201206387 六、發明說明： 【發明所屬之技術領域】 本發明大致上係關於用於產生高解析度、大視場 (「FOV」）影像之一傾斜内視鏡的自動機控制。本發明明 確言之係關於用於在顯像一體積時自動機控制内視鏡位置/ 定向及速度的内視鏡影像回饋。 【先前技術】 一内視鏡係具有從一身體内部顯像之能力的一裝置。一 内視鏡之實例包含（但不限於）任何類型之醫療鏡（例如，支 氣管鏡、結腸鏡、腹腔鏡等等）。特定言之，硬式内視鏡 係由沿内視鏡軸設置的—系列透鏡組成。 微創手術係透過若干小孔而予以進行。内視鏡常常用來 提供手術部位的視覺回饋。舉例而言，在完整内視鏡心臟 手術中内視鏡係用於提供心臟動脈的手術中即時視覺 化。歸因於其尺寸（通常<=1〇 mm)及至所考慮之一解剖目 標之相對距離，内視鏡僅提供一小區域的視覺化。對於一 名外科醫生而言，此可能在瞭解觀看區域的一相對位置時 造成問題。 為改良視場，手術中所使用之内視鏡常常是傾斜的，其 中透鏡與内視鏡軸成一角度。此容許一外科醫生藉由沿内 視鏡之軸線轉動該内視鏡及/或改變該内視鏡在解剖區之 插入角度而對一解剖區進行一手動視覺掃描。從掃描中， 外科醫生使用在螢幕上觀看之一系列内視鏡影像來建立手 、 的〜像圖。然而，使用一傾斜觀看内視鏡手動掃 150238.doc 201206387 描手術區域可能是一個令人疲倦的過程且鑑於手眼協作之 複雜性而傾向於發生錯誤。 在本技術中，已使用自動機系統以在一外科醫生使用一 计算機滑鼠或其他輸入裝置控制該自動機系統時固持内視 鏡。然而，雖然將自動機附接至内視鏡可改良該内視鏡之 操控’但是先前技術並未解決建立手術部位之一心像圓的 困難。 【發明内容】 本發明提供用於使用-自動機操控的傾斜内視鏡而在微 創手術巾獲得-手術王作區之完整視野的方法^此等方法 涉及移動内視鏡直至影像擷取掃描涵蓋所需工作區體積。 所收集、與相關聯内視鏡位置及^向（從已知自動機運動 學導出）組合的影像序列係用於重構-視覺、圖形或以其 他方式表u作區體積。雖然促進獲得—工作區體積之最 大影像涵蓋的機會，但是此等方法進—步促進最大化内視 鏡影像之重構品質之機會、最小化運行時間且遵從由插入 點及内視鏡纜線所施加的任何物理限制。 本發明之一形式為一種採用一自動機單元及一控制單元 的自動機型顯像系統。 該自動機單元包含用於產生包含一工作區體積之複數幅 内視鏡影像之一視訊流的一傾斜内視鏡，以及用於在該工 作區體積内移動該内視鏡的一自動機。 該控制單元包含—自動機控制器、—掃描控制模組及一 影像重構器。t亥自動機控制器命令該自動機在該工作區體 15023S.doc 201206387 積内執行該内視鏡的一或多次影像擷取掃描，各影像擷取 掃描包含該内視鏡在該工作區體積内之—或多次轉動運 動。該掃描控制模組將在該或該等影像擷取掃描期間所產 生之各内視鏡影像與該内視鏡在該工作區體積内之一對應 轉動姿勢相連結。該影像重構器從經產生之内視鏡影像與 該内視鏡在該工作區體積内之該對應轉動姿勢之連結中重 構該工作區體積的一體積影像。 本發明之一第二形式包含一種用於使一傾斜内視鏡產生 ι a工作區體積之複數幅内視鏡影像之一視訊流的自動 機顯像方法。該自動機方法涉及命令一自動機執行該内視 鏡在該工作區體積内之一或多次影像擷取掃描，各影像擷 取掃描包含該自動機控制該内視鏡在該工作區體積内的一 或多次轉動運動。該自動機顯像方法進一步包含將該或該 等影像拮員取掃描期間所產生之各内視鏡影像與該内視鏡在 該工作區體積内的一對應轉動姿勢相連結，以及從該等經 產生之内視鏡影像與該内視鏡在該工作區體積内之該對應 轉動姿勢之連結中重構該工作區體積的體積影像。 從結合附圖閱讀之本發明之多項實施例之下列詳細描述 中’本發明之前述形式及其它形式以及本發明之各種特徵 及優點將變得進一步顯而易見。詳細描述及圖式僅為本發 明之說明性而非限制性，本發明之範圍係由隨附技術方案 及其等效物界定。 【實施方式】 如圖1所展不’一種自動機顯像系統採用一自動機單元 150238.doc 201206387 多個控制單凡2〇 ’用於涉及從一工作區體積之兩個⑺或 =個内視鏡影像重構-體積影像的任何自動機程序。此類 動機程序之實例包含（但不限於）醫療程序、裝配線程序 及涉及行動自動機之程序。特定言之，可利用該自動機系 統用於包合（但不限於）微創心臟手術（例如，冠狀動脈繞道 移植術或二尖瓣換置）、微創腹部手術（腹腔鏡檢查）(例 如，攝護腺切除術或膽囊切除術）及天然開口經腔道内視 鏡手術的醫療程序。 自動機單元1G包含-自動機u及硬式附接至該自動機u 的一傾斜内視鏡12。 中自動機11廣義上係定義為在結構上經組態為機 動化控制於操縱料自動機㈣所需之―末端效應器之 -或多個接頭的任何自動機裝置。實務上，自動㈣可具 有包含—末端效應器平移、—末端效應器軸線轉動之-五 ()個自由度’以及接頭之三⑺個轉動自由度的一最小 值。 本文中，内視鏡12廣義上係定義為在一工作區體積内具 有用於顯像之-傾斜視場的任何裝置。出於本發明之目的 之内視鏡12之實例包含（但不限於）任何類型之可撓或硬式 鏡（例如，内視鏡、關節鏡、支氣管鏡、膽道鏡、結腸 鏡、膀胱鏡、十二指腸鏡、胃鏡、宮腔鏡、腹腔鏡、喉 鏡月®至鏡、耳鏡、推腸鏡、鼻喉镜、乙狀結腸鏡、鼻赛 镜、胸腔镜等等）及類似於配備—影像系統之一鏡的任何 裝置（例如具有顯像之-巢套插管）。顯像是局部的，且表 150238.doc 201206387 面影像可光學地以光纖、透鏡或小型（例如基於CCD)成像 系統來獲得》 本文中，出於本發明之目的，術語「工作區體積」廣義 上係定義為任何三維空間，其中可將内視鏡丨2插入該三維 空間中用於在工作區體積内顯像（若干）目標且/或用於顯像 該工作區體積之邊界，本文中，術語「内視鏡影像」廣義 上係定義為内視鏡12之一視場内之一工作區體積之一影 像’且本文中’術語「體積影像」廣義上係定義為工作區 體積之兩幅（2)或多幅内視鏡影像的一拼接。 實務上’内視鏡12係安裝至自動機11的末端效應器。自 動機11之末端效應器之一姿勢為末端效應器在自動機^之 一座標系統内之一位置及一定向。在將内視鏡12插入一工 作區體積内的情況下，内視鏡12在工作區體積内之視場的 任何給定姿勢對應於末端效應器在自動機座標系統内的一 獨特姿勢。因此，可將工作區體積内由内視鏡12所產生之 各個別内視鏡影像與該工作區體積内内視鏡12的對應姿勢 相連結。 本文現將提供圖2A至圖2E之描述以促進自動機單元1〇 的操作瞭解。明確言之’在一球形工作區體積4〇内插入一 内視鏡30且一自動機（未展示）執行涉及内視鏡3〇在工作區 體積40内之一或多次轉動運動的一影像操取掃描及涉及内 視鏡30相對於工作區體積40之一預定區域之一固定姿勢的 一内視鏡影像擷取。 如圖2 A所展示，内視鏡3 〇係透過一插入點41而插入至在 150238.doc 201206387 工作區體積40之一下半球40b内設置一視場32之工作區體 積内的一深度。另外’内視鏡30具有與垂直於插入點41之 一工作區軸線42重合的一内視鏡軸線3 1 ^在一影像擷取掃 描期間’内視鏡30繞内視鏡軸線31自轉且掃描涵蓋$360。 的轉動以獲取工作區體積40的兩幅或多幅内視鏡影像。 如圖2B所展示，内視鏡30係透過一插入點41而插入至在 工作區體積40之一上半球40a内設置一視場32之工作區體 積内的一深度。内視鏡軸線3 1再次與工作區軸線42重合， 其中’在一影像擷取掃描期間，内視鏡30繞内視鏡轴線3 i 自轉且掃描涵蓋S360°的轉動以獲取工作區體積4〇的兩幅 或多幅内視鏡影像。 如圖2C所展示，内視鏡30係透過一插入點41而插入至在 工作區體積40之一下半球40b内設置一視場32之工作區體 積内的一深度，且内視鏡3 0係相對工作區軸線42傾斜。在 一影像擷取掃描期間’内視鏡30在掃描涵蓋$360。的轉動 的情況下同時繞内視鏡轴線3 1自轉且圍繞工作區軸線42旋 轉以獲取工作區體積40的兩幅或多幅内視鏡影像。或者， 内視鏡30可僅繞内視鏡軸線31自轉或僅圍繞工作區轴線42 旋轉。 如圖2D所展示，内視鏡30係透過一插入點41而插入至在 工作區體積40之一上半球40a内設置一視場32之工作區體 積内的一深度’且内視鏡3 0係相對工作區軸線42傾斜。在 一影像擷取掃据期間，内視鏡30在掃描涵蓋$360。的轉動 的情況下同時繞内視鏡轴線3 1自轉且圍繞工作區軸線42旋 150238.doc 201206387 轉以獲取工作區體積40的兩幅或多幅内視鏡影像。再者， 或者，内視鏡30可僅繞内視鏡轴線3 1自轉或僅圍繞工作區 軸線42旋轉。 如圖2E所展示，内視鏡30係透過一插入點41而插入且内 視鏡3 0係相對於工作區軸線42傾斜，藉以使視場32導向在 相對插入點41之一遠端運動中心（「RCM」）點42處。内視 鏡影像擷取涉及内視鏡30獲取定中心於RCM 42上的一内 視鏡影像。 實務上，工作區體積可或可非為球形。然而，在任一情 形中，熟悉此項技術者將從圖2A至圖2E之描述中明白如 何執行各種形狀與尺寸之工作區體積的影像擷取掃描及内 視鏡影像摘取。 往回參考圖1，控制單元2〇包含一自動機控制器21及一 影像重構器22。 本文中，自動機控制器2 1廣義上係定義為在結構上經組 態以提供一或多個自動機控制命令（r RCC」）25給自動機 11用於控制一影像擷取掃描或一内視鏡影像擷取所需之自 動機11之末端效應器之一姿勢的任何控制器。更特定言 之，自動機控制命令25支配在一影像擷取掃描期間達成自 動機11之末端效應器之所需轉動所需的各自動機接頭之確 定移動，且因此支配在該影像擷取掃描期間内視鏡12在一 工作區體積内之-所要轉動β類似地，自動機控制命令Μ 支配在-内視鏡影像擷取期間達成自動機u之末端效應器 之所要固定姿勢所需的各自動機接頭之確定移動，且因此 I50238.doc 201206387 支配在該内視鏡影像擷取期間内視鏡12在—卫作區體積内 之所要固定姿勢。 、本文中’影像重構器22廣義上較義為在結構上經組態 以執行用於從兩幅（2)或多幅重疊内視鏡影像產生_體積影 ㈣任何圖框拼接演算法的任何襄置。實務上，影像重構 器22可需要或可無需内視鏡12之校準。 舉例而言’當在圖3所展示之—平面上投射—圓台，如 圖2八至2〇所展示之内視鏡2〇的一次轉動可產生重疊内視 鏡影像51之-環狀序列5Ge在連續重疊内視鏡影像^中， 可觀察特徵部之轉動，諸如(例如)如圖4所展示之内視鏡影 像51a及51b之特徵部的轉動。 明確言之，緊靠内視鏡影像川及川的兩個突出特徵部 而繪出線52及53。隨著内視鏡轉動，特徵部隨著從内視鏡 影像51a至内視鏡影像51b一些特徵部會消失、且—些新特 徵部會在内視鏡影像51b内出現而轉動。若所觀察之序列 5〇之表面實際上為如所展示之-平面，則唯—影像榻取掃 描期間之已知自動機姿勢便足以重構序列5()之表面之圖 (拼集)。然而，由於所觀察之目標之形狀及深度未知，故 而須在諸圖框之間重構並且追蹤影像特徵部。 各特徵部在任何給定點 列方程式[1]而從影像在 出： 在影像重構器22之一實施例中， 之空間中之速度、及、可根據下 空間中之導數込及Iy及時間（It)而得 dx*Vx)+(Iy*Vy)-lt=〇 方程式[1]之一解為速度場-光流。明確言之，可俨 150238.doc 201206387 圖框之間有有限μ & 、 特徵部移動後使用一加窗（windowed)方 法且僅近似估外食 & 由中之導數而求解方程式[1]。藉由内視鏡 轉動而產生的所預 、 頂期之光流應取決於各點中之速度或轉動 及深度而位在呈士 、 /、有一可變向量長度的同心圓上。然而，所 追蹤之特徵部中之雜訊及缺陷將造成向量具有長度誤差及 角度誤差二者。跨越影像之平均誤差可經量化且用來校正 、鏡之速度。右正從體積中升高内視鏡，則光流將從影 像之中心流動°亦可^義如轉動情形中的相同誤差。 十於人轉動中之影像拼集而言，在知道那些特徵重疊 且那些特徵屬於新影像區段後，使用光流來對準連續内視 鏡~像的影像。若考慮轉動速度與視訊棟取之圖框速率之 間之比望各區段中有兩幅以上的圖框重疊。為改良拼 集之穩健性’可藉由對該等值求平均值或中間值而跨越許 多視野計算影像之各像素。跨越影像之該等值之平均偏差 (例如’標準方差)可用來評估重構之品質。或者，為改良 影像品質及清晰度，可以其至圖框邊界之距離來加權像素 值’從而以較高值加權靠近中心的像素。 對於用於多個影像序列之不同深度處之影像拼集而言， 可如在内視鏡影像之間相同地精確進行拼集。在此情形 中，替代一單個成像序列之拼接，相同方法可用來拼接多 個影像序列》 圖5A至圖5D繪示内視鏡圖框之一循序拼接。首先，將 如圖5B所展示之一内視鏡影像61拼接至如圖沾所展示的 一内視鏡影像60，其中醒目顯示内視鏡影像6〇以強調影像 150238.doc 12 201206387 拼接《接著，將如圖5C所展示之一内視鏡影像62拼接至内 視鏡影像61，其中再次醒目顯示内視鏡影像6〇以強調影像 拼接。一體積影像63中之進一步影像拼接結果係如圖51)所 展示，其中再次醒目顯示内視鏡影像6 〇以強調影像拼接。 往回參考圖丨，如本文將結合圖6之後續描述進一步闡 述，控制單元20包含一掃描控制模組23，本文中該掃描控 制模組23廣義上係定義為在結構上經組態用於將在一影像 擷取掃描或一内視鏡影像擷取期間所產生之各内視鏡影像 與内視鏡12在工作區體積内之對應姿勢相連結的任何模 組。另外，如本文將結合圖7之後續描述進一步闡述，掃 掐控制模組23可在結構上經組態用於控制工作區體積之各 影像擷取掃描期間内視鏡12的一轉動速度及/或一深度。 實務上，掃描控制模組23可由整合於自動機控制器21内、 整合於影像重構器22内、分散於自動機控制器21與影像重 構器22之間，或安裝於連接至自動機控制器^及/或影像 重構器22之一分離裝置上的硬體、軟體及/或韌體來實 施0 圖6繪示代表本發明之一自動機顯像方法的一流程圖 70。流程圖70之—階段371涵蓋由系統之—使用者手動進 行或由掃描控制模組23㈣進行的—初始深度定位及内視 鏡12之工作區㈣μ。實務上，視㈣之初始深度定 位及工作區軸線定向取決於該系統的自動機應用。 流程圖70之mS72涵蓋由影像重構器22及掃描控制 模，、且23執行由自動機控制器21命令之影像擷取掃描及視訊 150238.doc •13· 201206387 流13之處理。在此階段期間’掃描控制模組23將影像擷取 掃描期間所產生之各内視鏡影像與内視鏡12在工作區體積 内之對應轉動姿勢相連結以致使影像重構器22能夠適當拼 接内視鏡影像。 若僅欲執行多次影像擷取掃描，則流程圖7〇循環通過階 段S7 1-S73 ’其中掃描控制模組23以確保影像擷取掃描之 全部内視鏡影像充分重疊以促進體積影像之重構的一方式 對在工作區體積内之内視鏡進行深度定位與工作區軸線定 向。 若僅欲執行一次影像擷取掃描或在完成多次擷取掃描 後’流程圖70之一選用階段S74涵蓋内視鏡在工作區體積 内之深度定位與工作區軸線定向，此階段獲取該工作區體 積之一居中RCM或其他預定區域的一内視鏡影像。 貫務上’多次影像擷取掃描之次數及階段S74之需求係 取決於自動機程序及工作區體積之大小及形狀。 圖8 A至圖8 E繪不流程圖7 0的示例性執行。明確言之， 圖8Α展示通過一工作區體積100之一插入點1〇1而插入内視 鏡90直至内視鏡90之一頂部儘可能靠近工作區體積1〇〇的 底部，而在工作區體積1〇〇内無任何物體或光反射遮擋正 由内視鏡90產生的視訊流。此階段容許顯像工作區體積 1〇〇之底表面的最大區域。插入深度之確定可經由觀察而 手動完成或由追蹤影像中之一或多個解剖特徵部的掃描控 制模組23自動完成。 一旦處於圖8Β所展示之此組態，則内視鏡90繞其内視鏡 150238.doc -14 - 201206387 軸線完全轉動一次以獲取接近工作區體積100之底部之— 圓形圖框序列的影像掃描。當從上面觀看時投射掃描循著 一環面110。環面110之一寬度及因此由掃描涵蓋之面積取 決於傾斜内視鏡90之視角及從内視鏡90之一頂部至所顯像 . 之表面的一距離。工作區體積100之一 RCM 102並未被此 * 第一掃描涵蓋，因此將稍後在過程中解決。 一完整轉動在用於影像重構器22之初始内視鏡影像與最 終内視鏡影像之間提供適當重疊以重構連續條帶。舉例而 言，圍繞内視鏡軸線之一全程轉動通常將花費五（5)至十 (10)秒。手術領域中所使用之内視鏡通常以每秒約三十 (30)圖框進行。因此，在每十（1〇)度的轉動下，將存在約 八（8)幅内視鏡影像。 如圖8C所展示，在其定向固定的情況下，升高内視鏡9〇 使其離開工作區體積1〇〇以備下次掃描。此時，内視鏡9〇 在與如圖8D所展示之先前掃描之方向相反的方向上繞其内 視鏡軸線轉動。此將防止纜線變過度纏繞。從上面觀之， 累積涵蓋的區域現為兩個同心環面1丨〇及u丨且其等之間有 一些重疊。重疊之量應經選定以便促進重構，且由升高内 視鏡90之量來控制。如前，此決定可藉由觀察而得到，或 . 由追蹤影像中之解剖特徵部的掃描控制模組23自動得到。 在相反方向升高内視鏡9〇且進行轉動掃描的過程繼續直 至視野由於插入點101處之工作區體積1〇〇而被遮掩，或者 換吕之，直至從一垂直插入角度收集新資料的能力被耗 盡。 150238.doc -15- 201206387 若在内視鏡90之第一次反覆中，該内視鏡9〇之頂部未觸 碰工作區體積100的底部，則表示RCM 1〇2至表面上之投 射的-圓形區域將不可見q視覺化此區力，需傾斜内視 鏡90，藉此如圖8E所展示般使内視鏡9〇平行於工作區體積 的底表面。 取決於自動機程序及使用者偏好，維持内視鏡9〇垂直於 插入點101可足以在各影像擷取掃描期間視覺化與自動機 程序有關的全部結構。替代地或者同時地，在一或多次影 像擷取掃描期間，可如圖2(:及/或圖2D所展示般定向内視 鏡90。 明確言之，對於工作區體積1〇〇之上半部而言，繞插入 點1 01傾斜内視鏡9〇以使透鏡更靠近工作區體積i 〇〇。傾斜 之里使得與一先前掃描仍存在一些重疊。在内視鏡9〇現相 對於插入平面成角度的情況下，内視鏡90之視場接著圍繞 垂直於通過插入點丨〇丨之平面的工作區軸線旋轉。内視鏡 90同時亦繞其軸自轉’因此淨效應為内視鏡9〇之透鏡面對 總是最接近内視鏡9〇之一頂部之工作區體積1〇〇之表面。 當從上面觀看時’所得投射掃描採取如前之一環面的形 接著’内視鏡90同時升高且成角度以觀看工作區體積 100的甚至更高區段。接著，在相反方向上進行兩種同時 轉動（繞插入軸線法線自轉及圍繞軸旋轉）以緩解纜線纏繞 問題。此過程繼續直至收集不到進一步資訊或内視鏡9〇之 角度到達由工作區體積100所約束的一限度。餘留待被涵 150238.doc -16 - 201206387 蓋之最終區域為工作區體積100之底部處的區域，其與垂 直於插入平面且通過插入點101的一向量對齊成一直線。 再者，在此點處，内視鏡90之頂部應靠近插入點i 〇1，因 此使其傾斜並且轉動使得内視鏡9〇之透鏡面對底部。影像 應足夠大以涵蓋刺餘區域。一旦獲取影像，則從工作區體 積1 〇 〇移除内視鏡9 0且將影像序列重構成一.連貫的體積影 像，因此完成該方法。 頰似地，對於工作 將内視鏡90插入至插入平面直至工作區體積1〇〇内之可能 之最低深度。第-次掃描等同於請所展示的掃描。然 而，在各内視鏡升高步驟，升高運動組合内視鏡繞插入點 101的傾斜以減小連續掃描之透鏡至表面距離之間之像 差。完成各升高及傾斜以在與由先前掃描涵蓋之表面充分 重疊的情況下設置涵蓋新表面的一新掃描。初始掃描之後 之各次掃歸及繞插入平面法線及軸的同時轉動，以造成 内視鏡9〇之透鏡面對最靠近内視鏡90之表面的-方式同牛 化。重複掃描直至無新表面可見或者内視鏡9〇無法作㈣ 進一步之傾斜。接著定向内視鏡9〇以在被移除 積底面以完成程序。 取體 往回參寺圖6，可以用认々 各次掃描之内視鏡之—固定II 動速度及諸次掃描之間之 U疋轉 流程70。或者，圖7繪 進仃 f不代表促進對用於各次掃描之内拍 鏡之轉動速度及諸次掃播内視 的本發明之一影像回饋方法的-流程二IS調 的本發明之-制“丄.内視鏡之冰度距離之調整 流 150238.doc 201206387 程請之-階段S81涵蓋最初設定用於最初掃描之内視鏡 之轉動速度及用於最初掃描之内視鏡之最低深度的掃描控 制模組23，以及發布自動機命令25以達成工作區體積内之 内視鏡之設定深度及内視鏡在用於第一次掃描之設定轉動 速度下之轉動的自動機控制器21。在該第一次掃描之後， 流程圖80之一階段S82涵蓋分析體積影像之正進行中重構 之品質的掃描控制模組23。流程圖8〇將為各後續掃描而保 持循環在階段S81及S82中直至掃描控制模組23認知到體積 I像之JL進行f重構之品質無法令人接受之時。在該情形 中，流程圖80之一階段S83涵蓋為後續掃描調整内視鏡之 轉動速度且/或調整後續掃描之間之深度距離的模組Μ。 實務上’可將流程圖8〇整合於流程圖7〇之階段s7i中（圖 6) ° 舉例而言，在最初執行階段S71的情況下，將内視鏡 12(圓1)之一轉動速度設定為自動機u(圖⑽一最大轉動 速度。此將確保最快的可能操作。對於各圖框而言，影像 重構器22(圖1)進行如本文先前所述之拼集且模組^(圖〇 使用向里、平均值或二者評估重構之品質。若重構之品質 低於一使用者指定臨限’則模組23通知自動機控制器 21 (圖1)減小内視鏡12之轉動速度以確保更順暢的圖框轉 變，其將透過窗大小而影響光流計算且透過容許來自更多 巾田圖框之強度訊息而影響重構。可使用與影像品質之線性 關係或者由一預定放大倍數來減小速度。 實務上，即使品質改善仍可降低轉動速度。考慮到重構 150238.doc -18- 201206387 將最好地罪近目標，預期轉動速度將隨時間而降低。 類似地，可類似地進行内視鏡深度之控制。明確言之， 内視鏡12之深度非為—連續值。在各次掃描後將内視鏡升 高一特定量。因此，期望有更少之重疊及重構之更低品 質。展現以下是簡單的：對於相㈤之深度增加，將連續掃 私之間之5G /。的重疊增加^倍。此重疊可確保有足夠的資 。孔進行成功的掃描。因此，可在最初執行階段s71時將重 疊之ΐ叹疋成最初值（例如，>5〇%)。隨著演算法計算拼 集，可減小深度增量以達成較佳影像拼集。 :藉由下列而進行影像拼集之品質控制：1)將所預期之 光机向里疋向與所觀察之光流向量定向相比較舉例而 言，誤差指示符可為所預期之向量與所觀察之向量之間之 一角度；或者2)在將影像轉換成一參考圖框之後將兩個或 夕個t/像之間之像素值相比較^兩種措施均是對於每像素 可藉由取平均 '均方根或此項技術中已知之組合誤 差的任何其他統計方法而為—拼集步驟歸納該等措施。 4回乡考圖1 ’控制單元2G進—步包含-影像顯示模組 24:本文中該影像顯示模組以廣義上係定義為在結構上經 組態用於騎在—經重構之體積影像内之卫作區體積之一 視汛流的一影像之任何模組。舉例而言，可如圖9A所展示 般重構心臟120之—體積影像⑵，其中隨著如圖祁及圖 9C所展示般經由自動機u控制内視鏡12，在—經重構之體 積影㈣内醒目顯示來自内視鏡12的—視訊流122。此致 使内視鏡12之—使用者在使用者正經由内視鏡12進行-手 150238.doc -19- 201206387 術時能夠視需要視覺化體積影像】2J。 貫務上’影像顯示模組24可由整合於 J田1 β於自動機控制器21 …於影像重構器22内、分散於自動機控制器幻與影 像重構器22之間，或安裝於連接至自動機㈣㈣及/或 影像重構器22之一分離裝詈卜的麻挪紅λ 雕衷置上的硬體、軟體及/或韌體來 實施。 從本文圖！至圖9之描述中，一般技術者將明白本發明之 眾多優點包含(但不限於)最大化一體積影像之重構品質及 覆蓋的機會，最小化-體積影像之重構之運行時間，以及 遵從由工作區體積之插人點及内視鏡I線所施加的任何物 理限制。此外，雖財文以在各次掃描之後升高内視鏡來 4田述圖8’但疋實務上可在各次掃描之後下降該内視鏡。 儘管已參考示例性態樣、特徵及實施描述本發明，然而 :揭示之系統及方法不限於此類示例性態樣、特徵及/或 貫&事貫Ji從本文所提供之描述巾，熟悉此項技術者 將顯而易見在不脫離本發明之精神或範圍的情況下，所揭 示之系統及方法可以有修飾例、替代例及增強例。相應 地，本發明明確涵蓋在本發明之範圍内的此類修飾例、替 代例及增強例。 【圖式簡單說明】 圖1繪不根據本發明之一自動機顯像系統的一示例性實 施例》 圖2A至圖2E綠示圖1之自動機顯像系统的多種示例性顯 像模式β 150238.doc -20· 201206387 圖3繪不根據本發明的—示例性影像榻取掃描。 圖4繪不此項技術已知的—示例性影像處理。 圖5A至圖5D緣不根據本發明從内視鏡影像示例性產生 一體積影像。 圖6繪示根據本發明之一 施例的一流程圖代表。 圖7繪示根據本發明之— 例的一流程圖代表。 自動機顯像方法之一示例性實 影像回饋方法之一示例性實施 圖8A至圖8E繪示根據本發 的流程圖。 明不例性執行圖3及圖4繪示 不例性顯示一體積影像及 圖9A至圖9C繪示根據本發明 一内視鏡影像。 【主要元件符號說明】 10 自動機單元 11 自動機 12 傾斜内視鏡 13 視訊流 20 控制單元 21 自動機控制器 22 影像重構器 23 掃描控制模組 24 影像顯示模組 25 自動機命令 30 内視鏡 150238.doc •21- 201206387 31 内視鏡軸線 32 視場 40 工作區體積 40a 工作區體積之上半球 40b 工作區體積之下半球 41 插入點 42 工作區軸線 50 序列 51a 内視鏡影像 51b 内視鏡影像 52 線 53 線 60 内視鏡影像 61 内視鏡影像 62 内視鏡影像 63 體積影像 90 内視鏡 100 工作區體積 101 插入點201206387 VI. Description of the Invention: TECHNICAL FIELD OF THE INVENTION The present invention generally relates to automaton control for producing a tilted endoscope of one of high resolution, large field of view ("FOV") images. The present invention is expressly related to endoscopic image feedback for automata control of endoscope position/orientation and speed when developing a volume. [Prior Art] An endoscope has a device that develops the image from inside the body. An example of an endoscope includes, but is not limited to, any type of medical lens (e.g., bronchoscope, colonoscope, laparoscope, etc.). In particular, rigid endoscopes consist of a series of lenses placed along the axis of the endoscope. Minimally invasive surgery is performed through a number of small holes. Endoscopes are often used to provide visual feedback of the surgical site. For example, in complete endoscopic cardiac surgery endoscopy is used to provide instant visualization of the heart artery during surgery. Due to its size (typically <= 1 mm) and the relative distance to one of the anatomical targets considered, the endoscope provides only a small area of visualization. For a surgeon, this may cause problems when understanding a relative position of the viewing area. To improve the field of view, the endoscopes used in surgery are often tilted, with the lens at an angle to the axis of the endoscope. This allows a surgeon to perform a manual visual scan of an anatomical region by rotating the endoscope along the axis of the endoscope and/or changing the insertion angle of the endoscope at the anatomical region. From the scan, the surgeon uses a series of endoscope images on the screen to create a hand-like image. However, using a tilt-view endoscope to manually scan the surgical area may be a tiresome process and tends to go wrong due to the complexity of hand-eye collaboration. In the present technology, an automaton system has been used to hold an endoscope when a surgeon controls the automaton system using a computer mouse or other input device. However, although attaching an automaton to an endoscope can improve the manipulation of the endoscope', the prior art does not address the difficulty of establishing a centroid of one of the surgical sites. SUMMARY OF THE INVENTION The present invention provides a method for obtaining a complete field of view of a minimally invasive surgical towel using a tilted endoscope controlled by an automaton - the method of moving the endoscope until image capture scanning Covers the required workspace volume. The sequence of images collected, combined with associated endoscopic position and orientation (derived from known automata kinematics), is used for reconstruction-visual, graphical or otherwise zonal volume. While facilitating access to the largest image coverage of the workspace volume, these methods further facilitate the opportunity to maximize the quality of the reconstructed image of the endoscope image, minimize run time and comply with the insertion point and endoscope cable Any physical restrictions imposed. One form of the present invention is an automatic model development system employing an automaton unit and a control unit. The automaton unit includes a tilting endoscope for generating a video stream of a plurality of endoscope images including a workspace volume, and an automaton for moving the endoscope within the workspace volume. The control unit includes an automaton controller, a scan control module, and an image reconstructor. The automaton controller commands the automaton to perform one or more image capture scans of the endoscope in the work area 15023S.doc 201206387, and each image capture scan includes the endoscope in the work area - or multiple rotations within the volume. The scan control module couples each of the endoscopic images generated during the or the image capture scan with a corresponding rotational posture of the endoscope within the volume of the work area. The image reconstructor reconstructs a volumetric image of the volume of the workspace from the resulting combination of the endoscopic image and the corresponding rotational posture of the endoscope within the volume of the workspace. A second form of the present invention comprises an auto-developing method for causing a tilting endoscope to generate a video stream of a plurality of endoscope images of a working area volume. The automaton method involves commanding an automaton to perform one or more image capture scans of the endoscope in the volume of the work area, each image capture scan including the automaton controlling the endoscope within the volume of the work area One or more rotations. The automaton imaging method further includes associating, and from each of the endoscope images generated during the scanning of the image trapping and the endoscope in a corresponding rotational posture within the volume of the working area A volumetric image of the volume of the workspace is reconstructed from the resulting endoscopic image and the corresponding rotational posture of the endoscope within the volume of the workspace. The above-described and other aspects of the present invention, as well as the various features and advantages of the present invention, will become more apparent from the detailed description of the invention. The detailed description and drawings are intended to be illustrative and not restrict [Embodiment] As shown in Fig. 1, an automaton visualization system adopts an automata unit 150238.doc 201206387. Multiple control units are used for two (7) or = within a working area volume. Mirror Image Reconstruction - Any automaton program for volumetric images. Examples of such motivational programs include, but are not limited to, medical procedures, assembly line procedures, and procedures involving an automated machine. In particular, the automaton system can be utilized for inclusion (but not limited to) minimally invasive cardiac surgery (eg, coronary bypass grafting or mitral valve replacement), minimally invasive abdominal surgery (laparoscopic) (eg , prostatectomy or cholecystectomy) and medical procedures for natural open endoscopic surgery. The automaton unit 1G includes an automaton u and a tilting endoscope 12 that is rigidly attached to the automaton u. The intermediate automaton 11 is broadly defined as any automaton device that is structurally configured to be machined to control the "end effector" or multiple joints required by the manipulator automaton (4). In practice, the automatic (4) may have a minimum value including - end effector translation, - five () degrees of freedom of the end effector axis rotation, and three (7) rotational degrees of freedom of the joint. Herein, endoscope 12 is broadly defined as any device having a slanted field of view for visualization within a workspace volume. Examples of endoscope 12 for purposes of the present invention include, but are not limited to, any type of flexible or rigid mirror (eg, endoscope, arthroscopy, bronchoscopy, choledochoscopy, colonoscopy, cystoscopy, duodenum) Mirror, gastroscope, hysteroscopy, laparoscopic, laryngoscope monthly to mirror, otoscope, colonoscopy, laryngoscope, sigmoidoscopy, nasal goggles, thoracoscopy, etc.) and similar to one of the imaging systems Any device of the mirror (eg with a visualization - nested cannula). The visualization is local and the surface image can be obtained optically, by fiber, lens or small (eg CCD based) imaging system. For the purposes of the present invention, the term "work area volume" is broadly defined. The upper system is defined as any three-dimensional space in which the endoscope 丨 2 can be inserted into the three-dimensional space for imaging (several) targets within the workspace volume and/or for visualizing the boundary of the workspace volume, The term "endoscope image" is broadly defined as one of the workspace volumes in one of the fields of view of the endoscope 12 and the term "volume image" is defined herein as two of the volume of the workspace. (2) or a mosaic of multiple endoscopic images. In practice, the endoscope 12 is attached to the end effector of the robot 11. One of the end effectors of the self-motivator 11 is in the position and direction of the end effector in the standard system of the robot. In the case where the endoscope 12 is inserted into the volume of a work area, any given pose of the field of view of the endoscope 12 within the volume of the work area corresponds to a unique pose of the end effector within the automaton coordinate system. Therefore, the individual endoscopic images produced by the endoscope 12 within the working volume can be coupled to the corresponding posture of the endoscope 12 within the working volume. The description of Figures 2A through 2E will now be provided to facilitate an understanding of the operation of the robot unit 1A. Specifically, an endoscope 30 is inserted into a spherical working area volume and an automaton (not shown) performs an image involving one or more rotational movements of the endoscope 3 in the working volume 40. The scanning and an endoscopic image capture involving the fixed posture of the endoscope 30 with respect to one of the predetermined areas of the workspace volume 40 is performed. As shown in Fig. 2A, the endoscope 3 is inserted through an insertion point 41 into a depth within the volume of the working area 32 of the field of view 32 in the lower hemisphere 40b of one of the 150238.doc 201206387 workspace volumes 40. In addition, the endoscope 30 has an endoscope axis 3 1 that coincides with a working area axis 42 perpendicular to the insertion point 41. ^ The endoscope 30 rotates and scans around the endoscope axis 31 during an image capture scan. Covers $360. Rotation to obtain two or more endoscopic images of the workspace volume 40. As shown in Fig. 2B, the endoscope 30 is inserted through an insertion point 41 into a depth within the volume of the working area in which a field of view 32 is disposed in an upper hemisphere 40a of the workspace volume 40. The endoscope axis 31 again coincides with the work area axis 42, wherein during the image capture scan, the endoscope 30 rotates about the endoscope axis 3 i and scans to cover the S360° rotation to obtain the work area volume 4 Two or more endoscope images of the skull. As shown in FIG. 2C, the endoscope 30 is inserted through an insertion point 41 into a depth within the volume of the working area in which a field of view 32 is disposed in one of the lower hemispheres 40b of the working volume 40, and the endoscope 30 is attached. It is inclined relative to the work area axis 42. During an image capture scan, the endoscope 30 covers $360 in the scan. In the case of rotation, it simultaneously rotates about the endoscope axis 31 and rotates around the working area axis 42 to obtain two or more endoscopic images of the working volume 40. Alternatively, the endoscope 30 may only rotate about the endoscope axis 31 or only about the work area axis 42. As shown in FIG. 2D, the endoscope 30 is inserted through an insertion point 41 into a depth within the volume of the working area in which a field of view 32 is disposed in one of the upper hemispheres 40a of the working volume 40 and the endoscope 3 0 It is inclined relative to the working area axis 42. During an image capture scan, the endoscope 30 covers $360 in the scan. In the case of rotation, the inner mirror axis 3 1 is rotated at the same time and rotated around the working area axis 42 150238.doc 201206387 to obtain two or more endoscope images of the working area volume 40. Still alternatively, the endoscope 30 may only rotate about the endoscope axis 3 1 or only about the work area axis 42. As shown in FIG. 2E, the endoscope 30 is inserted through an insertion point 41 and the endoscope 30 is tilted relative to the work area axis 42 to direct the field of view 32 to a distal center of motion relative to one of the insertion points 41. ("RCM") point 42. The endoscopic image capture involves the endoscope 30 acquiring an endoscopic image centered on the RCM 42. In practice, the workspace volume may or may not be spherical. However, in either case, those skilled in the art will appreciate from the description of Figures 2A through 2E how to perform image capture scans and endoscopic image capture of workspaces of various shapes and sizes. Referring back to Figure 1, the control unit 2A includes an automaton controller 21 and an image reconstructor 22. Herein, the automaton controller 2 1 is broadly defined as being structurally configured to provide one or more automaton control commands (r RCC) 25 for the automaton 11 to control an image capture scan or The endoscope image captures any controller of one of the desired end effectors of the automaton 11. More specifically, the automaton control command 25 governs the determined movement of the respective motive joints required to achieve the desired rotation of the end effector of the automaton 11 during an image capture scan, and thus governs during the image capture scan The endoscope 12 is rotated within a working area volume. Similarly, the automaton control command dictates the respective motives required to achieve the desired fixed posture of the end effector of the automaton during the endoscopic image capture. The determined movement of the joint, and therefore I50238.doc 201206387, governs the desired fixed posture of the endoscope 12 within the volume of the guarding area during the endoscopic image capture. In this context, the image reconstructor 22 is broadly defined as being structurally configured to perform any frame mosaic algorithm for generating _ volume shadows from two (2) or multiple overlapping endoscope images. Any device. In practice, image reconstructor 22 may or may not require calibration of endoscope 12. For example, 'when projected on a plane as shown in FIG. 3—a circular table, one rotation of the endoscope 2〇 as shown in FIGS. 2-8 to 2 can produce an overlapping endoscope image 51-loop sequence 5Ge In the continuously overlapping endoscope image ^, the rotation of the features can be observed, such as, for example, the rotation of the features of the endoscopic images 51a and 51b as shown in FIG. To be clear, lines 52 and 53 are drawn next to the two prominent features of the endoscope image of Sichuan and Sichuan. As the endoscope rotates, the features disappear as the features from the endoscopic image 51a to the endoscopic image 51b disappear, and some of the new features appear in the endoscopic image 51b. If the surface of the sequence observed is actually a plane as shown, then the known automaton pose during the image-reading scan is sufficient to reconstruct the map of the surface of sequence 5() (set). However, since the shape and depth of the observed object are unknown, it is necessary to reconstruct and track the image features between the frames. Each feature is from the image at any given point in equation [1]: in one embodiment of image reconstructor 22, the velocity in space, and, according to the derivative in the lower space, and Iy and time (It) and dx*Vx)+(Iy*Vy)-lt=〇 One of the equations [1] is solved as velocity field-optical current. To be clear, 俨 150238.doc 201206387 There is a finite μ & between the frames, the feature is moved and a windowed method is used and only the outer food & approximation is used to solve the equation [1]. The pre- and top-period light flow produced by the rotation of the endoscope should be on a concentric circle with a variable vector length depending on the speed or rotation and depth in each point. However, the noise and defects in the traced features will cause the vector to have both length and angular errors. The average error across the image can be quantified and used to correct the speed of the mirror. When the right side raises the endoscope from the volume, the light flow will flow from the center of the image. It can also be the same error as in the case of rotation. In the case of image stitching in a person's rotation, after knowing that the features overlap and those features belong to the new image segment, the optical flow is used to align the images of the continuous endoscope-image. Considering the ratio between the rotational speed and the frame rate of the video frame, there are more than two frames in each segment overlapping. To improve the robustness of the assembly, each pixel of the image can be computed across a plurality of fields of view by averaging or intermediate the values. The average deviation (e.g., 'standard deviation) across the equivalent of the image can be used to assess the quality of the reconstruction. Alternatively, to improve image quality and sharpness, the pixel value can be weighted by the distance from the border of the frame to weight the pixels near the center with higher values. For image mosaics at different depths of multiple image sequences, the assembly can be exactly the same as between the endoscope images. In this case, instead of splicing a single imaging sequence, the same method can be used to stitch multiple image sequences. Figure 5A through Figure 5D illustrate one of the endoscopic frames. First, an endoscope image 61 as shown in FIG. 5B is spliced to an endoscope image 60 as shown in the image, wherein the endoscope image 6 is highlighted to emphasize the image 150238.doc 12 201206387 stitching An endoscope image 62, as shown in FIG. 5C, is spliced to the endoscope image 61, wherein the endoscope image 6 is again highlighted to emphasize image stitching. The result of further image stitching in a volumetric image 63 is shown in Figure 51), where the endoscope image 6 is again highlighted to emphasize image stitching. Referring back to the drawings, as will be further explained herein in connection with the subsequent description of FIG. 6, the control unit 20 includes a scan control module 23, which is broadly defined herein as being structurally configured for Any module that combines each of the endoscopic images produced during an image capture scan or an endoscopic image capture with the corresponding pose of the endoscope 12 within the workspace volume. In addition, as will be further explained herein in conjunction with the subsequent description of FIG. 7, the broom control module 23 can be configured to control the rotational speed of the endoscope 12 during scanning of each image of the workspace volume. Or a depth. In practice, the scan control module 23 can be integrated into the automaton controller 21, integrated into the image reconstructor 22, dispersed between the automaton controller 21 and the image reconstructor 22, or installed in an automaton. A hardware, software, and/or firmware on the device is separated from one of the controllers and/or image reconstructors 22. FIG. 6 illustrates a flow chart 70 representative of an automaton imaging method of the present invention. Flowchart 70 - Stage 371 covers the initial depth positioning by the system-user or by the scan control module 23 (4) and the working area (4) μ of the endoscope 12. In practice, the initial depth positioning of view (4) and the orientation of the work area axis depend on the automaton application of the system. The mS72 of the flowchart 70 covers the image reconstructor 22 and the scan control mode, and 23 performs the image capture scan and video commanded by the automaton controller 21 150238.doc •13· 201206387 Stream 13 processing. During this phase, the scan control module 23 couples the endoscopic images generated during the image capture scan with the corresponding rotational posture of the endoscope 12 within the volume of the work area to enable the image reconstructor 22 to properly splicing. Endoscope image. If only a plurality of image capture scans are to be performed, the flowchart 7 loops through the stages S7 1-S73 'where the scan control module 23 ensures that all the endoscopic images of the image capture scan are fully overlapped to promote the weight of the volume image. One way of constructing the depth of view of the endoscope within the volume of the work area is to orient the axis of the work area. If only one image capture scan is to be performed or after multiple capture scans are completed, one of the flow charts 70 selects stage S74 to cover the depth positioning of the endoscope in the workspace volume and the orientation of the work area axis. An endoscope image of one of the zone volumes centered on the RCM or other predetermined area. The number of times of image capture and the number of stages S74 in the transaction depends on the size and shape of the automaton program and the size of the workspace. 8A through 8E illustrate an exemplary execution of flowchart 70. Specifically, FIG. 8A shows that the endoscope 90 is inserted through one insertion point 1〇1 of a work area volume 100 until the top of one of the endoscopes 90 is as close as possible to the bottom of the work area volume, while in the work area. There is no object or light reflection within the volume of 1 遮 to block the video stream being generated by the endoscope 90. This stage allows the maximum area of the bottom surface of the imaging work area volume. The determination of the insertion depth can be done manually by observation or automatically by the scan control module 23 that tracks one or more anatomical features in the image. Once in this configuration as shown in FIG. 8A, the endoscope 90 is fully rotated about its endoscope 150238.doc -14 - 201206387 axis to obtain an image of the circular frame sequence near the bottom of the workspace volume 100. scanning. The projection scan follows a torus 110 when viewed from above. The width of one of the toroids 110 and thus the area covered by the scan depends on the angle of view of the tilted endoscope 90 and a distance from the top of one of the endoscopes 90 to the surface of the image. One of the workspace volumes 100 RCM 102 is not covered by this *first scan and will therefore be resolved later in the process. A complete rotation provides an appropriate overlap between the initial endoscopic image for image reconstructor 22 and the final endoscopic image to reconstruct a continuous strip. For example, it may typically take five (5) to ten (10) seconds to rotate around one of the axes of the endoscope. Endoscopes used in the surgical field are typically performed in approximately thirty (30) frames per second. Therefore, there will be about eight (8) endoscope images every ten (1 〇) degrees of rotation. As shown in Fig. 8C, with its orientation fixed, the endoscope 9 is raised away from the working volume 1 〇〇 for the next scan. At this time, the endoscope 9 turns about its inner mirror axis in a direction opposite to the direction of the previous scan as shown in Fig. 8D. This will prevent the cable from becoming excessively entangled. Viewed from above, the cumulative coverage area is now two concentric toruses and u丨 with some overlap between them. The amount of overlap should be selected to facilitate remodeling and is controlled by increasing the amount of endoscope 90. As before, this decision can be obtained by observation, or automatically obtained by the scan control module 23 that tracks the anatomical features in the image. The process of raising the endoscope 9 in the opposite direction and performing the rotational scan continues until the field of view is obscured by the volume of the work area at the insertion point 101, or is replaced until a new data is collected from a vertical insertion angle. The ability is exhausted. 150238.doc -15- 201206387 If the top of the endoscope 90 does not touch the bottom of the workspace volume 100 in the first iteration of the endoscope 90, it indicates that the RCM 1〇2 is projected onto the surface. - The circular area will be invisible q to visualize this zone force, and the endoscope 90 needs to be tilted, whereby the endoscope 9 is parallel to the bottom surface of the workspace volume as shown in Figure 8E. Depending on the automaton program and user preferences, maintaining the endoscope 9" perpendicular to the insertion point 101 may be sufficient to visualize all of the structure associated with the automaton program during each image capture scan. Alternatively or simultaneously, during one or more image capture scans, the endoscope 90 can be oriented as shown in Figure 2 (and/or Figure 2D.) Clearly, for a workspace volume above 1〇〇 In the half, the endoscope 9 is tilted about the insertion point 101 to bring the lens closer to the workspace volume i. The tilt is such that there is still some overlap with a previous scan. The endoscope 9 is opposite to the front mirror 9 With the insertion plane angled, the field of view of the endoscope 90 is then rotated about a working area axis that is perpendicular to the plane through the insertion point 。. The endoscope 90 also rotates about its axis at the same time 'so the net effect is internal view The lens of the mirror 9 is facing the surface of the working area that is always closest to the top of one of the endoscopes 9 。. When viewed from above, the resulting projection scan takes the shape of the former torus and then the internal view. The mirror 90 is simultaneously raised and angled to view an even higher section of the workspace volume 100. Next, two simultaneous rotations (rotation around the axis of rotation and rotation about the axis) are performed in the opposite direction to alleviate the cable entanglement problem. This process continues until the collection is not To the further information or the angle of the endoscope 9 到达 reaches a limit bound by the workspace volume 100. The remaining area to be covered by the culvert 150238.doc -16 - 201206387 is the area at the bottom of the workspace volume 100, It is aligned with a vector perpendicular to the insertion plane and passing through the insertion point 101. Again, at this point, the top of the endoscope 90 should be close to the insertion point i 〇 1 so that it is tilted and rotated so that the endoscope 9 The lens of the cymbal faces the bottom. The image should be large enough to cover the spur area. Once the image is acquired, the endoscope 90 is removed from the workspace volume 1 and the image sequence is reconstructed into a coherent volume image. This method is completed. Cheeky, for the work, insert the endoscope 90 into the insertion plane until the lowest possible depth within 1 工作 of the working area volume. The first scan is equivalent to the scan shown. However, within each The mirror elevation step increases the tilt of the motion combination endoscope about the insertion point 101 to reduce the aberration between the lens and the surface distance of the continuous scan. The completion of each elevation and tilt to be covered by the previous scan A new scan covering the new surface is set with sufficient overlap of the surface. Each scan after the initial scan and the simultaneous rotation of the inserted plane normal and the axis, so that the lens of the endoscope 9〇 faces the closest end view The surface of the mirror 90 is the same as that of the cow. The scanning is repeated until no new surface is visible or the endoscope 9〇 cannot be further (4) further tilted. Then the endoscope 9 is oriented to remove the bottom surface to complete the procedure. Figure 6 of the body back to the temple, you can use the scanning of the endoscope - fixed II moving speed and the U 疋 turn between the scans 70. Or, Figure 7 painted 仃 f does not mean to promote the pair The image feedback method of the present invention for the rotation speed of the mirror in each scan and the scanning of the internal view - the second method of the present invention - the "ice. The distance of the ice mirror of the endoscope" Adjusting Flow 150238.doc 201206387 - The S81 includes a scan control module 23 that initially sets the rotational speed of the endoscope for initial scanning and the lowest depth of the endoscope for initial scanning, and issues an automaton command 25 to reach within the workspace volume The set depth of the endoscope and the automaton controller 21 of the endoscope for rotation at the set rotational speed for the first scan. After this first scan, one stage S82 of flowchart 80 encompasses scan control module 23 that analyzes the quality of the ongoing reconstruction of the volumetric image. Flowchart 8 will maintain the loop for each subsequent scan in stages S81 and S82 until the quality of the f-reconstruction of the JL that the scan control module 23 recognizes the volume I image is unacceptable. In this case, one stage S83 of flowchart 80 encompasses a module 调整 that adjusts the rotational speed of the endoscope for subsequent scans and/or adjusts the depth distance between subsequent scans. In practice, the flowchart 8 can be integrated into the stage s7i of the flowchart 7 (Fig. 6). For example, in the case of the initial execution phase S71, one of the endoscopes 12 (circle 1) is rotated. Set to automaton u (Fig. (10) - maximum rotational speed. This will ensure the fastest possible operation. For each frame, image reconstructor 22 (Fig. 1) performs the assembly and module as described previously herein. ^ (The figure uses the inward, average or both to evaluate the quality of the reconstruction. If the quality of the reconstruction is lower than a user specified threshold, the module 23 notifies the automaton controller 21 (Fig. 1) to decrease The rotational speed of the mirror 12 ensures a smoother frame transition, which affects the optical flow calculation through the window size and affects the reconstruction by allowing the intensity information from more rows of the field. The linearity of the image quality can be used. The relationship is either reduced by a predetermined magnification. In practice, even if the quality is improved, the rotation speed can be reduced. Considering that the reconstruction 150238.doc -18-201206387 will best offend the target, the expected rotation speed will be over time. Similarly, the endoscope can be similarly Control of the degree. Specifically, the depth of the endoscope 12 is not a continuous value. The endoscope is raised by a specific amount after each scan. Therefore, it is expected that there is less overlap and lower quality of reconstruction. The following is simple: for the depth increase of the phase (five), the overlap of 5G /. between successive sweeps is increased by 2. This overlap ensures that there is sufficient capital. The hole is successfully scanned. Therefore, at the beginning When the stage s71 is executed, the overlapping sighs are converted to the initial values (for example, > 5〇%). As the algorithm calculates the stitches, the depth increment can be reduced to achieve a better image set. Perform quality control of the image assembly: 1) Compare the expected optometry to the observed optical flow vector orientation. For example, the error indicator can be between the expected vector and the observed vector. One angle; or 2) comparing the pixel values between two or one t/images after converting the image into a reference frame. Both measures are for averaging the 'root mean square' for each pixel. Or any other statistical measure of the combined error known in the art And is - make do step induction of such measures. 4 Hometown Test Figure 1 'Control Unit 2G Step-Including-Image Display Module 24: In this paper, the image display module is defined in a broad sense as being structurally configured for riding-reconstructed volume One of the dimensions of the image of the turbulence in the image is any module of the image of the turbulence. For example, the volumetric image (2) of the heart 120 can be reconstructed as shown in Figure 9A, wherein the reconstructed volume is controlled via the automaton u as shown in Figure 祁 and Figure 9C. The video stream 122 from the endoscope 12 is highlighted in the shadow (4). As a result, the user of the endoscope 12 can visualize the volume image as needed when the user is performing the operation through the endoscope 12 150238.doc -19-201206387. The image display module 24 can be integrated into the JIA 1 β in the automaton controller 21 ... in the image reconstructor 22, dispersed between the automaton controller and the image reconstructor 22, or mounted on It is connected to the automaton (4) (4) and/or one of the image reconstructor 22 to separate the mounting hardware, software and/or firmware of the mounting device. From this picture! In the description of FIG. 9, one of ordinary skill in the art will appreciate that numerous advantages of the present invention include, but are not limited to, maximizing the quality of reconstruction and coverage of a volumetric image, minimizing the runtime of reconstruction of the volumetric image, and Obey any physical limitations imposed by the insertion point of the workspace volume and the line of the endoscope I. In addition, although the financial text raises the endoscope after each scan, it is practical to drop the endoscope after each scan. Although the invention has been described with reference to exemplary aspects, features, and implementations, the disclosed systems and methods are not limited to such exemplary aspects, features, and/or aspects of the present disclosure. It will be apparent to those skilled in the art that the disclosed systems and methods may be modified, substituted, and enhanced without departing from the spirit and scope of the invention. Accordingly, the present invention is intended to cover such modifications, alternatives, and modifications. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 depicts an exemplary embodiment of an automaton imaging system not according to the present invention. FIG. 2A to FIG. 2E illustrate various exemplary imaging modes of the automata development system of FIG. 150238.doc -20· 201206387 Figure 3 depicts an exemplary image-taking scan that is not in accordance with the present invention. Figure 4 depicts an exemplary image processing not known in the art. Figures 5A through 5D do not exemplarily produce a volumetric image from an endoscope image in accordance with the present invention. Figure 6 depicts a flow chart representation in accordance with one embodiment of the present invention. Figure 7 is a flow chart representation of an example of the invention in accordance with the present invention. One of the exemplary real image feedback methods of the automaton development method is an exemplary embodiment. Figs. 8A to 8E are flowcharts according to the present invention. FIG. 3 and FIG. 4 illustrate an example of a volumetric image and FIGS. 9A-9C illustrate an endoscope image in accordance with the present invention. [Main component symbol description] 10 Automated unit 11 Automaton 12 Tilting endoscope 13 Video stream 20 Control unit 21 Automaton controller 22 Image reconstructor 23 Scan control module 24 Image display module 25 Automaton command 30 Mirror 150238.doc •21- 201206387 31 Endoscope axis 32 Field of view 40 Work area volume 40a Work area volume Upper hemisphere 40b Work area volume Lower hemisphere 41 Insert point 42 Work area axis 50 Sequence 51a Endoscope image 51b Endoscope Image 52 Line 53 Line 60 Endoscope Image 61 Endoscope Image 62 Endoscope Image 63 Volume Image 90 Endoscope 100 Work Area Volume 101 Insertion Point
102 遠端運動中心/RCM 110 中心環面 111 中心環面 120 心臟 121 經重構之體積影像 122 視訊流 -22- 150238.doc102 Remote Motion Center/RCM 110 Center Torus 111 Center Torus 120 Heart 121 Reconstructed Volume Image 122 Video Stream -22- 150238.doc