efficiency.

Research on an artificial intelligence-based myopic maculopathy grading method using EfficientNet

PURPOSE

We aimed to develop an artificial intelligence-based myopic maculopathy grading method using EfficientNet to overcome the delayed grading and diagnosis of different myopic maculopathy degrees.

METHODS

The cooperative hospital provided 4642 healthy and myopic maculopathy color fundus photographs, comprising the four degrees of myopic maculopathy and healthy fundi. The myopic maculopathy grading models were trained using EfficientNet-B0 to EfficientNet-B7 models. The diagnostic results were compared with those of the VGG16 and ResNet50 classification models. The leading evaluation indicators were sensitivity, specificity, F1 score, area under the receiver operating characteristic (ROC) curve area under curve (AUC), 95% confidence interval, kappa value, and accuracy. The ROC curves of the ten grading models were also compared.

RESULTS

We used 1199 color fundus photographs to evaluate the myopic maculopathy grading models. The size of the EfficientNet-B0 myopic maculopathy grading model was 15.6 MB, and it had the highest kappa value (88.32%) and accuracy (83.58%). The model's sensitivities to diagnose tessellated fundus (TF), diffuse chorioretinal atrophy (DCA), patchy chorioretinal atrophy (PCA), and macular atrophy (MA) were 96.86%, 75.98%, 64.67%, and 88.75%, respectively. The specificity was above 93%, and the AUCs were 0.992, 0.960, 0.964, and 0.989, respectively.

CONCLUSION

The EfficientNet models were used to design grading diagnostic models for myopic maculopathy. Based on the collected fundus images, the models could diagnose a healthy fundus and four types of myopic maculopathy. The models might help ophthalmologists to make preliminary diagnoses of different degrees of myopic maculopathy.

A new convolutional neural network based on combination of circlets and wavelets for macular OCT classification

Artificial intelligence (AI) algorithms, encompassing machine learning and deep learning, can assist ophthalmologists in early detection of various ocular abnormalities through the analysis of retinal optical coherence tomography (OCT) images. Despite considerable progress in these algorithms, several limitations persist in medical imaging fields, where a lack of data is a common issue. Accordingly, specific image processing techniques, such as time-frequency transforms, can be employed in conjunction with AI algorithms to enhance diagnostic accuracy. This research investigates the influence of non-data-adaptive time-frequency transforms, specifically X-lets, on the classification of OCT B-scans. For this purpose, each B-scan was transformed using every considered X-let individually, and all the sub-bands were utilized as the input for a designed 2D Convolutional Neural Network (CNN) to extract optimal features, which were subsequently fed to the classifiers. Evaluating per-class accuracy shows that the use of the 2D Discrete Wavelet Transform (2D-DWT) yields superior outcomes for normal cases, whereas the circlet transform outperforms other X-lets for abnormal cases characterized by circles in their retinal structure (due to the accumulation of fluid). As a result, we propose a novel transform named CircWave by concatenating all sub-bands from the 2D-DWT and the circlet transform. The objective is to enhance the per-class accuracy of both normal and abnormal cases simultaneously. Our findings show that classification results based on the CircWave transform outperform those derived from original images or any individual transform. Furthermore, Grad-CAM class activation visualization for B-scans reconstructed from CircWave sub-bands highlights a greater emphasis on circular formations in abnormal cases and straight lines in normal cases, in contrast to the focus on irrelevant regions in original B-scans. To assess the generalizability of our method, we applied it to another dataset obtained from a different imaging system. We achieved promising accuracies of 94.5% and 90% for the first and second datasets, respectively, which are comparable with results from previous studies. The proposed CNN based on CircWave sub-bands (i.e. CircWaveNet) not only produces superior outcomes but also offers more interpretable results with a heightened focus on features crucial for ophthalmologists.

The differential value of radiomics based on traditional T1-weighted sequences in newborns with hyperbilirubinemia

BACKGROUND

On the basis of visual-dependent reading method, radiological recognition and assessment of neonatal hyperbilirubinemia (NH) or acute bilirubin encephalopathy (ABE) on conventional magnetic resonance imaging (MRI) sequences are challenging. Prior studies had shown that radiomics was possible to characterize ABE-induced intensity and morphological changes on MRI sequences, and it has emerged as a desirable and promising future in quantitative and objective MRI data extraction. To investigate the utility of radiomics based on T1-weighted sequences for identifying neonatal ABE in patients with hyperbilirubinemia and differentiating between those with NH and the normal controls.

METHODS

A total of 88 patients with NH were enrolled, including 50 patients with ABE and 38 ABE-negative individuals, and 70 age-matched normal neonates were included as controls. All participants were divided into training and validation cohorts in a 7:3 ratio. Radiomics features extracted from the basal ganglia of T1-weighted sequences on magnetic resonance imaging were evaluated and selected to set up the prediction model using the K-nearest neighbour-based bagging algorithm. A receiver operating characteristic curve was plotted to assess the differentiating performance of the radiomics-based model.

RESULTS

Four of 744 radiomics features were selected for the diagnostic model of ABE. The radiomics model yielded an area under the curve (AUC) of 0.81 and 0.82 in the training and test cohorts, with accuracy, precision, sensitivity, and specificity of 0.82, 0.80, 0.91, and 0.69 and 0.78, 0.8, 0.8, and 0.75, respectively. Six radiomics features were selected in this model to distinguish those with NH from the normal controls. The AUC for the training cohort was 0.97, with an accuracy of 0.92, a precision of 0.92, a sensitivity of 0.93, and a specificity of 0.90. The performance of the radiomics model was confirmed by testing the test cohort, and the AUC, accuracy, precision, sensitivity, and specificity were 0.97, 0.92, 0.96, 0.89, and 0.95, respectively.

CONCLUSIONS

The proposed radiomics model based on traditional TI-weighted sequences may be used effectively for identifying ABE and even differentiating patients with NH from the normal controls, which can provide microcosmic information beyond experience-dependent vision and potentially assist in clinical diagnosis and treatment.

Screening of Common Retinal Diseases Using Six-Category Models Based on EfficientNet

Purpose

A six-category model of common retinal diseases is proposed to help primary medical institutions in the preliminary screening of the five common retinal diseases.

Methods

A total of 2,400 fundus images of normal and five common retinal diseases were provided by a cooperative hospital. Two six-category deep learning models of common retinal diseases based on the EfficientNet-B4 and ResNet50 models were trained. The results from the six-category models in this study and the results from a five-category model in our previous study based on ResNet50 were compared. A total of 1,315 fundus images were used to test the models, the clinical diagnosis results and the diagnosis results of the two six-category models were compared. The main evaluation indicators were sensitivity, specificity, F1-score, area under the curve (AUC), 95% confidence interval, kappa and accuracy, and the receiver operator characteristic curves of the two six-category models were compared in the study.

Results

The diagnostic accuracy rate of EfficientNet-B4 model was 95.59%, the kappa value was 94.61%, and there was high diagnostic consistency. The AUC of the normal diagnosis and the five retinal diseases were all above 0.95. The sensitivity, specificity, and F1-score for the diagnosis of normal fundus images were 100, 99.9, and 99.83%, respectively. The specificity and F1-score for RVO diagnosis were 95.68, 98.61, and 93.09%, respectively. The sensitivity, specificity, and F1-score for high myopia diagnosis were 96.1, 99.6, and 97.37%, respectively. The sensitivity, specificity, and F1-score for glaucoma diagnosis were 97.62, 99.07, and 94.62%, respectively. The sensitivity, specificity, and F1-score for DR diagnosis were 90.76, 99.16, and 93.3%, respectively. The sensitivity, specificity, and F1-score for MD diagnosis were 92.27, 98.5, and 91.51%, respectively.

Conclusion

The EfficientNet-B4 model was used to design a six-category model of common retinal diseases. It can be used to diagnose the normal fundus and five common retinal diseases based on fundus images. It can help primary doctors in the screening for common retinal diseases, and give suitable suggestions and recommendations. Timely referral can improve the efficiency of diagnosis of eye diseases in rural areas and avoid delaying treatment.

A CT-based radiomics nomogram for predicting early recurrence in patients with high-grade serous ovarian cancer

PURPOSE

To develop and validate a radiomics nomogram for predicting early recurrence in high-grade serous ovarian cancer (HGSOC) patients.

MATERIALS AND METHODS

From May 2008 to December 2019, 256 eligible HGSOC patients were enrolled and divided into training (n = 179) and test cohorts (n = 77) in a 7:3 ratio. A radiomics signature (Radscore) was selected by using recursive feature elimination based on a support vector machine (SVM-RFE) and building a radiomics model for recurrence prediction. Independent clinical risk factors were generated by univariable and multivariable Cox regression analyses. A combined model was developed based on the Radscore and independent clinical risk factors and presented as a radiomics nomogram. Its performance was assessed by AUC, Kaplan-Meier survival analysis and decision curve analysis.

RESULTS

Seven radiomics features were selected. The radiomics model yielded AUCs of 0.715 (95% CI: 0.640, 0.790) and 0.717 (95% CI: 0.600, 0.834) in the training and test cohorts, respectively. The clinical model (FIGO stage and residual disease) yielded AUCs of 0.632 and 0.691 in the training and test cohorts, respectively. The combined model demonstrated AUCs of 0.749 (95% CI: 0.678, 0.821) and 0.769 (95% CI: 0.662, 0.877) in the training and test cohorts, respectively. In the combined model, PFS was significantly shorter in the high-risk group than in the low-risk group (P < 0.0001).

CONCLUSIONS

The radiomics nomogram performed well for early individualized recurrence prediction in patients with HGSOC and can also be used to differentiate high-risk patients from low-risk patients.

Attitudes of medical workers in China toward artificial intelligence in ophthalmology: a comparative survey

Background

In the development of artificial intelligence in ophthalmology, the ophthalmic AI-related recognition issues are prominent, but there is a lack of research into people’s familiarity with and their attitudes toward ophthalmic AI. This survey aims to assess medical workers’ and other professional technicians’ familiarity with, attitudes toward, and concerns about AI in ophthalmology.

Methods

This is a cross-sectional study design study. An electronic questionnaire was designed through the app Questionnaire Star, and was sent to respondents through WeChat, China’s version of Facebook or WhatsApp. The participation was voluntary and anonymous. The questionnaire consisted of four parts, namely the respondents’ background, their basic understanding of AI, their attitudes toward AI, and their concerns about AI. A total of 562 respondents were counted, with 562 valid questionnaires returned. The results of the questionnaires are displayed in an Excel 2003 form.

Results

There were 291 medical workers and 271 other professional technicians completed the questionnaire. About 1/3 of the respondents understood AI and ophthalmic AI. The percentages of people who understood ophthalmic AI among medical workers and other professional technicians were about 42.6 % and 15.6 %, respectively. About 66.0 % of the respondents thought that AI in ophthalmology would partly replace doctors, about 59.07 % having a relatively high acceptance level of ophthalmic AI. Meanwhile, among those with AI in ophthalmology application experiences (30.6 %), above 70 % of respondents held a full acceptance attitude toward AI in ophthalmology. The respondents expressed medical ethics concerns about AI in ophthalmology. And among the respondents who understood AI in ophthalmology, almost all the people said that there was a need to increase the study of medical ethics issues in the ophthalmic AI field.

Conclusions

The survey results revealed that the medical workers had a higher understanding level of AI in ophthalmology than other professional technicians, making it necessary to popularize ophthalmic AI education among other professional technicians. Most of the respondents did not have any experience in ophthalmic AI but generally had a relatively high acceptance level of AI in ophthalmology, and there was a need to strengthen research into medical ethics issues.

The Development and Validation of a CT-Based Radiomics Nomogram to Preoperatively Predict Lymph Node Metastasis in High-Grade Serous Ovarian Cancer

Purpose

To develop and validate a radiomics model for predicting preoperative lymph node (LN) metastasis in high-grade serous ovarian cancer (HGSOC).

Materials and Methods

From May 2008 to January 2018, a total of 256 eligible HGSOC patients who underwent tumor resection and LN dissection were divided into a training cohort (n=179) and a test cohort (n=77) in a 7:3 ratio. A Radiomics Model was developed based on a training cohort of 179 patients. A radiomics signature (defined as the Radscore) was selected by using the random forest method. Logistics regression was used as the classifier for modeling. An Integrated Model that incorporated the Radscore and CT_reported LN status (CT_LN_report) was developed and presented as a radiomics nomogram. Its performance was determined by the area under the curve (AUC), calibration, and decision curve. The radiomics nomogram was internally tested in an independent test cohort (n=77) and a CT-LN-report negative subgroup (n=179) using the formula derived from the training cohort.

Results

The AUC value of the CT_LN_report was 0.688 (95% CI: 0.626, 0.759) in the training cohort and 0.717 (95% CI: 0.630, 0.804) in the test cohort. The Radiomics Model yielded an AUC of 0.767 (95% CI: 0.696, 0.837) in the training cohort and 0.753 (95% CI: 0.640, 0.866) in the test. The radiomics nomogram demonstrated favorable calibration and discrimination in the training cohort (AUC=0.821), test cohort (AUC=0.843), and CT-LN-report negative subgroup (AUC=0.82), outperforming the Radiomics Model and CT_LN_report alone.

Conclusions

The radiomics nomogram derived from portal phase CT images performed well in predicting LN metastasis in HGSOC and could be recommended as a new, convenient, and non-invasive method to aid in clinical decision-making.