Preferred Language
Articles
/
fBb2j4oBVTCNdQwCuJ_G
Deep Transfer Learning for Improved Detection of Keratoconus using Corneal Topographic Maps
...Show More Authors
Abstract <p>Clinical keratoconus (KCN) detection is a challenging and time-consuming task. In the diagnosis process, ophthalmologists must revise demographic and clinical ophthalmic examinations. The latter include slit-lamb, corneal topographic maps, and Pentacam indices (PI). We propose an Ensemble of Deep Transfer Learning (EDTL) based on corneal topographic maps. We consider four pretrained networks, SqueezeNet (SqN), AlexNet (AN), ShuffleNet (SfN), and MobileNet-v2 (MN), and fine-tune them on a dataset of KCN and normal cases, each including four topographic maps. We also consider a PI classifier. Then, our EDTL method combines the output probabilities of each of the five classifiers to obtain a decision based on the fusion of probabilities. Individually, the classifier based on PI achieved 93.1% accuracy, whereas the deep classifiers reached classification accuracies over 90% only in isolated cases. Overall, the average accuracy of the deep networks over the four corneal maps ranged from 86% (SfN) to 89.9% (AN). The classifier ensemble increased the accuracy of the deep classifiers based on corneal maps to values ranging (92.2% to 93.1%) for SqN and (93.1% to 94.8%) for AN. Including in the ensemble-specific combinations of corneal maps’ classifiers and PI increased the accuracy to 98.3%. Moreover, visualization of first learner filters in the networks and Grad-CAMs confirmed that the networks had learned relevant clinical features. This study shows the potential of creating ensembles of deep classifiers fine-tuned with a transfer learning strategy as it resulted in an improved accuracy while showing learnable filters and Grad-CAMs that agree with clinical knowledge. This is a step further towards the potential clinical deployment of an improved computer-assisted diagnosis system for KCN detection to help ophthalmologists to confirm the clinical decision and to perform fast and accurate KCN treatment.</p>
Scopus Clarivate Crossref
View Publication
Publication Date
Thu Jan 01 2015
Journal Name
Journal Of Al-mansoor College
An Improvement to Face Detection Algorithm for Non-Frontal Faces
...Show More Authors

Publication Date
Sat Apr 05 2025
Journal Name
2025 Ieee 4th International Conference On Computing And Machine Intelligence (icmi)
From Pixels to Diagnosis: AI-Powered CNN for Pneumonia Detection
...Show More Authors

View Publication Preview PDF
Crossref (1)
Scopus Crossref
Publication Date
Fri Feb 08 2019
Journal Name
Iraqi Journal Of Laser
Photoacoustic Imaging for Tumor Detection: An in vitro Simulation Study
...Show More Authors

Photoacoustic is a unique imaging method that combines the absorption contrast of light or radio frequency waves with ultrasound resolution. When the deposition of this energy is sufficiently short, a thermo-elastic expansion takes place whereby acoustic waves are generated. These waves can be recorded and stored to construct an image. This work presents experimental procedure of laser photoacoustic two dimensional imaging to detect tumor embedded within normal tissue. The experimental work is accomplished using phantoms that are sandwiched from fish heart or blood sac (simulating a tumor) 1-14mm mean diameter embedded within chicken breast to simulate a real tissue. Nd: YAG laser of 1.064μm and 532nm wavelengths, 10ns pulse duration, 4

... Show More
View Publication Preview PDF
Publication Date
Fri Feb 08 2019
Journal Name
Journal Of The College Of Education For Women
Minimum Spanning Tree Algorithm for Skin Cancer Image Object Detection
...Show More Authors

This paper proposes a new method Object Detection in Skin Cancer Image, the minimum
spanning tree Detection descriptor (MST). This ObjectDetection descriptor builds on the
structure of the minimum spanning tree constructed on the targettraining set of Skin Cancer
Images only. The Skin Cancer Image Detection of test objects relies on their distances to the
closest edge of thattree. Our experimentsshow that the Minimum Spanning Tree (MST) performs
especially well in case of Fogginessimage problems and in highNoisespaces for Skin Cancer
Image.
The proposed method of Object Detection Skin Cancer Image wasimplemented and tested on
different Skin Cancer Images. We obtained very good results . The experiment showed that

... Show More
View Publication Preview PDF
Publication Date
Fri Jan 01 2021
Journal Name
Ieee Access
DNA Encoding and STR Extraction for Anomaly Intrusion Detection Systems
...Show More Authors

View Publication
Scopus (11)
Crossref (9)
Scopus Clarivate Crossref
Publication Date
Fri Nov 15 2024
Journal Name
Scientific Reports
Ag@WO3 core–shell nanocomposite for wide range photo detection
...Show More Authors

This study successfully synthesized high-performance photodetectors based on Ag-WO3 core–shell heterostructures using a simple and economical two-step pulsed laser ablation in water method and has investigated the electrical characteristics of the Ag@WO3 nanocomposite heterojunction. The Hall effect tests indicate that the synthesized Ag@WO3 exhibits n-type conduction with a Hall mobility of 1.25 × 103 cm2V-1S-1. Dark current–voltage properties indicated that the created heterojunctions displayed rectification capabilities, with the highest rectification factor of around 1.71 seen at a 5 V bias. A photodetector’s responsivity reveals the existence of two response peaks, which are situated in the ultraviolet and visible region. The ph

... Show More
View Publication
Scopus (12)
Crossref (10)
Scopus Clarivate Crossref
Publication Date
Fri Dec 30 2011
Journal Name
Iraqi Journal Of Chemical And Petroleum Engineering
Improved Method to Correlate and Predict Isothermal VLE Data of Binary Mixtures
...Show More Authors

Accurate predictive tools for VLE calculation are always needed. A new method is introduced for VLE calculation which is very simple to apply with very good results compared with previously used methods. It does not need any physical property except each binary system need tow constants only. Also, this method can be applied to calculate VLE data for any binary system at any polarity or from any group family. But the system binary should not confirm an azeotrope. This new method is expanding in application to cover a range of temperature. This expansion does not need anything except the application of the new proposed form with the system of two constants. This method with its development is applied to 56 binary mixtures with 1120 equili

... Show More
View Publication Preview PDF
Publication Date
Tue Jan 01 2019
Journal Name
Energy Procedia
Improved thermal and mechanical properties of CdBa2-x SrxCa2Cu3O8+δ superconducting compounds
...Show More Authors

View Publication
Scopus Clarivate Crossref
Publication Date
Mon Oct 05 2020
Journal Name
International Journal Of Advanced Science And Technology
Improved Merging Multi Convolutional Neural Networks Framework of Image Indexing and Retrieval
...Show More Authors

Improved Merging Multi Convolutional Neural Networks Framework of Image Indexing and Retrieval

Publication Date
Fri May 01 2020
Journal Name
International Journal Of Advanced Science And Technology
Improved Merging Multi Convolutional Neural Networks Framework of Image Indexing and Retrieval
...Show More Authors

Background/Objectives: The purpose of current research aims to a modified image representation framework for Content-Based Image Retrieval (CBIR) through gray scale input image, Zernike Moments (ZMs) properties, Local Binary Pattern (LBP), Y Color Space, Slantlet Transform (SLT), and Discrete Wavelet Transform (DWT). Methods/Statistical analysis: This study surveyed and analysed three standard datasets WANG V1.0, WANG V2.0, and Caltech 101. The features an image of objects in this sets that belong to 101 classes-with approximately 40-800 images for every category. The suggested infrastructure within the study seeks to present a description and operationalization of the CBIR system through automated attribute extraction system premised on CN

... Show More