Higher-order Feature Extraction Research Articles

A Guide to Annotation of Neurosurgical Intraoperative Video for Machine Learning Analysis and Computer Vision

Computer vision (CV) is a subset of artificial intelligence that performs computations on image or video data, permitting the quantitative analysis of visual information. Common CV tasks that may be relevant to surgeons include image classification, object detection and tracking, and extraction of higher order features. Despite the potential applications of CV to intraoperative video, however, few surgeons describe the use of CV. A primary roadblock in implementing CV is the lack of a clear workflow to create an intraoperative video dataset to which CV can be applied. We report general principles for creating usable surgical video datasets and the result of their applications. Video annotations from cadaveric endoscopic endonasal skull base simulations (n= 20 trials of 1-5 minutes, size= 8 GB) were reviewed by 2 researcher-annotators. An internal, retrospective analysis of workflow for development of the intraoperative video annotations was performed to identify guiding practices. Approximately 34,000 frames of surgical video were annotated. Key considerations in developing annotation workflows include 1) overcoming software and personnel constraints; 2) ensuring adequate storage and access infrastructure; 3) optimization and standardization of annotation protocol; and 4) operationalizing annotated data. Potential tools for use include CVAT (Computer Vision Annotation Tool) and Vott: open-sourced annotation software allowing for local video storage, easy setup, and the use of interpolation. CV techniques can be applied to surgical video, but challenges for novice users may limit adoption. We outline principles in annotation workflow that can mitigate initial challenges groups may have when converting raw video into useable, annotated datasets.

Higher Order Feature Extraction and Selection for Robust Human Gesture Recognition using CSI of COTS Wi-Fi Devices.

Device-free human gesture recognition (HGR) using commercial off the shelf (COTS) Wi-Fi devices has gained attention with recent advances in wireless technology. HGR recognizes the human activity performed, by capturing the reflections of Wi-Fi signals from moving humans and storing them as raw channel state information (CSI) traces. Existing work on HGR applies noise reduction and transformation to pre-process the raw CSI traces. However, these methods fail to capture the non-Gaussian information in the raw CSI data due to its limitation to deal with linear signal representation alone. The proposed higher order statistics-based recognition (HOS-Re) model extracts higher order statistical (HOS) features from raw CSI traces and selects a robust feature subset for the recognition task. HOS-Re addresses the limitations in the existing methods, by extracting third order cumulant features that maximizes the recognition accuracy. Subsequently, feature selection methods derived from information theory construct a robust and highly informative feature subset, fed as input to the multilevel support vector machine (SVM) classifier in order to measure the performance. The proposed methodology is validated using a public database SignFi, consisting of 276 gestures with 8280 gesture instances, out of which 5520 are from the laboratory and 2760 from the home environment using a 10 × 5 cross-validation. HOS-Re achieved an average recognition accuracy of 97.84%, 98.26% and 96.34% for the lab, home and lab + home environment respectively. The average recognition accuracy for 150 sign gestures with 7500 instances, collected from five different users was 96.23% in the laboratory environment.

Multichannel feature selection using information maximization with application to multichannel EEG signals

An algorithm for feature selection based on information optimization is developed. This algorithm performs subspace mapping from multi-channel signals, where network modules (NM) are used to perform the mapping for each of the channels. The algorithm is based on maximizing the mutual information (MI) between input and output units of each NM, and between output units of different NMs. Such formulation leads to substantial redundancy reduction in output units, in addition to extraction of higher order features from input units that exhibit coherence across time and/or space useful in classification problems. A number of experiments were carried to validate the performance of the proposed algorithm with very promising results particularly in the case of multichannel EEG data analysis.

Efficient information theoretic extraction of higher order features for improving neural network-based spam e-mail categorization

A novel approach for spam e-mail filtering is herein considered based on information theoretic extraction of higher order features and the committee machines neural network models. An extensive experimental study is organized, the most extensive so far in the literature, based on widely accepted benchmarking e-mail data sets, comparing the proposed methodology with the Naïve Bayes spam filter as well as with the Boosting tree methodology, the linear models-based classification (classification via regression) and the nonlinear models-based classification using simple neural network models, including Multilayer Perceptrons. Moreover, several feature extraction approaches based on information theory are evaluated, comparing mainly the proposed higher order feature extraction methodology with information theoretic extraction of single features. It is shown that the former outperforms the latter and, moreover, that the proposed information theoretic Boolean features present a remarkably high spam categorization performance compared to that of their analog counterparts. Finally, it is shown that the committee machines mail categorization performance compares very favorably to the other rival methods' performance, including the Bayes spam filter which is the most widely used approach in the e-mail services market.

Multi-Channel Subspace Mapping Using an Information Maximization Criterion

A new hybrid information maximization (HIM) algorithm is derived. This algorithm is able to perform subspace mapping of multi-channel signals, where the input (feature) vector for each of the channels is linearly transformed to an output vector. The algorithm is based on maximizing the mutual information (MI) between input and output sets for each of the channels, and between output sets across channels. Such formulation leads to a substantial redundancy reduction in the output sets, and the extraction of higher order features that exhibit coherence across time and/or space. In this paper, we develop the proposed algorithm and show that it combines efficiently the strengths of two well-known subspace mapping techniques, namely the principal component analysis (PCA) and the canonical correlation analysis (CCA). Unlike CCA, which is limited to two channels, the HIM algorithm can easily be extended to multiple channels. A number of simulations and real experiments are conducted to compare the performance of HIM to that of PCA and CCA.

Role of primary somatic sensory cortex in the categorization of tactile stimuli: effects of lesions.

We lesioned the right primary somatic sensory (S1) cortex in two monkeys trained to categorize the speed of moving tactile stimuli. Animals performed the task by pressing with the right hand one of two target switches to indicate whether the speed of a probe moving across the glabrous skin of the left hand was low or high. Sensory performance was evaluated with psychometric techniques and motor behavior was monitored by measuring the reaction (RT) and movement (MT) times before the experiment and throughout the 60 days after the ablation of SI cortex. After the lesion, there was a slight increase in the RTs but no change in the MTs, indicating that removal of SI cortex did not affect the animals' capacity to detect the stimuli. However, monkeys lost their ability to categorize the stimulus speeds. This effect was observed from the 1st day after the lesion until the end of the study. We conclude that somatosensory areas outside SI can by themselves process tactile information in a limited way and that the extraction of higher-order features that takes place during the categorization task requires the intervention of SI cortex.

A computational structure for preattentive perceptual organization: graphical enumeration and voting methods

Presents an efficient computational structure for preattentive perceptual organization. By perceptual organization the authors refer to the ability of a vision system to organize features detected in images based on viewpoint consistency and other Gestaltic perceptual phenomena. This usually has two components, a primarily bottom up preattentive part and a top down attentive part, with meaningful features emerging in a synergistic fashion from the original set of (very) primitive features. In this work the authors advance a computational structure for preattentive perceptual organization. The authors propose a hierarchical approach, using voting methods to build associations through consensus and relational graphs to represent the organization at each level. The voting method is very efficient in terms of time and space and performs impressively for a wide range of organizations. The graphical representation allows the ready extraction of higher order features, or perceptual tokens, because the relational information is rendered explicit. >