Entropy based integrated diagnosis for enhanced accuracy and removal of variability in clinical inferences

Background: Medical imaging is a thrust area in clinical diagnosis for internal tissue/cell abnormalities like growth of a tumor. Statistical analysis of these images has enables the use of intelligent virtual vision to eradicate the natural limitation of human vision in terms of 2-dimensional representation only. This, sometimes, leads to variability in diagnosis between classical rule based and statistical diagnosis. Objectives: In this paper, we propose to utilize the concept of entropy calculation for more accurate clinical inferences. Methods: The classical and statistical diagnosis is first represented through a probabilistic data set which is then infused to obtain integrated diagnosis. Relative entropies of these integrated diagnostic results with original classical and statistical results are calculated to make final decision on clinical stage of disease. Results and Conclusions: The proposed technique enhances accuracy of diagnosis towards current stage of disease through medical imaging by eliminating variability in inferences drawn through different approaches or different clinical experts. The approach has been verified through preliminary testing on ultrasound images taken for abnormal tissue growth resulting in a tumor.