肌張力不全障礙(Dystonia)為一種自發性的運動障礙疾病，患者在發病時會表現出不正常的肌肉痙攣或不正常的姿勢。當局部性手部肌肉發生肌張力不全障礙時, 稱為寫字型手部痙攣症(Writer’s Cramp)。 臨床上， 因為後天性的肌張力不全障礙患者之主要致病機轉尚未完全被了解，導致診斷只能依賴醫師經驗。因此，本研究研究目的為探討病人與正常人大腦連結的異同，並以此建立一個可以輔助臨床進行寫字型手部痙攣症的診斷系統。 本研究收集研究對象為寫字型手部痙攣症(Writer’s Cramp)患者15名與正常受試者15名進行15分鐘內自行數8秒手腕伸展時的腦波訊號。首先利用EEG資料分析患者腦區事件相關電位是否與正常人有所不同，接著，利用動態因果模型(DCM)分析患者與正常人腦區連結異同，之後進行機器學習邏輯斯迴歸分析確立診斷模型。 研究結果顯示，患者各個運動腦區Beta頻帶強度明顯小於正常人，但Theta頻帶強度卻明顯大於正常人。在大腦功能性連結的部分，我們則發現寫字型手部痙攣症患者共有11條連結與正常人有所不同。其中10條為抑制性變高的連結，說明過度的抑制對患者在進行非特定任務時為相當重要的，其中患側前運動皮質區以及對側感覺運動皮質區分別擁有5條異常的連結為較重要的腦區。輔助診斷系統的部分，我們則討論了時間以及頻帶對診斷系統的影響，我們利用較短的時間且將所有頻帶納入統計，分類方法為邏輯斯迴歸分析便可得到最高的準確率，可達到92%。時間的部分，我們僅用了在執行運動任務當時的腦波資料便可得到高準確率，說明了運動當時的神經網路連結比起運動前更為重要。頻帶的部分，我們則認為納入所有頻帶比起僅加入Alpha以及Beta頻帶擁有更高的準確率，這說明了所有頻帶的交互作用對於診斷是否為肌張力不全障礙患者來說重要。我們也發現theta頻帶對於肌張力不全障礙患者來說也有其重要性，可能是由於肌張力不全障礙患者也會產生痛的感覺導致。此外，根據輔助診斷系統判斷特徵連結也發現，與患側前運動皮質區有關的連結也確實為我們用來鑑別是否為病人相當重要的腦區，與我們統計出來的結果相符。未來我們也可進一步探討寫字型手部痙攣症患者好手壞手有何不同以及找出肌張力不全障礙患者特定的大腦連結模型，讓我們可以對於肌張力不全障礙有更多的了解。 ;Dystonia is a neurological movement disorder that abnormal contractions of muscles result in the twisting of fixed postures or muscle spasm. Writer’s cramp is described as a particular form of dystonia that affects only a small group of hand muscles during specific tasks.Despite writer′s cramp is neurological in origin, there are no clinical tools for diagnosis except the observations by experts. Given the neurological origin of writer′s cramp, the aim of this study was to investigate the neuronal alternations in patients with writer′s cramp as compared to normal controls. 30 subjects were recruited for this study (15 Writer’s cramp patients and,15 healthy subjects). They were instructed to do the self-initiated wrist extension repeat task every 8 seconds for 15 mins. During the movement task, 32-channel EEG data (10–20 EEG montage) and 2 channel EMG were measured at a sampling rate of 250 Hz. We studied the event-related potentials and the motor networks using Dynamic Causling Model(DCM) to find the significant differences between groups. Machine learning methods were employed to separate the patients from the controls based on the network features. The statistical results show that the power of beta oscillations was smaller while that of theta oscillations was greater in dystonia patients when compared to the healthy subjects. Regarding the brain network, we found 11 abnormal connections in patients, of which 10 were inhibitory, indicating over-inhibition is important for patients when they perform nonsymptomic task. Furthermore, 5 abnormal connections engaged ipsilateral premotor cortex(PM) and contralateral sensory motor cortex(SM1) indicate their importance. The best classification accuracy is 92% when we used DCM beta features over the peristimilus time of -500 to 2000 ms. In conclusion, the network alternations seen in patients with dystonia can serve as biomarker features that separate patients from the healthy control.
