Towards Enhanced Accuracy in Medical Diagnostics -- A Technique Utilizing Statistical and Clinical Data Analysis in the Context of Ultrasound Images

Abstract

Biomedical imaging remains as a critical thrust area in the healthcare domain with colossal challenges and opportunities associated with analyzing and intelligent utilization of the enormous amounts of data. This research explores the domain of ultrasound images from the perspective of statistical and clinical data modeling and analytics. We propose a technique towards enhancing diagnostic accuracy of the current state and prediction of future states in the context of anomalous/cancerous tissue growth. The proposed technique utilizes both statistical and clinical data associated with ultrasound scans of the cancerous tissue in developing a probabilistic data model towards enhanced inferences with respect to current state of disease and future estimates of survivability. Preliminary evaluation of the proposed technique shows that combining statistical parametric estimates with clinical rule based prediction yields a more accurate model of disease diagnosis. One of the implications of the proposed technique is the potential of practicing evidence based medicine rooted in factual statistical and clinical data thereby reducing variability with respect to diagnostics and corresponding treatment.