AHOP-FIP: An Adaptive Hybrid Optimized Preprocessing Framework for Uterine Fibroid Detection in Ultrasound Images

Abstract

Ultrasound imaging is essential in the diagnosis of uterine fibroids due to its accessibility, low cost, and radiation-free nature. Clinical interpretation is, however, often degraded by multiplicative speckle noise, less-than-ideal contrast, and non-uniform tissue echogenicity, which are all detrimental to automated analysis. To resolve these inherent issues, this paper proposes AHOP-FIP (Adaptive Hybrid Optimized Preprocessing of Fibroid Images), a structured preprocessing pipeline designed to ensure maximal fibroid visibility and preservation of diagnostically important structural and textural data. The two synergistic methods Speckle Reducing Anisotropic Diffusion (SRAD) for edge-aware denoising and Contrast-Limited Adaptive Histogram Equalization (CLAHE) for localized contrast amplification are jointly coupled. Compared to standalone CLAHE, SRAD and Wavelet Denoising on an open-source Kaggle fibroid dataset, AHOP-FIP achieves 35.5 dB PSNR, 0.93 SSIM and 95.7% classification accuracy using ResNet-50 classifier—it beats the baselines in all measures.