3 results
Search Results
Now showing 1 - 3 of 3
Article Citation - WoS: 18Citation - Scopus: 23A Novel Hybrid Machine Learning Based System To Classify Shoulder Implant Manufacturers(Mdpi, 2022) Sivari, Esra; Guzel, Mehmet Serdar; Bostanci, Erkan; Mishra, AlokIt is necessary to know the manufacturer and model of a previously implanted shoulder prosthesis before performing Total Shoulder Arthroplasty operations, which may need to be performed repeatedly in accordance with the need for repair or replacement. In cases where the patient's previous records cannot be found, where the records are not clear, or the surgery was conducted abroad, the specialist should identify the implant manufacturer and model during preoperative X-ray controls. In this study, an auxiliary expert system is proposed for classifying manufacturers of shoulder implants on the basis of X-ray images that is automated, objective, and based on hybrid machine learning models. In the proposed system, ten different hybrid models consisting of a combination of deep learning and machine learning algorithms were created and statistically tested. According to the experimental results, an accuracy of 95.07% was achieved using the DenseNet201 + Logistic Regression model, one of the proposed hybrid machine learning models (p < 0.05). The proposed hybrid machine learning algorithms achieve the goal of low cost and high performance compared to other studies in the literature. The results lead the authors to believe that the proposed system could be used in hospitals as an automatic and objective system for assisting orthopedists in the rapid and effective determination of shoulder implant types before performing revision surgery.Article Citation - WoS: 9Citation - Scopus: 9Proton Therapy for Mandibula Plate Phantom(Mdpi, 2021) Senirkentli, Guler Burcu; Ekinci, Fatih; Bostanci, Erkan; Guzel, Mehmet Serdar; Dagli, Ozlem; Karim, Ahmad M.; Mishra, AlokPurpose: In this study, the required dose rates for optimal treatment of tumoral tissues when using proton therapy in the treatment of defective tumours seen in mandibles has been calculated. We aimed to protect the surrounding soft and hard tissues from unnecessary radiation as well as to prevent complications of radiation. Bragg curves of therapeutic energized protons for two different mandible (molar and premolar) plate phantoms were computed and compared with similar calculations in the literature. The results were found to be within acceptable deviation values. Methods: In this study, mandibular tooth plate phantoms were modelled for the molar and premolar areas and then a Monte Carlo simulation was used to calculate the Bragg curve, lateral straggle/range and recoil values of protons remaining in the therapeutic energy ranges. The mass and atomic densities of all the jawbone layers were selected and the effect of layer type and thickness on the Bragg curve, lateral straggle/range and the recoil were investigated. As protons move through different layers of density, lateral straggle and increases in the range were observed. A range of energies was used for the treatment of tumours at different depths in the mandible phantom. Results: Simulations revealed that as the cortical bone thickness increased, Bragg peak position decreased between 0.47-3.3%. An increase in the number of layers results in a decrease in the Bragg peak position. Finally, as the proton energy increased, the amplitude of the second peak and its effect on Bragg peak position decreased. Conclusion: These findings should guide the selection of appropriate energy levels in the treatment of tumour structures without damaging surrounding tissues.Article Citation - WoS: 12Citation - Scopus: 17A Novel Framework Using Deep Auto-Encoders Based Linear Model for Data Classification(Mdpi, 2020) Karim, Ahmad M.; Kaya, Hilal; Guzel, Mehmet Serdar; Tolun, Mehmet R.; Celebi, Fatih V.; Mishra, AlokThis paper proposes a novel data classification framework, combining sparse auto-encoders (SAEs) and a post-processing system consisting of a linear system model relying on Particle Swarm Optimization (PSO) algorithm. All the sensitive and high-level features are extracted by using the first auto-encoder which is wired to the second auto-encoder, followed by a Softmax function layer to classify the extracted features obtained from the second layer. The two auto-encoders and the Softmax classifier are stacked in order to be trained in a supervised approach using the well-known backpropagation algorithm to enhance the performance of the neural network. Afterwards, the linear model transforms the calculated output of the deep stacked sparse auto-encoder to a value close to the anticipated output. This simple transformation increases the overall data classification performance of the stacked sparse auto-encoder architecture. The PSO algorithm allows the estimation of the parameters of the linear model in a metaheuristic policy. The proposed framework is validated by using three public datasets, which present promising results when compared with the current literature. Furthermore, the framework can be applied to any data classification problem by considering minor updates such as altering some parameters including input features, hidden neurons and output classes.
