State-of-the-art review on deep learning in medical imaging.

Deep learning (DL) is affecting each and every sphere of public and private lives and becoming a tool for daily use. The power of DL lies in the fact that it tries to imitate the activities of neurons in the neocortex of human brain where the thought process takes place. Therefore, like the brain, it tries to learn and recognize patterns in the form of digital images. This power is built on the depth of many layers of computing neurons backed by high power processors and graphics processing units (GPUs) easily available today. In the current scenario, we have provided detailed survey of various types of DL systems available today, and specifically, we have concentrated our efforts on current applications of DL in medical imaging. We have also focused our efforts on explaining the readers the rapid transition of technology from machine learning to DL and have tried our best in reasoning this paradigm shift. Further, a detailed analysis of complexities involved in this shift and possible benefits accrued by the users and developers.

Symtosis: A liver ultrasound tissue characterization and risk stratification in optimized deep learning paradigm.

Background and Objective Fatty Liver Disease (FLD) - a disease caused by deposition of fat in liver cells, is predecessor to terminal diseases such as liver cancer. The machine learning (ML) techniques applied for FLD detection and risk stratification using ultrasound (US) have limitations in computing tissue characterization features, thereby limiting the accuracy. Methods Under the class of Symtosis for FLD detection and risk stratification, this study presents a Deep Learning (DL)-based paradigm that computes nearly seven million weights per image when passed through a 22 layered neural network during the cross-validation (training and testing) paradigm. The DL architecture consists of cascaded layers of operations such as: convolution, pooling, rectified linear unit, dropout and a special block called inception model that provides speed and efficiency. All data analysis is performed in optimized tissue region, obtained by removing background information. We benchmark the DL system against the conventional ML protocols: support vector machine (SVM) and extreme learning machine (ELM). Results The liver US data consists of 63 patients (27 normal/36 abnormal). Using the K10 cross-validation protocol (90% training and 10% testing), the detection and risk stratification accuracies are: 82%, 92% and 100% for SVM, ELM and DL systems, respectively. The corresponding area under the curve is: 0.79, 0.92 and 1.0, respectively. We further validate our DL system using two class biometric facial data that yields an accuracy of 99%. Conclusion DL system shows a superior performance for liver detection and risk stratification compared to conventional machine learning systems: SVM and ELM.