https://ir.lib.ncu.edu.tw/handle/987654321/92945 Search the Channel s https://ir.lib.ncu.edu.tw/simple-search https://ir.lib.ncu.edu.tw/handle/987654321/99393 title: 使用人工智慧在不使用參考物體下測量大腸息肉大小;Using artificial intelligence measuring colon polyp size without a reference object abstract: 背景 息肉尺寸為決定大腸鏡追蹤間隔之關鍵因素之一。本研究提出一種無需參考物即可於大腸鏡檢查過程中量測息肉尺寸大小的人工智慧模型。 材料與方法 本研究建立之息肉尺寸估測迴歸模型，係以兩個獨立之 SegFormer 模型輸出為基礎，其中一者用於息肉分割；另一者用於深度估測。模型初期以仿體在模擬大腸模型進行訓練，隨後透過 1,304 張臨床影像進行遷移式學習。測試階段使用獨立於訓練集之外的52個息肉，共178張影像。以套圈法作為尺寸比較之真實值，採用Olympus 290系列大腸鏡（視野角170°）進行拍攝。息肉依尺寸分為三組：≦5mm、5–10mm 與 ≧10mm。統計分析包括誤差率分析、召回率、精確度、Bland-Altman分析圖、配對 t 檢定及 Cohen′s kappa，以評估套圈法與AI 模型之間在量化測量與分類一致性上的表現。 結果 於測試模型中，AI 模型在三種息肉尺寸組別之誤差率分別為 10.74%、12.36% 與 9.89%，平均誤差率為 11.47%。整體召回率為 0.846；三組息肉尺寸的精確度分別為 0.870、0.911 與 0.857，平均精確度為 0.879；整體 F1 分數為 0.861。Cohen′s kappa 值為 0.792，顯示兩種方法間具高度一致性。Bland-Altman 分析顯示兩方法間之平均偏差為 -0.03 mm，一致性界限介於 -1.654mm 至 1.596mm。 結論 本研究所建立之人工智慧模型可於結腸鏡檢查中，在無需參考物的情況下，準確量測大腸息肉尺寸，具臨床應用潛力與發展前景。 ;Background Polyp size is one of the key factors in determining colonoscopy surveillance intervals. We present an artificial intelligence model for colon polyp size measurement that does not require a reference object during a colonoscopy. Materials and Methods The regression model for polyp size estimation was developed using the outputs from two independent SegFormer models, one for polyp segmentation and the other for depth estimation. Initially, colonoscopic images of polyp phantoms were used to build the model, followed by transfer learning on 1,304 real-world images. For model testing, 178 polyp images from 52 polyps, independent of the training set, were evaluated. A snare was used as the ground truth for size comparison with the AI-based model. Olympus 290 series colonoscope with field of view (FOV) of 170 angle was used in this study. Polyps were categorized into three size groups: ≦5 mm, 5–10 mm, and ≧10 mm. Statistical analysis include error rate analysis, recall, precision, Bland–Altman plot, paired t-test, and Cohen′s kappa to evaluate both the quantitative agreement and categorical consistency between the snare method and AI-based model. Results The error rates for the snare method and the AI-based model across the three polyp size groups in testing model were 10.74%, 12.36%, and 9.89%, respectively, with an average error rate of 11.47%. The overall recall was 0.846, and precision rates for the three size groups were 0.870, 0.911, and 0.857, resulting in an average precision of 0.879. The overall F1 score was 0.861. Cohen′s kappa value between the two methods across the three groups was 0.792. Bland-Altman analysis showed a mean bias of -0.03 mm between the two methods, with limits of agreement, from -1.654 mm to 1.596 mm. Conclusion Our AI-based model shows promise as an accurate tool for colorectal polyp size measurement without the need for a reference object during screening colonoscopy. <br> https://ir.lib.ncu.edu.tw/handle/987654321/98223 title: The Characteristics and Neurophysiological Correlates of Statistical Learning of Positional Regularities in Readers of Different Orthographies abstract: 統計學習（Statistical Learning, SL）是指從環境中提取規律的能力，這種 能力與智力或工作記憶等其他基本認知能力有所不同，本研究聚焦探討其本質和特 性。過去的研究文獻已顯示SL與語言習得及閱讀能力有關，但大多數研究僅使用同 一種SL 測驗派典，主要評估個體對視覺或聽覺刺激中時間規律性的敏感度，這與掌 握語言中的語音與構詞面向較為相關。相較之下，過往研究對空間規律性的敏感度 ——例如對意符文字（如中文）中筆畫在不同空間位置的排列一致性——卻仍少有 探討。 為填補這個研究方向的空缺，本博士論文針對時間位置與空間位置這兩種規 律性，採用平行的 SL 作業，並探討語言背景如何影響這些作業中所測量到的 SL 能 力。具體而言，本博士論文利用本人於碩士論文中所發展、創新的視覺位置 SL (positional SL，PSL) 測驗，其中使用幾何圖形作為刺激材料，分別呈現時間與空 間位置的規律性，測量受試者對這兩種規律性的敏感度。本人過往的研究結果顯示， 無論是母語為拼音文字或意符文字的讀者，都能從兩種 PSL 測驗中學習到規律。不 過，在拼音文字讀者身上，只有空間 PSL 的表現能預測其早期中文識字能力。同時， 腦電波實驗也紀錄到意符文字讀者的大腦所呈現的事件相關電位 (event-related potentials，ERPs)，對於違背時間規律性之刺激材料在 vMMN 與 N400 成分的反應。 然而，這兩個創新SL 測驗的信度與所測量之能力的特異性，以及在拼音文字讀者身 上所引發的電生理反應，尚未被系統性的檢驗。有鑑於此，本博士論文假設意符文 字讀者和拼音文字讀者對時間規律性有相近的敏感度，但前者較後者有更佳的空間 規律性敏感度；在腦電波資料中，本人也預期意符文字讀者和拼音文字讀者在時間 規律性PSL作業中有類似的表現，但在空間規律性PSL作業中的表現則有所不同。 i 此外，儘管過去研究指出人們能同時學習多種規律，但對於時間與空間位置 規律同時呈現在視覺、聽覺感官頻道的情況下，受試者的學習表現尚未被實驗檢視。 為了比較受試者在不同感官頻道的時間、空間規律性 PSL 作業中之表現異同，本博 士論文也使用和視覺PSL作業難度相若的聽覺PSL作業。來自這些作業的實驗結果可 以反映在透過視覺管道學會閱讀之前，聽覺處理歷程在語言發展早期所扮演的角色。 本人假設當時間規律性和空間規律性同時呈現時，受試者可以學會兩種規律，另外 也假設受試者對於在視覺管道中所展現的空間規律性和在聽覺管道中所展現的時間 規律性，會有更好的敏感度。 為了達到以上的目標，本博士論文設計了三組實驗。首先，透過比較拼音文 字與意符文字讀者在兩種 PSL 作業中的表現，確認了這兩種測驗的信度與特異性， 並發現意符文字讀者對空間位置規律的敏感度高於對時間規律的敏感度，顯示閱讀 經驗會影響 SL 表現。本論文的第二組實驗，透過 ERPs 的成分（如 vMMN、N400 與 P600）與跨嘗試間相關性分析（inter-trial coherence，ITC），確認了不同語言背 景的讀者在學習時間與空間位置規律時所涉及的神經處理機制有所不同。本論文的 第三組實驗，評估了當時間與空間的位置規律性，以交錯方式或同時呈現在視覺與 聽覺兩種感官頻道中，受試者在SL作業中的表現。這組實驗的結果顯示，無論感官 頻道為何，受試者普遍對空間位置規律較時間位置規律更為敏感。 總結而言，本博士論文透過行為實驗與神經電生理證據，確認了人類對時間 與空間位置規律的統計學習能力並不相同，並指出語言經驗（特別是所閱讀的文字 系統）對統計學習表現的關鍵影響。未來的研究可進一步探討不同類型規律敏感度 的細緻差異，以及語言背景在統計學習機制中所扮演的角色。;This study investigates the nature of statistical learning (SL), defined as the ability to extract regularities from the environment, which is distinct from core cognitive abilities such as IQ and working memory. Previous research has demonstrated that SL correlates with language acquisition and reading proficiency. However, this relationship has been predominantly examined using a single SL paradigm that measures sensitivity to temporal contingencies within a stream of visual or auditory stimuli, paralleling aspects of language such as phonology and morphology. In contrast, sensitivity to spatial contingencies, analogous to the consistency of stroke patterns at different spatial positions within the characters of logographic writing systems (e.g., Chinese), remains unexplored. This study aims to address these gaps by differentiating between sensitivity to temporal and spatial regularities and examining how language backgrounds influence SL. To this end, I introduced two novel visual SL paradigms developed in my master′s research: the temporal and spatial positional SL (PSL) tests. These paradigms involve nonverbal geometric shapes that exhibit regularities in either temporal or spatial positions. My earlier work revealed that both alphabetic and logographic readers could extract regularities from both PSL formats, with no difference in both temporal and spatial PSL tests between groups. Critically, among alphabetic readers, only spatial PSL performance predicted early Chinese orthography acquisition. ERP recordings in logographic readers further demonstrated visual mismatch negativity (vMMN) and N400 effects in response to violations of spatial and temporal regularities, respectively. Despite these findings, the reliability and specificity of PSL tests and their associated ERP correlates in alphabetic readers remain unexamined. In this dissertation, it was hypothesized that sensitivity to spatial regularities would be more pronounced in logographic than alphabetic readers, whereas sensitivity to temporal regularities would be comparable between alphabetic and logographic readers. Similarly, it was also hypothesized that the EEG responses from both groups would be similar in the temporal PSL test, while it would be distinct in the spatial PSL test. Furthermore, while previous studies have shown that SL of multiple regularities simultaneously is possible, participants′ performance on learning simultaneously presented temporal and spatial regularities has yet to be tested. To compare participants’ PSL of temporal and spatial regularities across different modalities, comparable tests in the auditory modality were also included. Such findings may reflect the role of auditory processing in early language development before learning to read via the visual modality. It was hypothesized that when temporal and spatial regularities are presented simultaneously, participants would be able to learn both regularities. In addition, participants would exhibit greater sensitivity to spatial and temporal regularities in the visual and auditory modality, respectively. To fulfill the aims outlined above, my Ph.D. research comprises three series of experiments. First, I established the reliability and specificity of temporal and spatial PSL abilities across alphabetic and logographic readers. The results further showed that reading experience influences PSL performance, with logographic readers exhibiting greater sensitivity to spatial regularities than alphabetic readers. Second, I identified the neural correlates of temporal and spatial PSL across different reader groups through ERP components (i.e., vMMN, N400, and P600) and inter-trial coherence (ITC) analyses. These findings provide evidence for distinct cognitive processes underlying temporal and spatial PSL, which were further modulated by language backgrounds. Third, I demonstrated participants′ temporal and spatial PSL when to-be-learned regularities are presented in either interleaved or concurrent formats across both visual and auditory modalities. Participants also showed greater sensitivity to spatial than to temporal regularities in both modalities. In summary, my Ph.D. research underscores the robustness of separate temporal and spatial PSL capabilities through validated experimental designs and distinct neurophysiological correlates. The findings also highlight the influence of reading experience on SL performance. Future research should focus on sophisticated explorations of sensitivity to distinct regularity types and the impact of language background within SL paradigms. <br> https://ir.lib.ncu.edu.tw/handle/987654321/95485 title: 從同步鼾聲聲學分析和睡眠動態核磁共振成像進行靜態顱面測量和動態上呼吸道塌陷觀察，並探討其與阻塞性睡眠呼吸中止症嚴重程度的關聯。;Statis Craniofacial Measurement and Dynamic Upper Airway Collapse Observed from Synchronized snoring acoustic analysis and Sleep MRI and Its Association with Severity of Obstructive Sleep Apnea abstract: 阻塞性睡眠呼吸中止症是一種睡眠呼吸障礙，其特徵是在睡眠期間上呼吸道反覆發生阻塞性塌陷，這會因長期缺氧導致嚴重的心血管和神經認知等相關後遺症。多項生理睡眠檢查是阻塞性睡眠呼吸中止的標準診斷工具，但它不能提供關於上呼吸道塌陷部位和型態的信息。對於接受陽壓呼吸器治療的患者來說，呼吸道塌陷的位置和型態並不重要，因為陽壓呼吸器是對整個塌陷的上呼吸道施加一個壓力支撐呼吸道避免呼吸道阻塞。然而，有一大部分患者面對陽壓呼吸器治療的依從性仍然是一個很大的挑戰，他們經常因為無法適應陽壓呼吸器轉而希望採取外科手術治療。對於外科醫生，識別上呼吸道塌陷部位變得至關重要。更好地理解動態上呼吸道塌陷的位置和型態可以讓外科醫師可以更精準化及個別化的製作符合患者特點及需求的外科手術計畫。目前各種影像學方法和清醒時上呼吸道內視鏡檢查方法已被用於預測氣道塌陷的情形。這些檢查的缺點在於患者在清醒或靜態狀態下並不能完全表現出呼吸道阻塞的位置和型態。因此，若要患者在睡眠時進行成像且不使用藥物鎮靜，那麼動態睡眠核磁共振檢查是一種非常好的動態上呼吸道塌陷研究工具。本研究希望通過動態睡眠核磁共振檢查，並在軸位、矢狀位和冠狀位進行分析，我們透過靜態顱面測量和動態呼吸道塌陷觀察探討其與疾病嚴重程度的相關性。本研究使用睡眠動態核磁共振成像，並同步進行打呼聲錄音以觀察輕度和嚴重阻塞性睡眠呼吸中止症患者的動態呼吸道塌陷的差別。核磁共振中線矢狀影像用於靜態顱面測量。我們從參與睡眠動態核磁共振研究的參與者中選擇了15名男性重度阻塞性睡眠呼吸中止症患者（平均呼吸中止低通氣指數70.3 ± 23次/小時）與15名輕度阻塞性睡眠呼吸中止症患者（平均呼吸中止低通氣指數7.8 ± 1.4次/小時），並與其年齡和身體質量指數進行了配對。靜態顱面測量包括上下頜關係和氣道長度等相關測量。通過盲檢評估軸位、矢狀面和冠狀面觀察軟顎後、舌根後和側咽壁的動態塌陷情形。使用雙變量分析將這些測量與阻塞性睡眠呼吸中止症的疾病嚴重程度（呼吸中止低通氣指數）進行相關分析，統計學上顯著性設置為P < 0.01。結果顯示來自睡眠動態核磁共振成像顯示側咽壁塌陷（b = 51.8，P < 0.001）和來自靜態顱面測量的上呼吸道氣道長度（b = 27.2，P < 0.001）與阻塞性睡眠呼吸中止症的疾病嚴重程度呈顯著正相關。研究結果表明側咽壁塌陷和上呼吸道道長度確實與阻塞性睡眠呼吸中止症嚴重程度有顯著關聯。這些標記可轉化為阻塞性睡眠呼吸中止症患者的常規臨床評估，用於評估個案的嚴重程度及治療的個別化規劃的依據。此研究提供了有關影響阻塞性睡眠呼吸中止症嚴重程度的靜態顱面測量和動態呼吸道塌陷因素的寶貴見解。;Obstructive sleep apnea (OSA), a condition characterized by repeated episodes where the upper airway collapses during sleep. This collapse interrupts normal breathing and is associated with significant health risks, including heart and brain-related issues. The typical method for diagnosing OSA is polysomnography , which effectively identifies sleep disruptions but doesn′t pinpoint where or how the airway collapses. While continuous positive airway pressure therapy for OSA involves using a device that delivers positive pressure to keep the airway open during sleep, many patients find this uncomfortable and do not consistently use it. This noncompliance often leads patients to consider surgical alternatives, which necessitates precise identification of the exact locations of airway collapse to tailor the surgical approach. Traditional imaging techniques performed while the patient is awake or using static images are not very effective because they fail to capture the airway′s behavior during sleep, when OSA actually occurs. As a result, sleep MRI, performed without sedating the patient, emerges as a superior method. It allows researchers to observe the airway dynamically as it behaves naturally during sleep, providing valuable insights into where and how the airway collapses. This study involved analyzing 15 male subjects with severe OSA, who had an average Apnea-Hypopnea Index (AHI) of 70.3 events per hour, and comparing them to 15 males with mild OSA, who had an average AHI of 7.8 events per hour. Both groups were matched by age and body mass index . We used MRI to measure static craniofacial structures and observe the dynamic airway collapse in the retropalatal (RP), retroglossal (RG), and lateral pharyngeal wall (LPW) areas, in various views (axial, sagittal, and coronal). These measurements were then correlated with OSA severity using AHI, and statistical significance was set at P < 0.01. The study′s findings demonstrated that collapse at the lateral pharyngeal wall (b = 51.8, P < 0.001) and the length of the upper airway (b = 27.2, P < 0.001) were significantly associated with severity of OSA. These results suggest that specific features observed through sleep MRI, like the extent of lateral pharyngeal wall collapse and the length of the upper airway, could be crucial markers for assessing the severity of OSA. Incorporating these MRI findings into routine clinical evaluations could help refine the approach to treatment, particularly surgical interventions, leading to more customized and effective care for patients with OSA. The study underscores the importance of understanding both the anatomical and dynamic aspects of the upper airway in managing OSA severity. <br> https://ir.lib.ncu.edu.tw/handle/987654321/93088 title: 夢境清醒度與注意力及現實監控的正向關聯性： 行為與電生理之研究;Dream lucidity positively correlates with attention and reality monitoring: behavioral and EEG study. abstract: 人在夢境中具有不同程度的意識或覺察力。通常在一般夢中，作夢者對自己所處的狀態沒有覺察力，認為自己所處的環境是現實世界;然而當作夢者處在清醒夢(Lucid dream)中，便可能具有邏輯思考、自主控制夢的能力，甚至可以提取清醒時經歷過之事件的相關記憶。這種對於『現實與夢境差異的覺察力』，我們將之稱為「夢境清醒度 (Dream lucidity)」。透過腦造影的研究可以發現，當作夢者處在清醒夢中時，與執行功能相關的前額葉皮質區，會比作夢者處在非清醒夢中時更加被活化。此外，夢境清醒度高的人，能夠覺察夢境中的狀態，與清醒狀態之間的差異，這樣的覺察力與「現實監控 (reality monitoring)」也就是「區別經驗事件屬於內部產生或是外部經歷」的能力相似，而執行功能也在現實監控中扮演不可或缺的腳色，因此推測夢境清醒度的高低與否可能與某些認知功能相關，然而哪些認知功能跟夢境清醒度之間具有關聯性為何目前仍不清楚。因此本研究使用行為與腦波實驗檢視與夢境清醒度相關的認知功能，並探討夢境清醒度與現實監控的表現相關性為何。 實驗一針對夢境清醒度與四種認知功能作業之間的關聯性進行探討，其中包含注意力網絡作業 (attentional network task)、作業轉換作業 (task-switching task)、多數函數作業 (majority function task)、以及運作廣度作業 (operational span)。結果顯示，處在警覺狀態 (alert state)下的受試者，其夢境清醒度與注意力網絡作業中的衝突解決 (conflict resolution)具有正向相關性，也與利用多數函數作業所估算出訊息複雜度處理反應時間函數的截距有負相關。實驗二透過行為以及腦波研究，探討夢境清醒度與現實監控的表現之間的關聯性。結果顯示，夢境清醒度高的受試者，在現實監控的表現上比較好，特別是更能夠清楚判斷『曾經看過的物件』的來源。實驗三更進一步探討，夢境清醒度高的受試者如何透過記憶特徵(memory characteristics)分辨不同來源的物件。結果顯示，夢境清醒度高的受試者，針對想像以及曾經看過的物件回報相關的記憶特徵時，所回報感官相關的記憶特徵差異較小。總結以上結果，我們認為夢境清醒度高的受試者，不僅能夠有效率的進行注意力網絡中的衝突解決，且因為在記憶想像與曾經看過的物件之間的感官記憶特徵差異更小，更能夠能夠清楚判斷記憶的來源。 ;Dreams can be associated with different levels of consciousness or awareness. In extreme cases such as lucid dreaming, the dreamer can think logically, voluntarily control the dream’s narrative, and even retrieve memories of events experienced before (Voss et al., 2013). The degree of awareness of this difference between the dream context and reality is a continuous spectrum known as “dream lucidity.” Recent functional neuroimaging studies showed that the activation of the anterior prefrontal cortex (PFC), a region critical for executive function, was greater in lucid REM sleep than non-lucid REM sleep (Dresler et al., 2012). In addition, dream lucidity is related to the discrimination between self-generated precepts in a dream and externally derived experiences when awake, a capacity resembles “reality monitoring” in which executive function is also involved. Thus, we hypothesized that cognitive functions are indispensable in dream lucidity. However, whether and how cognitive functions correlate with dream lucidity remains unclear. The current thesis employed behavioral and event-related brain potential (ERP) experiments to identify the components of cognitive functions related to dreaming lucidity and determine how dream lucidity correlates with reality monitoring performance. The first experiment was carried out to examine the correlation between dream lucidity and cognitive functions revealed by the attentional network task, the task-switching task, the majority function task, and the operational span task. Results showed that dream lucidity positively correlated with the performance of conflict resolution in the attentional network task for participants with positive alerting scores. The second experiment examined the correlation between dream lucidity and reality monitoring performance in behavioral and EEG recording aspects. Results showed that participants with high dream lucidity have a greater reality monitoring performance, especially in the source judgment of perceived items. The third experiment further examined whether and how the participant with high dream lucidity differentiated different sources of items based on memory characteristics compared to participants with low dream lucidity. Results showed that people with high dream lucidity tended to report a smaller difference between the sensory features of their memories of perceived and imagined events. These findings suggest that the greater a person’s dream lucidity, the more efficiently he or she can resolve incongruent conditions and differentiate the memory sources because their memories of imagined and actual events correspond to a smaller difference in sensory features. <br>