https://www.um.edu.mt/library/oar/handle/123456789/141711 2026-05-31T22:50:18Z https://www.um.edu.mt/library/oar/handle/123456789/145972 Title: Enhancing spatial feature development from imagery using computer vision aided by GIS Abstract: This dissertation addresses the challenge of detecting and validating spatial features of differing geometric complexity. It focuses on buildings and swimming pools extracted from high-resolution satellite imagery and orthophotos for GIS applications. Conventional object detection workflows often lack mechanisms to reconcile computer vision outputs with authoritative spatial data. This limitation reduces their reliability for urban and environmental planning. The research is motivated by the need to integrate deep learning–based detection with GIS-based spatial validation to improve confidence and interpretability. The methodology uses the YOLOv11 object detection framework trained on approximately 1,000 manually annotated images. Both single-class and multi-class configurations are evaluated. Due to limitations in ArcGIS Pro’s native deep learning toolbox, inference is performed externally. Detection outputs are then reintroduced into the GIS environment. A novel GIS–CV integration pipeline is implemented using the arcpy library. Post-inference spatial refinement is applied using Intersection over Union (IoU) and Dice coefficient analysis. Authoritative planning basemap polygons are used to enable confidence reweighting. After spatial validation, the single-class swimming pool model achieved a mAP@0.5 of 0.78. It obtained a precision of 0.85 and a recall of 0.75 after 122 epochs. The runtime for this model was 0.289 hours. The building detection model achieved a mAP@0.5 of 0.45 after 100 epochs. It recorded a precision of 0.698 and a recall of 0.626, with a runtime of 0.254 hours. Pool mAP@0.5 increased from 0.74 to 0.78, while building mAP@0.5 increased from 0.439 to 0.45. A multi-class model detecting buildings, pools, and vegetation achieved an overall mAP@0.5 of 0.475. This model recorded a precision of 0.54 and a recall of 0.489. This main contribution is a custom GIS–CV pipeline with a novel post-inference validation framework. This approach enhances detection reliability and supports scalable integration of computer vision outputs into operational GIS workflows. Description: M.Sc. ICT(Melit.) 2026-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/145970 Title: A tool to support the diagnosis of Alzheimer’s disease Abstract: Alzheimer’s disease is a progressive neurodegenerative disorder affecting millions of individuals worldwide. Currently, there is no cure, making early recognition critical, as timely interventions can help slow functional deterioration and maintain quality of life. This dissertation aimed to develop a tool to support the diagnosis of Alzheimer’s disease. The tool is designed to complement existing assessment methods rather than replace them, supporting practitioners in their diagnostic process. The objectives were achieved by developing a Convolutional Neural Network (CNN) model to classify MRI scans into four stages of Alzheimer’s disease, creating a prototype web application to evaluate whether the integration of the model with it is feasible, and conducting interviews with domain experts to inform the tool’s features and functionalities. The prototype was then refined and re-evaluated with expert feedback. The resulting web application allows authorised medical specialists to log in, manage patient information, upload MRI scans, predict the stage of Alzheimer’s disease, and access comprehensive reports that include both current and past scans for comparison. This tool demonstrates the potential of integrating artificial intelligent assisted imaging analysis into clinical workflows to support more informed and efficient diagnostic decisions. Domain experts evaluated the tool as aesthetically pleasing, easy to follow, clear, and straightforward to use. Description: M.Sc.(Melit.) 2026-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/145969 Title: Artificial intelligence in breast positioning and quality assurance in mammography Abstract: Accurate breast positioning in mammography is essential for diagnostic image quality and quality assurance, yet image-evaluation systems such as PGMI (Perfect, Good, Moderate, Inadequate) are subjective, time-consuming and prone to inter-and intra observer variability. This dissertation aimed to train, test, and validate deep learning models for assessing breast positioning on medio-lateral oblique views using the posterior nipple line (PNL) criterion, and to evaluate radiographers’ perceptions of AI in breast positioning and quality assurance. A mediolateral-oblique subset of the VinDr-Mammo dataset (n=2,000) was matched by SOPInstanceUID to the deep-breast-positioning GitHub repository, and models were replicated. Two strategies were studied: (i) landmark regression (by replicating the U Net, Attention U-Net, CoordAtt U-Net and ResNeXt-50 models, and employing a novel HRNet), with Good/Bad labels derived post hoc via a deterministic PNL rule, and (ii) direct image-level classification (ResNeXt-50 replica, Optuna-tuned ResNeXt-50, ConvNeXt-Tiny, and EfficientNet-B3). The performance metrics for regression included per-landmark Euclidean error (mm) and pectoral-line angular error (°), while those for classification included macro-F1 and ROC-AUC on the test set. Results were reported as mean ±standard deviation across five seeds. In parallel, a prospective cross-sectional questionnaire was distributed amongst radiographers working in the mammography unit (n=9) at a local general public hospital in Malta. For regression, HRNet yielded the lowest landmark and angular errors and, via the PNL rule, the strongest derived classification (accuracy 94.20±1.04%; F1(Bad) 92.67±1.27%). For direct classification, ConvNeXt-Tiny provided the most balanced performance (macro-F1 82.64±2.07%; accuracy 83.40±2.22%), while EfficientNet-B3 was lower on macro-F1 (82.10±2.42%) but achieved the highest Sensitivity(Bad) (84.16±5.91%) and ROC-AUC (90.65±2.57%); both exceeded ResNeXt-50 baselines. Questionnaire response rate was 88.9%. PGMI was viewed as subjective (4.25/5) and time-consuming (3.75/5). Adoption enablers were workflow integration (n=6) and training (n=5); concerns were over-reliance (n=7), accountability (n=6) and reduced autonomy (n=5). Amongst the approaches evaluated, HRNet achieved the strongest landmark regression performance and consequently the best post-hoc PNL-derived Good/Bad grading, whereas for direct image-level classification, ConvNeXt-Tiny provided the most balanced overall performance, with EfficientNet-B3 achieving the highest ROC AUC and sensitivity for Bad cases. Questionnaire findings indicate that radiographers perceive practical value in AI support for positioning, particularly for improving consistency and enabling real-time feedback, while emphasising that adoption depends on training and workflow integration. However, external validation on independent datasets is required to confirm generalisable performance prior to prospective evaluation in clinical practice. Description: M.Sc.(Melit.) 2026-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/145968 Title: Auto-segmentation of organs at risk in prostate cancer patients for MR-Linac Abstract: Accurate and efficient delineation of pelvic organs-at-risk is essential for MRI-based prostate radiotherapy. Manual contouring is time-consuming and prone to observer variation. This study evaluates whether deep-learning auto-segmentation can provide reliable contours for prostate MRI planning by comparing a slice-wise Swin-UNet with a volumetric nnFormer. An expertly annotated, anonymised T2-weighted MRI cohort was used for development and testing (training n = 261; testing n = 66). Model performance was assessed against expert contours using the Dice Similarity Coefficient (DSC) and the 95th percentile Hausdorff distance, alongside a qualitative visual review to characterise typical failure modes. On the test set, nnFormer outperformed Swin-UNet for the bladder with DSC 0.77 versus 0.66 and HD95 1.69 versus 4.77 mm, and for the clinical target volume (CTV) with DSC 0.74 versus 0.63 and HD95 1.64 versus 3.12 mm. For the rectum, nnFormer achieved a lower HD95 of 1.93 versus 4.39 mm but a slightly lower DSC of 0.65 versus 0.75. Qualitatively, nnFormer produced more spatially coherent and anatomically faithful contours with narrower case-to-case variability. In contrast, Swin-UNet was more susceptible to artefacts and slice-to-slice intensity variation, with common failure modes including under-segmentation of overfilled bladders and fragmented rectal walls in the presence of gas. These findings indicate that automated segmentation of the clinical target volume and organs-at-risk on prostate MRI is feasible with both architectures. However, the volumetric nnFormer provides superior boundary fidelity and overall accuracy for bladder and CTV. Further research is required to confirm and expand these results. This includes validating the approach on MR-linac systems across multiple centres, incorporating uncertainty-aware quality assurance to better guide human review, and prospectively assessing how the method affects editing time and dosimetric outcomes in MRI-only and MR-guided adaptive workflows. Description: M.Sc.(Melit.) 2026-01-01T00:00:00Z