Image-based automated measurement model to predict pelvic organ prolapse

A tangible non-transitory computer readable storage media having computer-executable instructions for a computer processor to perform a method for predicting or diagnosing pelvic organ prolapse, the instructions comprising:

• receiving an image of a patient from a magnetic resonance imaging (MRI) procedure performed around a pelvic area of the patient;
• automatically identifying a location of a bladder and a rectum of the patient within the MRI image based on their intensity values displayed on the MRI image;
• automatically identifying a pubic bone of the patient within said MRI image, wherein the step of identifying the pubic bone comprises:
• automatically extracting and identifying keypoints on the pubic bone and on a pelvic floor on the MRI image based on a spatial relationship between the pelvic organs and pubic bone structures,
• automatically generating candidate bounding boxes substantially around the pubic bone structures,
• automatically extracting features within the bounding boxes, and
• classifying the candidate bounding boxes using an automated classifier having a learning algorithm that analyzes data used for classification analysis;
• wherein the classifier is constructed based on a training data set using a support vector machine (SVM) trained using a radial basis function kernel providing for non-linear decision surface;
• wherein the SVM-based classifier classifies each candidate bounding box using Haar-like and texture features, each candidate bounding box being classified as an enclosed pubic bone region and a partially enclosed pubic bone region;
• automatically identifying a coccyx and a sacral promontory of the patient within the MRI image, wherein the step of identifying the coccyx and the sacral promontory comprises:
• wherein using locations of the bladder, the rectum, and the pubic bone structures to localize the coccyx and the sacral promontory via a non-linear regression;
• and
• automatically linking the extracted features of the pubic bone structures and the localized coccyx and sacral promontory with clinical information to predict or diagnose pelvic organ prolapse.