A Futuristic Approach to Synaptic Fusion of INN and CNN Architectures for Tissue Classification

Radiology refers to a branch of medical specialization that focuses on image processing for diagnostics. It has seen immense advances in predictive methodologies in the past few decades, with its predictive capacities being mostly accounted for by machine learning and deep learning approaches. However, classical machine learning has shown comparatively poor performance with regard to non-linear, irrelevant, and highly correlated features in data, and has also shown high reliance on proper presentation of data. This paper aims to improve that notion and aid radiological diagnostics using advanced deep learning and image processing techniques, which will eventually result in improved diagnostic accuracy and productivity of radiologists. This experiment efficiently fuses hybrid neural networks by using CNN (Convolutional Neural Network Architecture) and INN (Involutional neural networks) architectures, which was then compared with other pre-trained models, trained and tested on multiple color models such as RGB (Red, Green, Blue), HSV (Hue, Saturation, Value), and YCbCr (Luminance, Chrominance), to get optimal efficiency. Our proposed Convolutional Neural Network architectures, which include two fused hybrid models of INN and CNN, when trained with carefully handcrafted features, outperform other pre-trained models in terms of micro accuracy.