Wyniki wyszukiwania - BazTech

Ograniczanie wyników

2 Biocybernetics and Biomedical Engineering

Znaleziono wyników: 2

Liczba wyników na stronie

Wyniki wyszukiwania

Sortuj według:

Ogranicz wyniki do:

Surgical phase classification and operative skill assessment through spatial context aware CNNs and time-invariant feature extracting autoencoders

Pradeep Chakka Sai, Sinha Neelam

Biocybernetics and Biomedical Engineering

2023

Vol. 43, no. 4

700--724

Automated surgical video analysis promises improved healthcare. We propose novel spatial context aware combined loss function for end-to-end Encoder-Decoder training for Surgical Phase Classification (SPC) on laparoscopic cholecystectomy (LC) videos. Proposed loss function leverages on fine-grained class activation maps obtained from fused multi-layer Layer-CAM for supervised learning of SPC, obtaining improved Layer-CAM explanations. Post classification, we introduce graph theory to incorporate known hierarchies of surgical phases. We report peak SPC accuracy of 96.16%, precision of 94.08% and recall of 90.02% on public dataset Cholec80, with 7 phases. Our proposed method utilizes just 73.5% of parameters as against existing state-of-the-art methodology, achieving improvement of 0.5% in accuracy, 1.76% in precision with comparable recall, with an order less standard deviation. We also propose DNN based surgical skill assessment methodology. This approach utilizes surgical phase prediction scores from the final fully-connected layer of spatial-context aware classifier to form multi-channel temporal signal of surgical phases. Time-invariant representation is obtained from this temporal signal through time- and frequency-domain analyses. Autoencoder based time-invariant features are utilized for reconstruction and identification of prominent peaks in dissimilarity curves. We devise a surgical skill measure (SSM) based on spatial-context aware temporal-prominence-of-peaks curve. SSM values are expected to be high when executed skillfully, aligning with expert assessed GOALS metric. We illustrate this trend on Cholec80 and m2cai16-tool datasets, in comparison with GOALS metric. Concurrence in the trend of SSM with respect to GOALS metric is obtained on these test videos, making it a promising step towards automated surgical skill assessment.

Sleep EEG analysis utilizing inter-channel covariance matrices

Gopan K. Gopika, Prabhu Sathvik S., Sinha Neelam

Biocybernetics and Biomedical Engineering

2020

Vol. 40, no. 1

527--545

Background: Sleep is vital for normal body functions as sleep disorders can adversely affect a person. Electroencephalographic (EEG) signals indicate brain functions and have characteristic signatures for various sleep stages. These enable the use of EEG as an effective tool for in-depth studies about sleep. Sleep stages are broadly divided as rapid eye movement (REM) and non-rapid eye movement (NREM). NREM is further divided into 3 stages. The objective of the work is to distinguish the given EEG epoch as wake, NREM1, NREM2, NREM3 and REM. DREAMS Subject Database containing 5 EEG channels is used here. This work focuses on utilizing EEG by exploiting variations in inter-dependencies of different brain regions during sleep. New method: Covariance matrices of the wavelet-decomposed channels are used to obtain the variations in inter-dependencies. The feature sets are: (1) simple matrix properties(MF) like trace, determinant and norm, (2) eigen-values (E1), (3) eigen-vector corresponding to the largest eigen-value (E2) and (4) tangent vectors obtained using Riemann geometry (RG-TS). The features are input to ensemble classifier with bagging. Subject-specific, All-subjects-combined and Leave-one-subject-out methods of analysis are carried out. Results: In all methods of analysis, RG-TS features give maximum accuracy (80.05%, 83.05% and 61.79%), closely followed by E1 (79.49%, 77.14% and 58.34%). Comparison with existing method: The proposed method obtains higher and/or comparable accuracy. This work also ensures no biasing of classifier due to unequal class distribution. Conclusion: The performances of RG-TS and E1 features reveal that the changes in interdependencies of pre-frontal and occipital lobe along with the central lobe can be used to distinguish the different sleep stages.