Cervical image visualization and improvement for clinical support in diagnosis and treatment

Cervical white light imaging (CWLI) plays a key role in diagnosing and managing cervical intraepithelial neoplasia (CIN) but is limited by uneven illumination, low contrast, and specular reflections. Cervical image visualization and improvement (CIVI) is designed to improve contrast and uniform brightness while maintaining the histological accuracy of colposcopic images to clearly identify tissue differences. CIVI combines multiple image enhancement techniques, including specular reflection removal, illumination correction, and local contrast enhancement, in a unified framework. This technique was validated on 52 histologically confirmed CWLI images (CIN 1-3). We performed a Local Binary Pattern (LBP) analysis to extract cervical epithelial texture features and quantified blood contrast using the Michelson contrast (MC). A qualitative assessment was conducted by five expert colposcopists using a four-level scale. The mean LBP distance between lesion and normal tissue showed a marked increase from 0.192 to 0.391. The mean MC for blood area and surrounding tissue improved from 0.216 to 0.290. Experts observed an improvement in image quality for all CIN grades, with an average score increase of + 0.71. These initial experimental results demonstrate the potential of the proposed framework to enhance the visualization of CIN, support better treatment decisions, and enable more personalized care.