http://irepo.futminna.edu.ng:8080/jspui/handle/123456789/120 2026-08-02T15:43:49Z http://irepo.futminna.edu.ng:8080/jspui/handle/123456789/31924 Title: Machine Learning-based Medical Image Compression Using Principal Component Analysis (PCA) and Autoencoders Authors: Sanni, Henry; Dada, Michael; Awojoyogbe, Bamidele Abstract: Medical imaging creates large datasets crucial for diagnosis, posing storage and transmission challenges in digital health systems. Traditional compression methods like JPEG and PNG often lose critical diagnostic details, necessitating more advanced techniques. This study explores machine learning-based compression using Principal Component Analysis (PCA) and Autoencoders to achieve high compression ratios while preserving diagnostic quality. The models were trained and evaluated on medical imaging datasets, comparing reconstruction error, compression ratio, and diagnostic accuracy retention. Results demonstrate that Autoencoders outperform PCA in preserving diagnostically relevant features at higher compression rates. Description: None 2024-11-08T00:00:00Z http://irepo.futminna.edu.ng:8080/jspui/handle/123456789/31923 Title: Evaluation of Generative Medical Artificial Intelligence based on Fine-tuned Large Language Models (LLM) Authors: Bobai, Terry; Dada, Michael; Awojoyogbe, Bamidele Abstract: The integration of large language models (LLMs) into medical applications has opened new frontiers in clinical decision support and patient management. Fine-tuned generative models can significantly improve diagnostics and personalized care but require rigorous evaluation for reliability and accuracy. While LLMs excel in natural language generation, their medical applications face challenges, including accuracy, clinical relevance, and ethical use. Ensuring these models align with real-world practices and data privacy is crucial for adoption. This study fine-tunes a GPT-style LLM using a curated medical dataset, comprising publicly available resources (e.g., PubMed articles, clinical guidelines, medical textbooks, electronic health records (EHR). Etc.) and anonymized patient records. The dataset was curated by removing irrelevant data, anonymizing sensitive information, and annotating key medical concepts for improved model learning. The model's performance was evaluated across tasks like generating medical summaries and proposing diagnoses, using accuracy, precision, recall, F1 score, and human evaluations as key metrics. The fine-tuned model achieved an accuracy of 85% in generating patient summaries and 80% for differential diagnostic suggestions. The F1 score was 0.82, reflecting a good balance of precision and recall. However, in complex cases, accuracy dropped to 70%, and factual inaccuracies emerged. Ethical concerns, such as biases in training data, were also noted, emphasizing the need for ongoing improvements and oversight. Generative medical AI based on fine-tuned LLMs holds significant potential for enhancing healthcare delivery. Nonetheless, challenges related to accuracy, transparency, and ethical compliance must be addressed before integration into clinical practice. Future work will focus on improving model robustness and alignment with regulatory standards. Description: None 2024-11-08T00:00:00Z http://irepo.futminna.edu.ng:8080/jspui/handle/123456789/31922 Title: Derivation of Multidimensional Force for Magnetic Tweezer as a Function of T₁ and T₂ Relaxation Times using Magnetic Resonance Diffusion and Langevin Equations Authors: Lazarus, John; Dada, Michael; Awojoyogbe, Bamidele; Abolarinwa, Simon Abstract: Magnetic tweezer is an instrument that has the ability to exert force in terms of magnitude and direction on biomolecules, such as deoxyribonucleic acid (DNA). It is a force spectroscopy technique with wide range of applications ranging from determination of elastic properties of DNA, study of DNA-protein interactions, study of topology of DNA, study of cancer cells, among others. Magnetic field is produced when magnetic tweezer applies force on a biomolecule. The force applied is along the axis of the molecule and field direction. Several models of magnetic tweezer have been developed overtime for measurement of force applied on a particle, but each developed model is marred with limitations such as magnetic hysteresis of the core material in the tweezer and low level spatial resolution. Also, there is limited availability of knowledge on the incorporation of nuclear magnetic resonance (NMR) relaxation time constants (T₁ and T₂), with NMR diffusion and Langevin equations to determine force applied on a particle along different axis. Therefore, this study derived a multidimensional force for magnetic tweezer from T₁ and T₂ relaxation times using NMR diffusion and Langevin equations. As this model helps to overcome the limitations encountered by the previous models. The NMR diffusion equation: ∂Mᵧ/∂t = D∇²Mᵧ + (F₀/T₀)yB₁(r,t), and Langevin equation: M(dv/dt) = -ζv + X(t) were utilised in the study to derive a multidimensional force at different boundary conditions. The equations are used to derive power spectrum from T₁ and T₂ relaxation times, given that T₀ = 1/T₁ + 1/T₂. The multidimensional force derived in this study is in x, y, and z orientation with each axis having its characteristic variable. The result obtained from this study can be used to simulate T₁ and T₂ relaxation times values of DNA-binding proteins, an essential protein which enhances DNA-protein interactions. The results of the simulation can depict different features of DNA, thereby availing knowledge on DNA classification, genetic abnormality in cancer cells, among others. Description: None 2024-11-08T00:00:00Z http://irepo.futminna.edu.ng:8080/jspui/handle/123456789/31921 Title: Derivation of Fredholm Integrals of the First Kind in terms of Analytical Solutions to Bloch Nuclear Magnetic Resonance (NMR) Flow Equation and Relaxometric Tissue Profiles Authors: Genyi, Terfa; Awojoyogbe, Bamidele; Dada, Michael; Olarinoye, Oyeleke Abstract: The Bloch NMR equation is fundamental to understanding nuclear magnetization in the presence of diffusion, relaxation, and magnetic fields. However, it becomes particularly complex when applied to fluid mixtures with varying diffusion coefficients and relaxation properties. This complexity necessitates advanced mathematical techniques for deriving accurate analytical solutions. In this study, we derive the Fredholm integrals of the first kind based on these analytical solutions to the Bloch MRI equation for relaxation time. Using these solutions as a modified kernel distribution function and Tikhonov regularization providing accurate estimates of the T1 and T2 maps, a Python algorithm has been developed for interpreting NMR signals in heterogeneous systems with useful outputs. This is particularly useful in developing improved methods for multidimensional correlation magnetic resonance imaging (MRI) for general tissue diagnosis. Furthermore, this method offers a new perspective on the inverse problem of reconstructing molecular dynamics and interactions from observed NMR relaxometric data distribution which provides vital insights into tissue microstructure that enhances diagnostic capabilities in clinical MRI. Description: None 2024-11-08T00:00:00Z