A High-Accuracy and Energy-Efficient Zero-Shot-Retraining Seizure-Detection Processor with Hybrid-Feature-Driven Adaptive Processing and Learning-Based Adaptive Channel Selection

Hits:

Indexed by:Essay collection

First Author:J. Liu

Correspondence Author:C. Wang*, and J. Zhou*

Co-author:X. Liu, X Wang, Z. Xie, Z. Zhong, J. Fan, H. Qiu, Y. Xu, H. Qin, Y. Long, Y. Zhou, Z. Shen, L. Zhou, L. Chang, S. Liu, S. Lin,

Journal:IEEE International Solid-State Circuits Conference (ISSCC 2024)

Included Journals:EI

Discipline:Engineering

First-Level Discipline:Electronic Science And Technology

Document Type:C

Date of Publication:2024-02-17

Abstract:Seizure detection processors using machine learning have been proposed to detect the seizure onset of patients for alert or stimulation purpose [1-4]. The existing designs can achieve high accuracy when large amount of seizure data from the patient is available for the training of classifier. However, unlike the collection of non-seizure data, the collection of seizure data with low occurrence requires the patients to undergo time-consuming and costly hospitalization, which is difficult in practice. To address this issue, recently [5] proposed the first zero-shot-retraining seizure detection processor achieving relatively high accuracy without seizure data from the patient for retraining (the zero-shot here means zero seizure data following [5]). Instead, only 2-minute non-seizure data from the patient is required to calibrate the clustered features extracted with a neural network (NN) pre-trained on the public seizure dataset. Although this addressed the aforementioned issue, the accuracy (sensitivity 90.3% & specificity 93.6%) of this design is still limited for practical use, and the energy consumption is large for wearable EEG monitoring devices like other seizure detection processors using NN, as shown in Fig. 1. In this work, we propose a zero-shot-retraining seizure detection processor requiring no seizure data from the patient for retraining as [5] but with much higher accuracy and energy efficiency. It has two major features: 1) A hybrid-feature-driven adaptive processing architecture with on-chip learning requiring no seizure data from the patient is proposed to achieve ultra-low energy consumption and high accuracy. 2) A learning-based adaptive channel selection technique is proposed to further reduce the energy consumption while maintaining high accuracy.