Face Digital Image Speckle Correlation (DISC) Application as a Non-Invasive, Accessible Tool to Detect Acoustic Neuroma

Acoustic neuroma, a benign tumor on the vestibulocochlear nerve, often causes hearing loss and facial asymmetry. Each year in the United States, approximately 3,000 individuals are diagnosed with the disease. The traditional method of early diagnosis using an EEG is inaccessible, costly, and uncomfortable. To address these challenges, our project aims to develop a face digital image speckle correlation (DISC) application for iOS to facilitate real-time diagnosis by mapping facial muscle movement through advanced image processing techniques. This approach provides a convenient, accessible, affordable, and non-invasive tool for early diagnosis in patients and doctors, eventually generating better outcomes.<br/><br/>We developed a DISC application that will assist the diagnosis of acoustic neuroma by standardizing and streamlining the process of photo-taking and cropping, which ensures that patients are consistently positioned, aiding doctors in accurately identifying facial asymmetry, thus enhancing diagnostic accuracy and accessibility.<br/><br/>For image processing and heatmap application, a Python script in a Jupyter Notebook managed image acquisition, preprocessing, and speckle correlation analysis. Utilizing OpenCV, we performed edge detection, contour detection, and region of interest extraction. The user interface, developed with Swift and Apple’s CoreML library, applied a mask to ensure correct user positioning during photo capture. Both images were converted to grayscale and downscaled to 10 by 10 pixels for easier processing.<br/><br/>Tests on project members demonstrated the DISC application’s effectiveness in generating heatmaps, accurately identifying facial asymmetry by highlighting stronger vector displacements in areas of facial movement, such as a smirk. Bilinear interpolation, used to calculate displacement at intermediate points within a planar face mesh, highlighted variations in facial movement, indicating potential asymmetry associated with acoustic neuroma.<br/><br/>The DISC application shows significant promise as a preliminary diagnostic tool for acoustic neuroma by effectively detecting facial asymmetry. By integrating Python and OpenCV for image processing with Swift and CoreML for facial segmentation, we have developed a robust and user-friendly platform. The deployment on Azure for OpenCV integration into Xcode to interlink all GUI elements ensures scalability and reliability.<br/><br/>Overall, the DISC application offers a user-friendly, time-efficient, and accurate solution for diagnosing acoustic neuroma through the detection of facial muscle movements by standardizing the photo-taking process and providing real-time analysis. As it assists early detection and prompts medical consultation, the application would ultimately ease healthcare barriers and improve patient outcomes.