Abstract
<p class="MsoNormal" style="text-align: justify;"><strong style="mso-bidi-font-weight: normal;"><span style="font-size: 10.0pt; font-family: 'Arial',sans-serif; mso-ascii-theme-font: minor-bidi; mso-hansi-theme-font: minor-bidi; mso-bidi-theme-font: minor-bidi;">Abstract</span></strong><span style="font-size: 10.0pt; font-family: 'Arial',sans-serif; mso-ascii-theme-font: minor-bidi; mso-hansi-theme-font: minor-bidi; mso-bidi-theme-font: minor-bidi;">: </span></p>
<p class="MsoNormal" style="text-align: justify;"><span style="font-size: 10.0pt; font-family: 'Arial',sans-serif; mso-ascii-theme-font: minor-bidi; mso-hansi-theme-font: minor-bidi; mso-bidi-theme-font: minor-bidi;">Background: </span></p>
<p class="MsoNormal" style="text-align: justify;"><span style="font-size: 10.0pt; font-family: 'Arial',sans-serif; mso-ascii-theme-font: minor-bidi; mso-hansi-theme-font: minor-bidi; mso-bidi-theme-font: minor-bidi;">As a result of the availability of high-speed computing devices, disease screening procedures in modern hospitals have significantly improved over the last few decades. As a result of this invention of deep learning procedures (DP), this work implemented modern diagnostic schemes to achieve accurate and fast results when screening patients for diseases with the aid of medical data. </span></p>
<p class="MsoNormal" style="text-align: justify;"><span style="font-size: 10.0pt; font-family: 'Arial',sans-serif; mso-ascii-theme-font: minor-bidi; mso-hansi-theme-font: minor-bidi; mso-bidi-theme-font: minor-bidi;">Method: </span></p>
<p class="MsoNormal" style="text-align: justify;"><span style="font-size: 10.0pt; font-family: 'Arial',sans-serif; mso-ascii-theme-font: minor-bidi; mso-hansi-theme-font: minor-bidi; mso-bidi-theme-font: minor-bidi;">This study applied pre-trained DP to detect Diabetic Foot Ulcers (DFU) from the test images. This work consists following stages; (i) Resizing, augmenting, and enhancing images, (ii) deep-features mining with a chosen DP, (iii) features reduction using 50% dropout and serial features-fusion, and (iv) Binary-classification through five-fold cross-validation. Two types of disease detection procedures implemented during the investigation: (a) Conventional deep-features and (b) fused deep-features (FD). </span></p>
<p class="MsoNormal" style="text-align: justify;"><span style="font-size: 10.0pt; font-family: 'Arial',sans-serif; mso-ascii-theme-font: minor-bidi; mso-hansi-theme-font: minor-bidi; mso-bidi-theme-font: minor-bidi;">Result: </span></p>
<p class="MsoNormal" style="text-align: justify;"><span style="font-size: 10.0pt; font-family: 'Arial',sans-serif; mso-ascii-theme-font: minor-bidi; mso-hansi-theme-font: minor-bidi; mso-bidi-theme-font: minor-bidi;">As a result of this study, the FD obtained with VGG16 and ResNet101 enabled 99.5% accuracy in DFU detection using SoftMax classifier. </span></p>
<p class="MsoNormal" style="text-align: justify;"><span style="font-size: 10.0pt; font-family: 'Arial',sans-serif; mso-ascii-theme-font: minor-bidi; mso-hansi-theme-font: minor-bidi; mso-bidi-theme-font: minor-bidi;">Conclusion: </span></p>
<p class="MsoNormal" style="text-align: justify;"><span style="font-size: 10.0pt; font-family: 'Arial',sans-serif; mso-ascii-theme-font: minor-bidi; mso-hansi-theme-font: minor-bidi; mso-bidi-theme-font: minor-bidi;">This demonstration confirmed that the proposed scheme is effective in detecting DFU from the chosen database.</span></p>
Publisher
Pharaoh Academy International Publishing Co., Limited