CASTS: CLS-Guided Adaptive Sparse Token Selection with Multi-Layer Feature Fusion for Interpretable Auditory Brainstem Response Classification
Authors:
- Zhiyong ZhouZZZhiyong Zhou
- Yin LiuYLYin Liu
- Michael ChesnayeMCMichael Chesnaye
- Feng YangFYFeng Yang
- Tiecheng SongTSTiecheng Song
- Anyong QinAQAnyong Qin
- Fan ZhangFZFan Zhang
- Jinlong WangJWJinlong Wang
Preprint
- 13 September 2026
- preprint
- Published by Elsevier in SSRN Electronic Journal
Abstract
The auditory brainstem response (ABR) is an important objective tool for hearing assessment, but its manual interpretation remains labor-intensive and subject to inter-observer variability. Although deep learning approaches applied to the actual ABR voltage measurements have shown promising performance, their clinical use may be constrained by limited access to raw data, as commercial ABR systems often store recordings in vendor-specific formats and require proprietary software or device-specific export procedures. In contrast, ABR report images are routinely generated in clinical practice and provide a more accessible source for automated analysis. However, existing image-based approaches often struggle to capture long-range waveform dependencies and may be distracted by non-informative backgrounds. To address these challenges, we propose CASTS, an interpretable framework for automated ABR report image classification. CASTS uses a pretrained ViT-B/16 backbone with multi-layer feature fusion to integrate hierarchical waveform-image representations. A classification (CLS)-guided top-k patch selection module is also introduced to identify and aggregate waveform-relevant patch tokens according to their similarity to the CLS token, thereby reducing the influence of background-dominant regions. Evaluated on a public dataset containing 186 ABR report images, CASTS achieved a mean classification accuracy of 94.88% and an AUC of 0.976, outperforming representative convolutional and Transformer-based feature extraction baselines. Ablation studies confirmed the effectiveness of the proposed components. Occlusion-based sensitivity analysis further showed that CASTS primarily focused on clinically meaningful waveform structure, especially wave V, rather than non-informative background areas. These results suggest that CASTS provides an effective and interpretable solution for automated ABR report image analysis.Keywords
This publication has 27 references indexed in Scilit:
- Optimizing auditory brainstem response detection through NSGA-II guided feature selectionNeural Computing and Applications, 2025
- Detection of auditory brainstem response peaks using image processing techniques in infants with normal hearing sensitivityBiomedical Signal Processing and Control, 2023
- Optimising weighted averaging for auditory brainstem response detectionBiomedical Signal Processing and Control, 2023
- Comparison of machine learning models to classify Auditory Brainstem Responses recorded from children with Auditory Processing DisorderComputer Methods and Programs in Biomedicine, 2021
- Objective auditory brainstem response classification using machine learningInternational Journal of Audiology, 2019
- ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design2018
- Objective measures for detecting the auditory brainstem response: comparisons of specificity, sensitivity and detection timeInternational Journal of Audiology, 2018
- Visualizing and Understanding Convolutional Networks2014
- Auditory brainstem response threshold estimation: subjective threshold estimation by experienced clinicians in a computer simulation of the clinical testInternational Journal of Audiology, 2004
- New Techniques of Hearing AssessmentOtolaryngologic Clinics of North America, 1994