Data mining process for identification of non-spontaneous saccadic movements in clinical electrooculography

A hybrid unsupervised-Deep learning tandem for electrooculography time series analysis

Medical data are often tricky to get mined for patterns even by the generally demonstrated successful modern methodologies of deep learning. This paper puts forward such a medical classification task, where patient registers of two of the categories are sometimes hard to be distinguished because of samples showing characteristics of both labels in turn in several repetitions of the screening procedure. To this end, the current research appoints a pre-processing clustering step (through self-organizing maps) to group the data based on shape similarity and relabel it accordingly. Subsequently, a deep learning approach (a tandem of convolutional and long short-term memory networks) performs the training classification phase on the ‘cleaned’ samples. The dual methodology was applied for the computational diagnosis of electrooculography tests within spino-cerebral ataxia of type 2. The accuracy obtained for the discrimination into three classes was of 78.24%. The improvement that this duo brings over the deep learner alone does not stem from significantly higher accuracy results when the performance is considered for all classes. The major finding of this combination is that half of the presymptomatic cases were correctly found, in opposition to the single deep model, where this category was sacrificed by the learner in favor of a good accuracy overall. A high accuracy in general is desirable for any medical task, however the correct identification of cases before the symptoms become evident is more important.

Automated Detection of Presymptomatic Conditions in Spinocerebellar Ataxia Type 2 Using Monte Carlo Dropout and Deep Neural Network Techniques with Electrooculogram Signals

Application of deep learning (DL) to the field of healthcare is aiding clinicians to make an accurate diagnosis. DL provides reliable results for image processing and sensor interpretation problems most of the time. However, model uncertainty should also be thoroughly quantified. This paper therefore addresses the employment of Monte Carlo dropout within the DL structure to automatically discriminate presymptomatic signs of spinocerebellar ataxia type 2 in saccadic samples obtained from electrooculograms. The current work goes beyond the common incorporation of this special type of dropout into deep neural networks and uses the uncertainty derived from the validation samples to construct a decision tree at the register level of the patients. The decision tree built from the uncertainty estimates obtained a classification accuracy of 81.18% in automatically discriminating control, presymptomatic and sick classes. This paper proposes a novel method to address both uncertainty quantification and explainability to develop reliable healthcare support systems.

Robust EOG-based saccade recognition using multi-channel blind source deconvolution

Human activity recognition (HAR) is a research hotspot in the field of artificial intelligence and pattern recognition. The electrooculography (EOG)-based HAR system has attracted much attention due to its good realizability and great application potential. Focusing on the signal processing method of the EOG-HAR system, we propose a robust EOG-based saccade recognition using the multi-channel convolutional independent component analysis (ICA) method. To establish frequency-domain observation vectors, short-time Fourier transform (STFT) is used to process time-domain EOG signals by applying the sliding window technique. Subsequently, we apply the joint approximative diagonalization of eigenmatrix (JADE) algorithm to separate the mixed signals and choose the "clean" saccadic source to extract features. To address the problem of permutation ambiguity in a case with a six-channel condition, we developed a constraint direction of arrival (DOA) algorithm that can automatically adjust the order of eye movement sources according to the constraint angle. Recognition experiments of four different saccadic EOG signals (i.e. up, down, left and right) were conducted in a laboratory environment. The average recognition ratios over 13 subjects were 95.66% and 97.33% under the between-subjects test and the within-subjects test, respectively. Compared with "bandpass filtering", "wavelet denoising", "extended infomax algorithm", "frequency-domain JADE algorithm" and "time-domain JADE algorithm, the recognition ratios obtained relative increments of 4.6%, 3.49%, 2.85%, 2.81% and 2.91% (within-subjects test) and 4.91%, 3.43%, 2.21%, 2.24% and 2.28% (between-subjects test), respectively. The experimental results revealed that the proposed algorithm presents robust classification performance in saccadic EOG signal recognition.