Background: Accurate estimation of the age of an ischaemic stroke lesion is clinically important because therapeutic decisions are strongly dependent on time from stroke onset. Non-contrast computed tomography (NCCT) remains the most accessible first-line neuroimaging modality for stroke assessment particularly in resource-constrained settings. However, temporal changes in infarct appearance may be subtle and difficult to characterise consistently by visual assessment alone; quantitative texture analysis combined with machine learning may provide an objective approach to automated lesion-age classification. This study compared the performance of an Artificial Neural Network (ANN) and a k-Nearest Neighbour (k-NN) classifier for classifying ischaemic stroke lesion age using statistical texture features extracted from NCCT images. Methods: A prospective cross-sectional purposive study was conducted among 197 patients with clinically diagnosed ischaemic stroke who underwent NCCT at three diagnostic centres in Nigeria. Following radiological confirmation, 316 regions of interest (ROIs) comprising 158 normal brain tissue ROIs, 58 acute lesions, 50 subacute lesions and 50 chronic lesions were analysed. Four classes of statistical texture descriptors—Histogram, Grey-Level Co-occurrence Matrix (GLCM/COM), Run-Length Matrix (RLM) and Absolute Gradient were extracted from the selected ROIs. Feature reduction was performed using the Fisher coefficient and the most discriminating parameters from each texture class were used for classification. A multilayer feed-forward ANN and a k-NN classifier with k = 1 were implemented using scikit-learn. The dataset was divided into 80% training and 20% testing subsets using stratified sampling. Classification performance was assessed using sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), F1-score, Cohen's kappa, and area under the receiver operating characteristic curve (AUROC).