Facial Expression-based Parkinson's Disease Severity Diagnosis via Feature Fusion and Adaptive Class Balancing
Yintao Zhou, Wei Huang, Zhengyu Li, Jing Huang, Meng Pang
TL;DR
The paper tackles Parkinson's disease (PD) severity diagnosis using facial expressions, addressing limitations of single-expression cues and imbalanced PD-stage distributions. It introduces a two-component method: attention-based feature fusion (AFF) to integrate features from six basic emotions, with a ResNet-18 extractor processing each expression and a fusion mechanism computing $W_{attn} = \sigma( FC( AvgPool(F)) + FC(MaxPool(F)) )$ where $F$ is the concatenated $6d$ feature, producing a fused representation. It also introduces an adaptive class balancing loss $L_{adapt} = -\sum_{i=1}^{C} \alpha_i (1 - p_t)^{\gamma} y_i \log(p_i)$ with $\alpha_i = N_{max}/N_i$ to emphasize minority and hard samples. Experiments on the PDFE PD facial expression dataset (augmented with public non-PD facial-expression datasets) show improved performance over baselines, with ablation studies confirming the effectiveness of AFF and adaptive class balancing.
Abstract
Parkinson's disease (PD) severity diagnosis is crucial for early detecting potential patients and adopting tailored interventions. Diagnosing PD based on facial expression is grounded in PD patients' "masked face" symptom and gains growing interest recently for its convenience and affordability. However, current facial expression-based approaches often rely on single type of expression which can lead to misdiagnosis, and ignore the class imbalance across different PD stages which degrades the prediction performance. Moreover, most existing methods focus on binary classification (i.e., PD / non-PD) rather than diagnosing the severity of PD. To address these issues, we propose a new facial expression-based method for PD severity diagnosis which integrates multiple facial expression features through attention-based feature fusion. Moreover, we mitigate the class imbalance problem via an adaptive class balancing strategy which dynamically adjusts the contribution of training samples based on their class distribution and classification difficulty. Experimental results demonstrate the promising performance of the proposed method for PD severity diagnosis, as well as the efficacy of attention-based feature fusion and adaptive class balancing.