Convolutional Neural Networks (CNNs) are the prevalent technology in computer vision and have become increasingly popular for medical imaging data classification and analysis. In this field, due to the scarcity of medical data, pretrained ResNets on ImageNet can be considered a suitable first approach. This paper examines the medical imaging classification accuracy of conventional basic custom CNNs compared to ImageNet pretrained ResNets on various medical datasets in an effort to give more information about the importance of medical data and its preprocessing techniques for disease studies. Microscope-extracted cytological images were examined along with chest X-rays, MRI brain scans, and melanoma photographs. The medical images were examined in various sets, class combinations, and resolutions. Augmented image datasets and asymmetrical training and validation splits among the classes were also examined. Models were developed after they were tested and fine-tuned with respect to their network size, parameter values and network methods, image resolution, size of dataset, multitude, and genre of class. Overfitting was also examined, and comparative studies regarding the computational cost of different models were performed. The models achieved high accuracy in image classification that varies depending on the dataset and can be easily incorporated in future over-the-internet medical decision-supporting (telemedicine) environments. In addition, it appeared that conventional basic custom CNN overperformed ImageNet pretrained ResNets. The obtained results indicate the importance of utilizing medical image data as a testbed for improvements in CNN classification performance and the possibility of using CNNs and data preprocessing techniques for disease studies.

The adequacy and efficacy of simple and hybrid machine learning and Computational Intelligence algorithms were evaluated for the classification of potential prostate cancer patients in two distinct categories, the high- and the low-risk group for PCa. The evaluation is based on randomly generated surrogate data for the biomarker PSA, considering that reported epidemiological data indicated that PSA values follow a lognormal distribution. In addition, four more biomarkers were considered, namely, PSAD (PSA density), PSAV (PSA velocity), PSA ratio, and Digital Rectal Exam evaluation (DRE), as well as patient age. Seven simple classification algorithms, namely, Decision Trees, Random Forests, Support Vector Machines, K-Nearest Neighbors, Logistic Regression, Naïve Bayes, and Artificial Neural Networks, were evaluated in terms of classification accuracy. In addition, three hybrid algorithms were developed and introduced in the present work, where Genetic Algorithms were utilized as a metaheuristic searching technique in order to optimize the training set, in terms of minimizing its size, to give optimal classification accuracy for the simple algorithms including K-Nearest Neighbors, a K-means clustering algorithm, and a genetic clustering algorithm. Results indicated that prostate cancer cases can be classified with high accuracy, even by the use of small training sets, with sizes that could be even smaller than 30% of the dataset. Numerous computer experiments indicated that the proposed training set minimization does not cause overfitting of the hybrid algorithms. Finally, an easy-to-use Graphical User Interface (GUI) was implemented, incorporating all the evaluated algorithms and the decision-making procedure.

Based on the need for radiobiological databases, in this work, we mined experimental ionizing radiation data of human cells treated with X-rays, γ-rays, carbon ions, protons and α-particles, by manually searching the relevant literature in PubMed from 1980 until 2024. In order to calculate normal and tumor cell survival α and β coefficients of the linear quadratic (LQ) established model, as well as the initial values of the double-strand breaks (DSBs) in DNA, we used WebPlotDigitizer and Python programming language. We also produced complex DNA damage results through the fast Monte Carlo code MCDS in order to complete any missing data. The calculated α/β values are in good agreement with those valued reported in the literature, where α shows a relatively good association with linear energy transfer (LET), but not β. In general, a positive correlation between DSBs and LET was observed as far as the experimental values are concerned. Furthermore, we developed a biophysical prediction model by using machine learning, which showed a good performance for α, while it underscored LET as the most important feature for its prediction. In this study, we designed and developed the novel radiobiological ‘RadPhysBio’ database for the prediction of irradiated cell survival (α and β coefficients of the LQ model). The incorporation of machine learning and repair models increases the applicability of our results and the spectrum of potential users.

Despite the increasing application of machine learning and computational intelligence algorithms in medicine and physiotherapy, accurate classification and prognosis algorithms for postoperative patients in the rehabilitation phase are still lacking. The present study was carried out in two phases. In Phase I, classification performance of simple machine learning algorithms applied on data of patients suffering of reverse total shoulder arthroplasty (RTSA), examining algorithms’ classification accuracy and patients’ rehabilitation prognosis. In Phase II, hybrid computational intelligence algorithms were developed and applied in order to search for the minimum possible training set that achieves the maximum classification and prognostic performance. The data included features like age and gender, passive range of available motion of all movements (preoperative and postoperative), visual analog pain scale (preoperative and postoperative), and total rehabilitation time. In Phase I, K-nearest neighbors (ΚΝΝ) classification algorithm and K-means clustering algorithm (GAKmeans) were applied. Also, a genetic algorithm (GA)-based clustering algorithm (GAClust) was also applied. To achieve 100% performance on the test set, KNN used 80% of the data in the training set, whereas K-means and GAClust used 90% and 53.3%, respectively. In Phase II, additional computational intelligence algorithms were developed, namely, GAKNN (Genetic Algorithm K-nearest neighbors), GAKmeans, and GA2Clust (genetic algorithm-based clustering algorithm 2), for genetic algorithm optimization of the training set. Genetic algorithm optimization of the training set using hybrid algorithms in Phase II resulted in 100% performance on the test set by using only 35% of the available data for training. The proposed hybrid algorithms can reliably be used for patients’ rehabilitation prognosis.