An ensemble statistical evaluation of medical image embedding with SqueezeNet neural network

The trio of data, technology, and man constitutes a formidable tripartite synergy towards enhancing health informatics through data science. This avails state-of-the-arts which employ tools like the interquartile range, probability density function, predictive analytics etc. the opportunity of medical data evaluation for pathology and medical diagnosis purposes. However, such evaluations are seldom carried out on medical image embedding acquired through transfer learning. This study therefore employs the SqueezeNet deep embedder on computerized tomography scan signals for feature extraction of pneumonia attributes. An ensemble statistical tool is used for the evaluation after the feature selection of significant attributes by analysis of variance. To answer research question that seeks to discover most significant data feature, experimental result returns an external data attribute as one with the most discriminative information for pneumonia detection. An interquartile range of 40000 to 240000 with a dispersed probability density function in the second quartile also indicates a positive case of pneumonia medical condition