<p>Breast cancer remains a leading cause of cancer-related mortality among women worldwide, emphasizing the urgent need for accurate and early diagnostic methods. Hyperspectral imaging (HSI) offers a non-invasive, high-resolution approach for tissue analysis by combining spatial and spectral information across hundreds of contiguous bands. In this study, we present a line-scanning hyperspectral imaging system, integrated with probability density function (PDF) analysis and automated classification algorithms, for the detection and delineation of malignant breast tissue. Diffuse reflectance spectra were extracted from ex-vivo breast biopsies, and PDF analysis identified 445&#xa0;nm as the optimum diagnostic wavelength, showing the highest divergence between malignant and normal tissues. A total of 70 biopsy samples (35 malignant, 35 normal), confirmed by mammography and histopathology, were analyzed. K-means clustering was applied at the selected wavelength to segment malignant regions, with performance validated against pathologist annotations. The system achieved sensitivity of 97%, specificity of 95%, and overall accuracy of 95%, aligning closely with pathology reports. Receiver operating characteristic (ROC) analysis yielded an area under the curve (AUC) above 0.95, confirming strong diagnostic performance. Beyond classification, the system provides visual outputs that assist in tumor margin delineation, supporting surgical planning and personalized treatment strategies. These findings demonstrate the feasibility of integrating PDF-guided wavelength selection with HSI for robust breast cancer diagnostics, with potential applications in intraoperative guidance and digital pathology.</p>