Advancements in Image Enhancement for Improved Knee Ultrasound Segmentation Towards Measurement of Lateral Patellar Tendon Displacement

Patellofemoral Pain Syndrome (PFPS), characterized by anterior knee pain around the patella, requires accurate diagnosis for effective management. Traditional methods rely on clinician expertise, which can be subjective and time-consuming. This study introduces an automated system to measure lateral patellar tendon displacement, specifically the Patellar Tendon-Trochlear Groove (PT-TG) distance, using preprocessed ultrasound images to diagnose PFPS. While Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) offer high-resolution imaging, they are costly and carry risks such as radiation. Ultrasound imaging, on the other hand, is non-invasive, cost-effective, and convenient, allowing for repeated examinations. However, ultrasound images often suffer from low contrast, blurred edges, speckle noise, and shadows, complicating accurate segmentation and analysis. Therefore, image preprocessing is crucial to enhance ultrasound image quality to further extract the medical values embedded within the images. This study applies Contrast Limited Adaptive Histogram Equalization (CLAHE) and Gamma Correction for contrast enhancement, along with Rank Order Filter for speckle noise reduction. Data were collected from the Research Center for the Health Sciences at the University of Santo Tomas, involving 26 participants, 12 with PFPS and 14 without. Results indicate that these preprocessing algorithms significantly enhance image quality, leading to better segmentation of the patellar tendon and trochlear sulcus, and more accurate PT-TG distance measurement. The error rate in measuring lateral patellar tendon displacement decreased from 54.13% to 25.91% post-preprocessing. The proposed algorithm shows potential in aiding PFPS diagnosis, reducing the diagnostic burden on clinicians, and potentially improving patient outcomes.