A DenseNet CNN-based liver lesion prediction and classification for future medical diagnosis

Abstract

Liver disease diagnosis is a major medical challenge in developing nations. Every year around 30 billion people face liver failure issues resulting in their death. The past liver abnormality detection models have faced less accuracy and high theory of constraint metrics. The lesion on the liver hasn't been identified clearly with earlier models, so an advanced, efficient, and effective liver disease detection is essential. To overcome the limitations of existing models, this approach proposes a deep liver abnormality detection with DenseNet convolutional neural network (CNN) based deep learning technique. This work collected liver Computed Tomography (CT) scan images from Kaggle dataset for training in the initial stage. The pre-processing has been performed with region-growing segmentation, and training is performed through DenseNet CNN. The real-time test images are collected from Government General Hospital Vijayawada (10,000 samples), verified on proposed DenseNet CNN to diagnose whether the input has a liver lesion. Finally, the results obtained and derived confusion matrix summarizes the performance of the proposed methodology with following metrics of accuracy at 98.34%, sensitivity at 99.72%, recall at 97.84%, throughput at 98.43% and detection rate at 93.41%. The comparison results reveals that the proposed technique attains more accuracy and outperforms the other pioneer methodologies.