紅外線熱影像顯微游離皮瓣術後監測系統：基於獨立成分分析法之溫度變化偵測

Abstract

目前癌症中發生率與死亡率成長速度最快速的癌症其中之為口腔癌；在口腔癌的治療方面，目前臨床中最常使用的方法是透過外科手術進行病灶的切除，病患在切除病灶後，需要利用病患身體上其他部位的組織來重建其患處區域以保持其美觀性；其中，顯微游離皮瓣手術是多年來可靠又重要重建方法。然而，儘管游離皮瓣重建手術是由富有經驗的顯微手術醫師操刀，仍有少數游離皮瓣會在五至七天內發生血循障礙，且自得知皮瓣發生阻塞到執行再探查的時間間隔越長，其也會增加再次手術的失敗機率，因此如何監控游離皮瓣血液灌流狀況，提早偵測發生血管阻塞時間點，並即時警示醫師執行再探查，或重新手術接合，至關重要！ 目前臨床游離皮瓣的監控與照護，主要是透過護理人員定期執行游離皮瓣的常規檢查，但這不僅需要耗費相當龐大的人力資源，同時也會因為護理人員照護經驗的不同而產生判斷上的差異；而其他監控游離皮瓣血管是否阻塞的方法，不但需要較高的成本，而且也無法提供一套非侵入式與非接觸式，甚至是連續的監控系統；因此，在先期研究中，提出利用紅外線具備低成本、非侵入性、非接觸性、無放射性、快速且可反覆成像的等特性，在不接觸游離皮瓣表面的情況下，測量其組織溫度的變化，是一項重要的起步；且透過因素分析演算法能夠有效的移除人體生理之共同的因素，以觀察到真實的溫度變化資訊。但本研究發現如果存在兩個以上會影響溫度變化的獨立因素時，以目前觀察值的數量及組合方式可能不適用因素分析的方法，因為沒有足夠的資料去能夠解讀更為複雜之情況，且若所估計出之共同因素並非所有共同影響之因素的線性組合，而是先有某項因素後出現某項因素，這樣因素分析演算法就無法將其分離乾淨。 因此本研究基於先期研究的理論基礎，希望利用獨立成分分析演算法能夠解析出所觀察訊號之特有訊號，與具備更多的彈性與更多資訊可以解釋所發生之不同現象之資訊的優勢，且這些因素不容易由因素分析演算法來詮釋，希望透過獨立成分分析演算法這項優勢與能力，發展更適用於紅外線熱影像之游離皮瓣術後監測系統之溫度變化偵測演算法。 目前本研究已進行四次的動物實驗以及於臨床收案共二十三位病患，且皆已通過台北榮民總醫院動物照護及使用委員會(IACUC核准編號：2017-181)與臨床試驗委員會(IRB編號：2016-01-006BC)之審核。其中有一位病患發生靜脈血管阻塞的情況，除經因素分析其紅外線熱影像溫度變化的結果，發現游離皮瓣阻塞之皮瓣溫度有明顯下降的現象已外，經獨立成分分析演算法之結果我們可以發現，有兩項獨立因素可能影響著游離皮瓣區域之溫度變化，且其中一項獨立因素發生溫度下降趨勢的時間點早於因素分析之結果，另一項則同於因素分析之結果，且而此現象與護理紀錄比對顯示，其游離皮瓣溫度變化皆早於護理人員之判斷的可能性，但由於目前阻塞案例太少，因此無法做更進一步的驗證與討論，希望在未來能夠有更多資料進行皮瓣阻塞溫度變化的驗證，並發展出一套提供臨床監測皮瓣血管阻塞的輔助監測工具。

One of the most common cancer with the highest mortality rate is Oral cancer. The most frequently adopted approach of curative treatment of oral cancer is by resecting cancer legions and reconstructing the affected regions by unaffected tissues of the patient. Free flap surgery is a reliable reconstruction method operated by many medical professionals. However, circulatory compromises were sometimes observed within five to seven days after surgeries, even if the operations were performed by experienced microvascular surgeons. Furthermore, research shows that that the success rate of surgeries diminishes due to late detection of circulatory compromises. Therefore, monitoring and early detection of circulatory compromises signs are crucial to the free flap surgery. At present, clinical free flap care and monitoring are performed mainly through scheduled inspections by nursing staff. The caring process is labor-intensive and the success rate of detection varies owing to subjective judgement and experiences of carers. Other available methods of monitoring includes the continuity usage of monitor systems, which are invasive and incur higher cost. It is noticeable that an alternative method of monitoring is imminent. Therefore, previous scholars suggested infrared imaging as an alternative method to current monitoring for its inexpensive, non-invasive, non-contacting, non-radioactive, real-time, and repeatable features. Through the Factor Analysis algorithm, it could effectively remove the common factors of human physiology and observe the changes of real temperature. However, the method does not apply in more complex situations. If the estimation of common factors is not a linear combination of all factors but one factor occurs after another, the Factor analysis algorithms is unable to segregate the factors. Because if the estimated of the common factor is not a linear combination of all factors, but a certain factor occurs first and then a factor occurs later, then Factors Analysis algorithms can't separate them. Therefore, based on theoretical basis of prior researches, this study aimed to analyze the unique features of the observed signals by using the Independent Component Analysis algorithms for its flexibility. Plus, the factors could not be easily explained by the Factor Analysis algorithm. Through the advantages and capabilities of the Independent Component Analysis algorithm, the study hope to propouse a temperature variation detection algorithm that is more suitable for the Infrared thermography monitoring system for microvascular free flap after surgery. Out of the four animal experiments and twenty-three cases monitored in the study, only one patient developed venous thrombosis. Through the Factors Analysis algorithm, we found that the temperature of the free flap had significantly decreased. According to the results of the Independent Component Analysis algorithm, we found that there was two independent factors that may affect the temperature of the free flap region. One of the independent factors could detect the drop of the temperature earlier than the result of the factor analysis, and the other is the same as the result of the Factor Analysis. Both of the results showed the potential of early detection in temperature changes comparing to nursing records. However, due to the limitation of sampling, the study was unable to provide further verifications and discussions. It is hoped that more data will be available in the future to support the aim analytical methods. The goal of this study it to setup a foundation in the development of an auxiliary monitoring tool to monitor free flap pedicel thrombosis after a lesion removal surgery.