Nearest Neighbor-Based Strategy to Optimize Multi-View Triplet Network for Classification of Small-Sample Medical Imaging Data.

Multi-view classification with limited sample size and data augmentation is a very common machine learning (ML) problem in medicine. With limited data, a triplet network approach for two-stage representation learning has been proposed. However, effective training and verifying the features from the representation network for their suitability in subsequent classifiers are still unsolved problems. Although typical distance-based metrics for the training capture the overall class separability of the features, the performance according to these metrics does not always lead to an optimal classification. Consequently, an exhaustive tuning with all feature-classifier combinations is required to search for the best end result. To overcome this challenge, we developed a novel nearest-neighbor (NN) validation strategy based on the triplet metric. This strategy is supported by a theoretical foundation to provide the best selection of the features with a lower bound of the highest end performance. The proposed strategy is a transparent approach to identify whether to improve the features or the classifier. This avoids the need for repeated tuning. Our evaluations on real-world medical imaging tasks (i.e., radiation therapy delivery error prediction and sarcoma survival prediction) show that our strategy is superior to other common deep representation learning baselines [i.e., autoencoder (AE) and softmax]. The strategy addresses the issue of feature's interpretability which enables more holistic feature creation such that the medical experts can focus on specifying relevant data as opposed to tedious feature engineering.

Feature extraction of linear predictors at spectral bands of interest.

Intra-cranial electroencephalograms (EEG) from two patients diagnosed with epilepsy are sampled at 1 kHz, enabling analysis and feature extraction at frequency bands above the gamma range. This study focuses on the extraction of linear features (including autoregressive, autoregressive-moving average and Fourier coefficients) obtained at both low (below 100 Hz) and high (100-500 Hz) bands of the signal spectrum. Comparisons of the performance of each feature are made based on a binary hypothesis test of statistical distributions from inter-ictal and pre-ictal epochs. Results are obtained from pre-ictal time periods as assessed by an expert epileptologist.

Optimisation and data mining techniques for the screening of epileptic patients.

Identifying abnormalities or anomalies by visual inspection on neurophysiologic signals such as ElectroEncephaloGrams (EEGs), is extremely challenging. We propose a novel Multi-Dimensional Time Series (MDTS) classification technique, called Connectivity Support Vector Machines (C-SVMs) that integrates brain connectivity network with SVMs. To alter noise in EEG data, Independent Component Analysis based on the Unbiased Quasi Newton Method was applied. C-SVM achieved 94.8% accuracy classifying subjects compared to 69.4% accuracy with standard SVMs. It suggests that C-SVM can be a rapid, yet accurate, technique for online differentiation between epileptic and normal subjects. It may solve other classification MDTS problems too.