CNN-based diagnosis models for canine ulcerative keratitis

Abstract

the purpose of this methodological study was to develop a convolutional neural network (cnn), which is a recently developed deep-learning-based image recognition method, to determine corneal ulcer severity in dogs. the cnn model was trained with images for which corneal ulcer severity (normal, superficial, and deep) were previously classified by veterinary ophthalmologists' diagnostic evaluations of corneal photographs from patients who visited the Veterinary Medical teaching Hospital (VMtH) at Konkuk University and 3 different veterinary ophthalmology specialty hospitals in Korea. The original images (depicting normal corneas (36) and corneas with superficial (47) ulcers, deep (47) ulcers), flipped images (total 520), rotated images (total 520), and both flipped and rotated images (total 1,040) were labeled, learned and evaluated with GoogLenet, Resnet, and VGGnet models, and the severity of each corneal ulcer image was determined. to accomplish this task, models based on tensorflow, an opensource software library developed by Google, were used, and the labeled images were converted into TensorFlow record (TFRecord) format. The models were fine-tuned using a CNN model trained on the imagenet dataset and then used to predict severity. Most of the models achieved accuracies of over 90% when classifying superficial and deep corneal ulcers, and ResNet and VGGNet achieved accuracies over 90% for classifying normal corneas, corneas with superficial ulcers, and corneas with deep ulcers. This study proposes a method to effectively determine corneal ulcer severity in dogs by using a CNN and concludes that multiple image classification models can be used in the veterinary field. Deep learning diagnostic tools for image recognition have recently been tested in many medical fields. In the field of imaging diagnostics, the use of such tools have been reported in The Veterinary Journal 1,2. Recent advances in computer hardware technology, such as high performance graphic processing units (GPUs), have permitted the development of deep neural networks (DNNs). Deep learning algorithms are an evolution of neural networks and are currently used in a variety of medical and industrial applications 3,4. Deep learning fundamentally consists of a deep neural network structure with several layers. An artificial neural network based on the backpropagation 5 algorithm was highly anticipated in the 1990s; such a network would utilize logic that corrects the error of each neuron after analyzing the error in the reverse direction at the output side when the error occurs. However, research has stagnated because learning in artificial neural network models becomes more difficult as the number of layers increases. Since the mid-2000s, artificial neural networks have been improved with respect to learning methods: huge amounts of data have been made available, and the hardware environment has been improved, leading to remarkable performance improvements. In addition, the dataset overfitting problem, which was a persistent issue with artificial n...

Prediction Model of Aryl Hydrocarbon Receptor Activation by a Novel QSAR Approach, DeepSnap–Deep Learning

“…Differences in prediction performances in terms of the parameter loss (Val), Acc (Val), BAC, F, AUC, Acc (Test), and MCC were analyzed using the Mann-Whitney U test [79][80][81]. For each of the 10 angles (38,38,38), (42,42,42), (50,50,50), (55,55,55), (65,65,65), (85, 85, 85), (105, 105, 105), (176, 176, 176), (300, 300, 300), and (360, 360, 360) in the two datasets Tra/Val/Test = 1:1:1 and 2:2:2, seven evaluation indicators of loss (Val), Acc (Val), BAC, F, AUC, Acc (Test), and MCC are represented as box plots. Significant differences are calculated for each angle.…”

“…The 3D chemical models were captured automatically as snapshots with user-defined angle increments with respect to the x-, y-, and z-axes. In this study, 10 angle increments were used: (38,38,38), (42,42,42), (50,50,50), (55,55,55), (65,65,65), (85, 85, 85), (105, 105, 105), (176, 176, 176), (300, 300, 300), and (360, 360, 360). The snapshots were saved as 256 × 256 pixel resolution PNG files (RGB) and divided into three types of datasets: training (Tra), validation (Val), and test (Test).…”

“…It has been shown that GoogLeNet performs better than AlexNet for classification, detection, and counting [48,[58][59][60][61][62][63]. For training the model, we used GoogLeNet with a 22 layer deep CNN, comprising two convolutional layers, two pooling layers (four MAX pools and one AVG pool), and nine "Inception" modules, in which each module has six convolution layers, one pooling layer, and 4 million parameters [64,65]; the deep CNN was implemented using open-source software on the CentOS Linux distribution 7.3.1611. We performed 25-and 5-fold experiments to investigate the effect of angle increments and dataset split ratios on prediction performance.…”

Vehicle Classification Based on FBG Sensor Arrays Using Neural Networks

“…On the other hand, the architecture of GoogLeNet consists of 22 layers (nine inception modules). The main motivation for the inception modules' (layers') creation is to make a deeper CNN network so that highly accurate results could be achieved [23,26,27]. For vehicle classification, several works using deep learning and convolutional neural networks were described in [20].…”

Res-WGAN: Image Classification for Plant Small-scale Datasets

“…Multilevel convolution and activation layers are used to extract high-dimensional features from raw images to simplify the final classification. Thus, CNN is more suitable for image processing than other traditional machine learning methods [7].…”