基於語音特徵之失智症篩檢方法

A Method for Dementia Screening Based on Speech Features

劉子維(Tz-Wei Liou)

指導教授 ： 鄭士康

國立臺灣大學/電機資訊學院/電機工程學研究所/碩士(2019年)

https://doi.org/10.6342/NTU201903392

摘要

失智症人口逐年攀升，影響到許多家庭及醫療系統。患者須及早治療才能延緩其惡化的速度，而傳統的診斷方法需要依賴專業醫護人員的經驗且費時，無法應付龐大的需求。因此我們提出一個以語句中的語音特徵做為輸入的失智辨識模型，來解決這樣的問題。我們將Mandarin_Lu及Aishell語音庫組成訓練資料，並嘗試解決資料不平衡的問題，以長短期記憶神經網路(Long-short term memory)作為基礎，訓練出一個失智辨識模型，在兩種語音庫組成的測試集上達到98%的準確率。

關鍵字

失智症 ； 深度學習 ； 語音處理

並列摘要

The population of people who suffer from dementia is increasing year by year. Early detection of dementia is very important for the patient. However, traditional testing tools rely on experienced doctor to conduct and take long time. Therefore, this thesis presents an approach to detect dementia automatically through acoustic features. The proposed method employs LSTM recurrent neural network on MFCC features extracted from spoken utterances to build a predictive model. We train the model on utterances which come from two speech corpora (Mandarin_Lu and Aishell) and deal with imbalanced data. Our model achieves an accuracy of 98% on test set.

並列關鍵字

Dementia ； Deep Learning ； Speech Processing

參考文獻

[1]S.Sarraf, D. D.Desouza, and J.Anderson, “DeepAD : Alzheimer ’ s Disease Classification via Deep Convolutional Neural Networks using MRI and fMRI,” pp. 1–32, 2016.

[2]A.Payan and G.Montana, “Predicting Alzheimer ’ s disease : a neuroimaging study with 3D convolutional neural networks,” pp. 1–9, 2015.

[3]M.Lu, “Attention-based Deep Multiple Instance Learning for MCI discriminative patch localization and diagnosis on brain MRI,” 國立台灣大學碩士論文, July, 2019.

[4]S.Olubolu, O.Id, J. S.Wong, and C. P.Wong, “Deep language space neural network for classifying mild cognitive impairment and Alzheimer-type dementia,” pp. 1–15, 2018.

[5]T.Niu, M.Bansal, and S.Karlekar, “Detecting Linguistic Characteristics of Alzheimer’s Dementia by Interpreting Neural Models,” 2014.