Performance study of augmentation techniques for HEp2 CNN classification

Ebrahim, Maad; Alsmirat, Mohammad A.; Al‐Ayyoub, Mahmoud

doi:10.1109/iacs.2018.8355460

Cited by 12 publications

(9 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For that reason, Many modifications are made in order to improve the results. These modifications includes changing with many parameters turnings, preprocessing of the image set, pre-augmentation on the image set that is inspired by [6], and followed the suggestions in Brownlee's blog [37]. The…”

Section: B Disk Herniation Type Classificationmentioning

confidence: 99%

“…Data augmentation is a technique used to increase the size of the dataset using a set of image transformation techniques before feeding them to train the system. It is very well known in the literature [6] that augmentation always increase the accuracy of CNN because it increases the size of the training data and it is currently one of the well known deep learning hyper parameters that can be tuned to enhance the deep learning accuracy [7].…”

mentioning

confidence: 99%

See 1 more Smart Citation

Deep Learning-Based Disk Herniation Computer Aided Diagnosis System From MRI Axial Scans

et al. 2022

Self Cite

View full text Add to dashboard Cite

Computer-aided diagnosis (CAD) systems have been the focus of many researchers in both computer and medical fields. In this paper, we build two convolutional neural network (CNN) based CAD systems for diagnosing lumbar disk herniation from Magnetic Resonance Imaging (MRI) axial scans. The first one is a disk herniation detection CAD system which is a binary CAD system that determines whether the case image contains disk herniation or not. The second system is a disk herniation type classification CAD system which can determine the type of the disk herniation in the image if one exists. To train and test the proposed systems, an image set is built and annotated with the help of a radiologist. In order to get rid of the "noisy" parts of the input images and reduce their complexity, we experiment with different ROI extraction methods. The image set is also preprocessed and enlarged using augmentation techniques to make it suitable to be used with CNN. There are many novel aspects of this work. First, the problems of disk herniation detection and recognition from axial scans are not well-studied in the literature. Second, we use deep learning techniques which produces ground-breaking results in many image processing tasks, but are yet to reach their full potential with medical image processing in general. Finally, we explore the use of many innovative techniques such as data augmentation and transfer learning, which greatly improves the accuracy of our models. The results of our systems are impressive with a 95.65% accuracy for the detection problem and a 91.38% accuracy for the recognition problem.

show abstract

Section: B Disk Herniation Type Classificationmentioning

confidence: 99%

mentioning

confidence: 99%

Deep Learning-Based Disk Herniation Computer Aided Diagnosis System From MRI Axial Scans

et al. 2022

Self Cite

View full text Add to dashboard Cite

show abstract

“…Data augmentation is a useful way to reduce the generalization error (overfitting) of models by increasing the amount of training data and adding a variety of distortions and noise to the training data. This aims to strengthen and enhance the robustness of the model [35]. We have applied four groups of data augmentation techniques on images and segmentation maps.…”

Section: Data Pre-processing and Augmentationmentioning

confidence: 99%

The 2ST-UNet for Pneumothorax Segmentation in Chest X-Rays using ResNet34 as a Backbone for U-Net

Abedalla¹,

Abdullah²,

Al‐Ayyoub³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Pneumothorax, also called a collapsed lung, refers to the presence of the air in the pleural space between the lung and chest wall. It can be small (no need for treatment), or large and causes death if it is not identified and treated on time. It is easily seen and identified by experts using a chest X-ray. Although this method is mostly error-free, it is time-consuming and needs expert radiologists. Recently, Computer Vision has been providing great assistance in detecting and segmenting pneumothorax. In this paper, we propose a 2-Stage Training system (2ST-UNet) to segment images with pneumothorax. This system is built based on U-Net with Residual Networks (ResNet-34) backbone that is pre-trained on the ImageNet dataset. We start with training the network at a lower resolution before we load the trained model weights to retrain the network with a higher resolution. Moreover, we utilize different techniques including Stochastic Weight Averaging (SWA), data augmentation, and Test-Time Augmentation (TTA). We use the chest X-ray dataset that is provided by the 2019 SIIM-ACR Pneumothorax Segmentation Challenge, which contains 12,047 training images and 3,205 testing images. Our experiments show that 2-Stage Training leads to better and faster network convergence. Our method achieves 0.8356 mean Dice Similarity Coefficient (DSC) placing it among the top 9% of models with a rank of 124 out of 1,475.

show abstract

“…Another preprocessing step the system performs is DA. While DA has been shown to be very effective for image processing tasks (Chatfield et al, 2014;He et al, 2016;Chollet, 2016;Ebrahim et al, 2018), it use in text processing tasks is still limited (Fadaee et al, 2017;Kafle et al, 2017). Since the training data is small, a data augmentation step is performed on Corpus-26 by applying random shuffling on Corpus.…”

Section: Systemmentioning

confidence: 99%

Team JUST at the MADAR Shared Task on Arabic Fine-Grained Dialect Identification

Talafha

Fadel

Al‐Ayyoub

et al. 2019

Proceedings of the Fourth Arabic Natural Language Processing Workshop

Self Cite

View full text Add to dashboard Cite

In this paper, we describe our team's effort on the MADAR Shared Task on Arabic Fine-Grained Dialect Identification. The task requires building a system capable of differentiating between 25 different Arabic dialects in addition to MSA. Our approach is simple. After preprocessing the data, we use Data Augmentation (DA) to enlarge the training data six times. We then build a language model and extract n-gram word-level and character-level TF-IDF features and feed them into an MNB classifier. Despite its simplicity, the resulting model performs really well producing the 4th highest F-measure and region-level accuracy and the 5th highest precision, recall, city-level accuracy and country-level accuracy among the participating teams.

show abstract

Performance study of augmentation techniques for HEp2 CNN classification

Cited by 12 publications

References 18 publications

Deep Learning-Based Disk Herniation Computer Aided Diagnosis System From MRI Axial Scans

Deep Learning-Based Disk Herniation Computer Aided Diagnosis System From MRI Axial Scans

The 2ST-UNet for Pneumothorax Segmentation in Chest X-Rays using ResNet34 as a Backbone for U-Net

Team JUST at the MADAR Shared Task on Arabic Fine-Grained Dialect Identification

Contact Info

Product

Resources

About