A New Approach for Detecting Fundus Lesions Using Image Processing and Deep Neural Network Architecture Based on YOLO Model

A deep learning approach to hard exudates detection and disorganization of retinal inner layers identification on OCT images

The purpose of the study was to detect Hard Exudates (HE) and classify Disorganization of Retinal Inner Layers (DRIL) implementing a Deep Learning (DL) system on optical coherence tomography (OCT) images of eyes with diabetic macular edema (DME). We collected a dataset composed of 442 OCT images on which we annotated 6847 HE and the presence of DRIL. A complex operational pipeline was defined to implement data cleaning and image transformations, and train two DL models. The state-of-the-art neural network architectures (Yolov7, ConvNeXt, RegNetX) and advanced techniques were exploited to aggregate the results (Ensemble learning, Edge detection) and obtain a final model. The DL approach reached good performance in detecting HE and classifying DRIL. Regarding HE detection the model got an AP@0.5 score equal to 34.4% with Precision of 48.7% and Recall of 43.1%; while for DRIL classification an Accuracy of 91.1% with Sensitivity and Specificity both of 91.1% and AUC and AUPR values equal to 91% were obtained. The P-value was lower than 0.05 and the Kappa coefficient was 0.82. The DL models proved to be able to identify HE and DRIL in eyes with DME with a very good accuracy and all the metrics calculated confirmed the system performance. Our DL approach demonstrated to be a good candidate as a supporting tool for ophthalmologists in OCT images analysis.

Enhanced object detection in pediatric bronchoscopy images using YOLO-based algorithms with CBAM attention mechanism

Background and Objective

Bronchoscopy is a widely used diagnostic and therapeutic procedure for respiratory disorders such as infections and tumors. However, visualizing the bronchial tubes and lungs can be challenging due to the presence of various objects, such as mucus, blood, and foreign bodies. Accurately identifying the anatomical location of the bronchi can be quite challenging, especially for medical professionals who are new to the field. Deep learning-based object detection algorithms can assist doctors in analyzing images or videos of the bronchial tubes to identify key features such as the epiglottis, vocal cord, and right basal bronchus. This study aims to improve the accuracy of object detection in bronchoscopy images by integrating a YOLO-based algorithm with a CBAM attention mechanism.

Methods

The CBAM attention module is implemented in the YOLO-V7 and YOLO-V8 object detection models to improve their object identification and classification capabilities in bronchoscopy images. Various YOLO-based object detection algorithms, such as YOLO-V5, YOLO-V7, and YOLO-V8 are compared on this dataset. Experiments are conducted to evaluate the performance of the proposed method and different algorithms.

Results

The proposed method significantly improves the accuracy and reliability of object detection for bronchoscopy images. This approach demonstrates the potential benefits of incorporating an attention mechanism in medical imaging and the benefits of utilizing object detection algorithms in bronchoscopy. In the experiments, the YOLO-V8-based model achieved a mean Average Precision (mAP) of 87.09% on the given dataset with an Intersection over Union (IoU) threshold of 0.5. After incorporating the Convolutional Block Attention Module (CBAM) into the YOLO-V8 architecture, the proposed method achieved a significantly enhanced m A P 0.5 and m A P 0.5 : 0.95 of 88.27% and 55.39%, respectively.

Conclusions

Our findings indicate that by incorporating a CBAM attention mechanism with a YOLO-based algorithm, there is a noticeable improvement in object detection performance in bronchoscopy images. This study provides valuable insights into enhancing the performance of attention mechanisms for object detection in medical imaging.

Neighbored-attention U-net (NAU-net) for diabetic retinopathy image segmentation

Background

Diabetic retinopathy-related (DR-related) diseases are posing an increasing threat to eye health as the number of patients with diabetes mellitus that are young increases significantly. The automatic diagnosis of DR-related diseases has benefited from the rapid development of image semantic segmentation and other deep learning technology.

Methods

Inspired by the architecture of U-Net family, a neighbored attention U-Net (NAU-Net) is designed to balance the identification performance and computational cost for DR fundus image segmentation. In the new network, only the neighboring high- and low-dimensional feature maps of the encoder and decoder are fused by using four attention gates. With the help of this improvement, the common target features in the high-dimensional feature maps of encoder are enhanced, and they are also fused with the low-dimensional feature map of decoder. Moreover, this network fuses only neighboring layers and does not include the inner layers commonly used in U-Net++. Consequently, the proposed network incurs a better identification performance with a lower computational cost.

Results

The experimental results of three open datasets of DR fundus images, including DRIVE, HRF, and CHASEDB, indicate that the NAU-Net outperforms FCN, SegNet, attention U-Net, and U-Net++ in terms of Dice score, IoU, accuracy, and precision, while its computation cost is between attention U-Net and U-Net++.

Conclusion

The proposed NAU-Net exhibits better performance at a relatively low computational cost and provides an efficient novel approach for DR fundus image segmentation and a new automatic tool for DR-related eye disease diagnosis.

One-Stage Detection without Segmentation for Multi-Type Coronary Lesions in Angiography Images Using Deep Learning

It is rare to use the one-stage model without segmentation for the automatic detection of coronary lesions. This study sequentially enrolled 200 patients with significant stenoses and occlusions of the right coronary and categorized their angiography images into two angle views: The CRA (cranial) view of 98 patients with 2453 images and the LAO (left anterior oblique) view of 176 patients with 3338 images. Randomization was performed at the patient level to the training set and test set using a 7:3 ratio. YOLOv5 was adopted as the key model for direct detection. Four types of lesions were studied: Local Stenosis (LS), Diffuse Stenosis (DS), Bifurcation Stenosis (BS), and Chronic Total Occlusion (CTO). At the image level, the precision, recall, mAP@0.1, and mAP@0.5 predicted by the model were 0.64, 0.68, 0.66, and 0.49 in the CRA view and 0.68, 0.73, 0.70, and 0.56 in the LAO view, respectively. At the patient level, the precision, recall, and F1scores predicted by the model were 0.52, 0.91, and 0.65 in the CRA view and 0.50, 0.94, and 0.64 in the LAO view, respectively. YOLOv5 performed the best for lesions of CTO and LS at both the image level and the patient level. In conclusion, the one-stage model without segmentation as YOLOv5 is feasible to be used in automatic coronary lesion detection, with the most suitable types of lesions as LS and CTO.

Automatic Detection of Microaneurysms in Fundus Images Using an Ensemble-Based Segmentation Method

In this study, a novel method for automatic microaneurysm detection in color fundus images is presented. The proposed method is based on three main steps: (1) image breakdown to smaller image patches, (2) inference to segmentation models, and (3) reconstruction of the predicted segmentation map from output patches. The proposed segmentation method is based on an ensemble of three individual deep networks, such as U-Net, ResNet34-UNet and UNet++. The performance evaluation is based on the calculation of the Dice score and IoU values. The ensemble-based model achieved higher Dice score (0.95) and IoU (0.91) values compared to other network architectures. The proposed ensemble-based model demonstrates the high practical application potential for detection of early-stage diabetic retinopathy in color fundus images.

An Efficient and Intelligent Detection Method for Fabric Defects based on Improved YOLOv5

Limited by computing resources of embedded devices, there are problems in the field of fabric defect detection, including small defect size, extremely unbalanced aspect ratio of defect size, and slow detection speed. To address these problems, a sliding window multihead self-attention mechanism is proposed for the detection of small targets, and the Swin Transformer module is introduced to replace the main module in the original YOLOv5 algorithm. First, to reduce the distance between several scales, the weighted bidirectional feature network is employed on embedded devices. In addition, it is helpful to improve the perception of small-target faults by incorporating a detection layer to achieve four-scale detection. At last, to improve the learning of positive sample instances and lower the missed detection rate, the generalized focal loss function is finally implemented on YOLOv5. Experimental results show that the accuracy of the improved algorithm on the fabric dataset reaches 85.6%, and the mAP is increased by 4.2% to 76.5%, which meets the requirements for real-time detection on embedded devices.

AIE-YOLO: Auxiliary Information Enhanced YOLO for Small Object Detection

Small object detection is one of the key challenges in the current computer vision field due to the low amount of information carried and the information loss caused by feature extraction. You Only Look Once v5 (YOLOv5) adopts the Path Aggregation Network to alleviate the problem of information loss, but it cannot restore the information that has been lost. To this end, an auxiliary information-enhanced YOLO is proposed to improve the sensitivity and detection performance of YOLOv5 to small objects. Firstly, a context enhancement module containing a receptive field size of 21×21 is proposed, which captures the global and local information of the image by fusing multi-scale receptive fields, and introduces an attention branch to enhance the expressive ability of key features and suppress background noise. To further enhance the feature expression ability of small objects, we introduce the high- and low-frequency information decomposed by wavelet transform into PANet to participate in multi-scale feature fusion, so as to solve the problem that the features of small objects gradually disappear after multiple downsampling and pooling operations. Experiments on the challenging dataset Tsinghua-Tencent 100 K show that the mean average precision of the proposed model is 9.5% higher than that of the original YOLOv5 while maintaining the real-time speed, which is better than the mainstream object detection models.

Special Issue "Computer Aided Diagnosis Sensors"

Sensors used to diagnose, monitor or treat diseases in the medical domain are known as medical sensors [...].