Developing an Original Computational Framework for Personalised Dosimetry in External Beam Radiotherapy for Cancer Patients in Nigeria

External beam radiotherapy for cancer in Nigeria is delivered with dosimetry protocols developed from population-average anatomical models that systematically underestimate the impact of Nigerian patient morphometric, tissue composition, and physiological characteristics on dose distribution, creating individualized dosimetry errors with consequences for tumour control and normal tissue toxicity. This dissertation develops an original computational framework for personalised dosimetry in external beam radiotherapy specifically incorporating Nigerian patient characteristics, making contributions to radiation therapy physics and biomedical engineering. The framework is developed through four integrated research phases. The first phase establishes a Nigerian patient anatomy computational phantom library from deformable CT scan data collected from one hundred and sixty cancer patients at the National Hospital Abuja and Obafemi Awolowo University Teaching Hospital, spanning cervical, head and neck, and breast cancer cases representing the highest-burden cancer types in Nigeria. The second phase develops a Monte Carlo dose calculation engine incorporating Nigerian-patient-specific tissue composition parameters derived from dual-energy CT imaging and laboratory tissue composition measurements. The third phase constructs a machine learning-based dose prediction model trained on the Nigerian patient phantom library to enable rapid patient-specific dose distribution estimation without full Monte Carlo computation. The fourth phase performs clinical validation comparing framework-predicted dose distributions against measured dosimetry in phantom and clinical treatment plan verification. Original contributions include the Nigerian patient computational phantom library, the Nigerian tissue composition parameterization, and the phantom-trained rapid dose prediction model. Clinical implementation testing demonstrates personalised dosimetry corrections of up to 12 percent relative to standard protocols in high-BMI patients. Keywords: personalised dosimetry, radiotherapy, computational phantom, Monte Carlo, Nigeria.