<![CDATA[As a result, abnormal cells develop in the body, leading to a highly constitutive cell type that is a significant risk to the patient's functional capabilities and vital processes. The early and accurate recognition of such cells is crucial for accurate diagnosis and prognosis, and this recognition is made possible by medical imaging techniques, particularly magnetic resonance imaging (MRI). Despite advances in 3D learning models, several scientific studies involving deep convolutional networks (CNNs) still face numerous challenges. These challenges include the underutilization of spatial information, the inability of traditional data reduction techniques to minimise data dimensionality during the assembly phase, and suboptimal data processing during the data synchronisation or listening. In addition, some approaches require large volumes of data to achieve sufficient performance, which limits their applicability to real-world healthcare scenarios. This paper discusses the V-Net model that has been trained for a relatively long time to process volumetric 3D data, including a wide variety of very small sub-3D spatial volumes. This work used a large global MRI dataset, split into 80% for the training set and 20% for the test set. Before the tests, the images were preprocessed by resizing them to 128 × 128, applying Min-Max normalisation, and CLAHE (Contrast Limited Adaptive Histogram Equalisation) to enhance contrastof the images. The results showed that the proposed model achieved a 99% improvement in tumour detection performance over all other approaches. The findings indicate that employing specialised architectures like V-Net may significantly enhance the efficiency of medical diagnostic imaging specialists.]]>
Copyrights © 2026
