514515 8 5 7 8twf. doc/0 1 2 A7 B7 五、發明說明（/ ) 本發明是有關於一種獲取目標影像之邊界資料、外型 與邊界類別及影像導引之方法，且特別是有關於一種針對 超音波所攝取之腫瘤影像，獲取腫瘤之邊界資料及自動判 斷腫瘤之外型與邊界類別，加以判斷腫瘤良惡性之方法， 再利用獲取之邊界資料取得腫瘤在三度空間中之位置及 方向進而可導引器械至目標處。 （ 癌症是人類的主要死亡原因之一 9而超音波檢驗爲腫 瘤篩檢的重要工具。當超音波掃描完腫瘤後，以往皆係由 醫師根據臨床經驗觀察腫瘤外形及邊界之規則與否，以判 定腫瘤之良惡性，但由於超音波所攝取出之影像皆有雜訊 在其中，是以往往會造成醫師誤判之情形，也就是說，原 本可能是良性腫瘤，但因雜訊之影響5使得腫瘤之外形不 規則或邊界成鋸齒狀，導致醫師誤判此腫瘤爲惡性腫瘤。 以往亦有一種由外國人所提出之snake方法，可偵測 出腫瘤之輪廓’此方法需先由人爲方式沿著腫瘤之輪廓選 取一腫瘤區域，否則無法繼續運作，而由於惡性腫瘤是會 侵蝕好細胞之細胞組成，是以若爲惡性腫瘤，則應如圖1 所示其腫瘤100的輪廓是一個不規則形狀，而當醫師利用 此snake方法欲選取目標區域12〇時，必須沿著不規則外 形之腫瘤100輪廓進行描繪，是一件相當困難且不嚴謹之 作法，且當進行探測腫瘤之位置所在時，是以垂直探測方 式（如箭頭所示）進行偵測腫瘤之真實位置，但由於垂直 採測的點不夠密集，使得探測後連接所得之輪廓可能爲其 他形狀，且這樣的方式在同樣遇到雜訊時，如雜訊13〇 , 亦會造成誤判的情形，亦即以爲所探測到之雜訊爲腫瘤之 3 本紙張尺度適用中國國家標準（CNS)A4 X 297公--- --：;---------· κ-------訂--------- (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 514515 8578twf.doc/0 12 經濟部智慧財產局員工消費合作社印製 B7 五、發明說明（ >) … 一部份，造成腫瘤範圍擴大，是以此方法只能使用於具有' 規則形狀之腫瘤外形才可進行。 一 而在中華民國申請案號爲‘86106191之申請案中，亦 對某一器官進行輪廓之加強取得有所著墨，其發明名稱爲 「醫學影像之電腦化輪廓評估」，其評估步驟包括圖2之 流程步驟，即先在步驟200中，逆轉目標器官之超音波影 像而獲得負像，再在步驟202中，於正交方向成比例放大 負像而得放大影像，之後將放大且爲負像之影像與原有之 正向影像相加（步驟204)，以獲得含灰階的轉移與儲存 地形影像，附加邊緣資訊增強之技術，其中轉移與儲存地 形是照相使用之暗房技術，然後在步驟206中，二元化步 驟204之含灰階的轉移與儲存地形影像，以去除灰階，在 步驟208中，擴張與溶蝕二元化影像，獲取目標器官邊緣 輪廓之更平順外形圖，最後提供一種節略外形圖，其中擴 張與溶蝕操作和節略操作係使用影像處理套裝軟體（如， 「Optimas』得自 Bioscan) 〇 在86106191之申請案中，於步驟206中，雖然利用 所謂之二元化以去除灰階步驟來去除雜訊，但在文中只提 及係使用一種基本之數位影像操作，並未述及究爲何數位 影像操作以進行二元化操作。此86106191之申請案中， 關於解決習知之技術內容9只是將某器官之外形輪廓更淸 晰之描繪出以評估輪廓之大小，若運用在本案，只是將腫 瘤之外形輪廓大小詳加描繪出，並未對腫瘤位於人體內之 實際位置詳加指出。另外，86106191之申請案是針對攝護 腺腫大問題加以解決，但卻未對攝護腺需腫大到某一程度 4 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） -------------i--^n^-i------訂--------- (請先閱讀背面之注意事項再填寫本頁) 514515 85 78twf.doc/0 1 2 A7 B7 五、發明說明（^ ) 才需切除之技術加以提出。而關於描繪器官之輪廓外形之 技術手法，綜合上述步驟可知，其係運用許多技術才可完 成，包括暗房技術、擴張與溶蝕操作和節略操作等使用之 影像處理套裝軟體。 由於傳統的醫學超音波掃描，缺點在於醫師必需由平 面影像判斷病灶位置，其精確度不高且需豐富的經驗。在 手術時，醫師會使用超音波掃描來尋找手術後殘留的腫瘤 等病灶，但所找到的病灶常會因血水或其他組織的遮蔽而 使得醫師難以進行切除。在醫學診斷方面，影像中如果發 現有可疑組織，醫師必需再穿刺取樣然後化驗’證實是有 害的組織後才進行手術切除，而在穿刺取樣時，醫師必需 由影像中的刻度及對人體組織的知識槪略評估位置，遇到 組織太小而無法準確取樣時，醫師往往必需等待它擴大再 來取樣，對於需要即早切除的惡性腫瘤，是非常不好的處 理方式。 經濟部智慧財產局員工消費合作社印製 人請先閱讀背面之注意事1再填寫本頁) 而在以影像重建及定位導引之SPANS系統中（註： SPANS系統是由曾淸秀所發展的系統，中華民國發明專利 一^ 一* 一*五—^四號’八九年—*月一^日核准），即提升了手術 中的定位精確度，但是由於手術時會使原本的組織友病灶 移位，而SPANS系統所用的影像掃描資料是在手術前取得 的，無法即時顯示ώ手術時新的組織位置，因此醫師必需 以超音波掃描來輔助。如果能將超音波影像的資料顯示於 重建的立體影像中，則醫師能更容易快速的判斷出所尋找 的病灶，也能彌補SPANS系統無法即時更新資料的缺點。 有鑒於此，本發明提出一種獲取目標影像之邊界資 5 本纸張尺度適用中國國豕標準（CNS)A4規格（210 X 297公釐） t 514515 85 78twf.doc/0 1 2 五、發明說明（屮） (請先閱讀‘背面之注意事項再填寫本頁) .料、外型與邊界類別及影像導引定位之方法，係利用所提 出之簡易運算法，去除雜訊以得知如腫瘤在人體中之輪廓 大小，且可得知其位於人體中之位置，甚而經由比對可自 動得知而不需醫師判別即可辨識腫瘤之良惡性，但由上所 得知之位置乃是一二維空間位置，本發明可再進一步將之 形成三度空間位置，以利醫師找到實際位置及方向。 此獲取目標影像之邊界資料及外型與邊界類別之方 法，係適用於影像處理系統中，其中ROI矩形區域內含之 目標影儀，若爲檢驗腫瘤所攝取，則爲匯瘤影像。 此方法包括：取得ROI矩形區域中具有最小灰度投影 値之參考點，接著並以特定閥値二値化矩形區域，再以參 考點爲起點進行擴散，直至擴散至二値化交接點爲止，即 爲目標影像之實際範圍。之後獲取目標影像之邊界資料， 利用目標影像之中點至邊界點之距離分佈是否規則且對 稱分佈，來分類目標影像之外形5再利用邊界點之二次導 數變化是否平滑，以分類目標影像之邊界，其中分類目標. 影像之外形及邊界就是作爲腫瘤良惡性之判斷。 經濟部智慧財產局員工消費合作社印製 其中取得具有最小灰度投影値之參考點之方法包 括：分別計算矩形區域內橫軸上像素相對應之垂直灰度分 佈投影値（gray level projection, distribution in vertical direction)，分別計算矩形區域內縱軸上像素 相對應之水平灰度分佈投影値（gray level proj ect ion distribution in horizontal direction)。在依序排列 水平灰度分佈投影値及垂直灰度分佈投影値後”取得在二 維座標軸上對應最小灰度投影値之參考座標，此參考座標 6 - 本紙‘尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） 514515 8578twf.doc/012 A7 B7 經濟部智慧財產局員工消費合作·社印製 五、發明說明（夕） 即爲參考點在ROI矩形區域中之所在位置。 其中，本發明提出之二値化矩形區域之方法包括：計 算在矩形區域中之灰度平均値及灰度標準差’再利用灰度 平均値、灰度標準差及一係數之間的線性組合，取得臨界^ 灰度値。之後，將灰度値在臨界灰度値以下之像素’設爲 黑色像素，相對的’將灰度値在臨界灰度値以上之像素5 設爲白色像素，藉此將矩形區域內之所有像素分爲黑白像 素。 本發明又提出一種獲取目標影像之邊界資料之方 法，適用於超音波腫瘤影像處理系統中，此方法包括：取 得矩形區域中具有最小灰度値之參考點，接著利用上述之 二値化方法二値化矩形區域，之後以參考點爲起點進行擴 散，直至擴散至二値化交接點爲止，將腫瘤影像之實際範 圍自矩形區域中分割出，以標示已分割之腫瘤影像之數個 邊界點相對應之數個座標位置，進而得知腫瘤位置所在。 本發明再提出一種獲取目標.影像之外型類別之方 法，適用於一超音波腫瘤影像處理系統中，此方法包括： 標示腫瘤影像之邊界點對應之座標位置，以得知腫瘤位 置，利用腫瘤影像之中點至各個邊界點之距離分佈，計算 距離分佈之第一距離平均値及第一距離標準差。 ” 計算第一距離標準差及第一距離平均値之第一比 値，當第一比値小於近於零的預設値，則腫瘤影像之外形 係趨於圓形9表示腫瘤係良性腫瘤，當比値大於預設値， 則將巨標影像分割成四個象限，加以計算每一象限之第二 距離平均値及第二距離標準差。 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） --1--------·----------訂--------- (請先閱讀背面之注意事項再填寫本頁) * 514515 A7 8578twf.doc/012 五、發明說明（θ ) -—^— 計算弟一距離平均値及第二距離標準差間之第二比 値，當每-象限相對應之第二比値間皆近似_致，則目標 影像之外形係-對稱外形，表示腫瘤係良性腫瘤，但當每 -象限相對應之第二比_不—致，翻標影像之外形係 一不規則外形，表示腫瘤係惡性腫瘤。 本發明另再提出一種獲取目標影像之邊界類別之方 法，適用於-超音波腫瘤影像麵系統中，此方法包括： 在標7ΙΚ腫瘤影像之數個邊界點對應之數個座標位置，得知 腫瘤在人體之實際位置後，求得邊界點之二次導數分佈 値，而當一次導數分佈値小於預設値，則目標影像之邊界 係平滑曲線，表示腫瘤係良性腫瘤，否則在二次導數分佈 値大於預設値時，則目標影像之邊界係一不規則曲線，即 表示腫瘤係惡性腫瘤。 綜合上述’本發明即是先提出計算包含腫瘤（即腫瘤 影像）之ROI矩形區域內之垂直、水平灰度分佈投影値來 確定腫瘤之位置’接著以此位置爲起始點將腫瘤之外形分 離出並獲取腫瘤外形之邊界資料。又依據獲取之邊界資 料，由腫瘤影像之中心至邊界的距離分佈與邊界點之二次 導數變化大小來辨別腫瘤的外形與邊界的種類.。由上述之 方法即能有效將抽象的口語，如^平滑（smooth)、不規則 (i r regul ar)、橄欖球狀（ovaq )、樹葉狀（1 〇bul at ed ) 等以量化數據來表示，藉以有效將腫瘤加以分類，作爲癌 症自動診療系統之有效工具。 接著，本發明又提出一種影像定位導引方法，係應用 於影像導引系統中，此影像導引系統包括訊號收集器、超 8 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） —------1—·——-丨丨訂--------- (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 經濟部智慧財產局員工消費合作社印製 514515 8 5 78twf.doc/012 A7 137 五、發明說明（7 ) 音波換能ά及固定於超音波換能器之定位裝置，此影像導 引方法包括：擷取特徵點影像步驟、量測特徵點影像座標 步驟、量測特徵點裝置座標步驟、利用基因法則運算找出 轉換矩陣步驟、點選所欲顯示像素之步驟、顯示像素位置 於三維影像之步驟及手術導引步驟，將手術用之手術器械 模型及超音波病灶影像置於三維影像上來讓醫師知道手 術器械與病灶影像位置的相對關係。 ^ 綜合影像導引之方法，使得醫師能夠從影像的位置， 透過定位裝置的導引找到實際病灶位置及方向，可解決因 醫師不確定超音波影像中的實際位置及方向而無法順利 進行醫療行爲的狀況。 爲讓本發明之上述和其他目的、特徵、和優點能更明 顯易懂，下文特舉較佳實施例，並配合所附圖式，作詳細 說明如下= 圖式之簡單說明： 第1圖繪示的是習知snake方法探測腫瘤輪廓之方 法； 第2圖繪示的是習知案號爲86106191之申請案進行 輪廓評估之方法； 第3圖繪示的是本發明提供之一較佳實施例之一種影 像處理系統； 第4圖繪示的是本發明提供之一較佳實施例之一種獲 取目標影像之邊界資料及外型與邊界類別之方法之一流 程圖； 第5圖繪示的是本發明提供之一較佳實施例之一種獲 9 —.—,1—·——------訂--------- (請先閱讀背面之注意事項再填寫本頁) 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） 經濟部智慧財產局員工消費合作社印製 514515 85 7 8twf.doc/012 Λ7 B7 五、發明說明（$ ) 取目標影像之邊界資料之方法之一流程圖； 第6a〜6f圖繪示的是本發明依據圖5流程圖所提供之 一相對應不意圖； 第7圖繪示的是本發明提供之一較佳實施例之一種獲 取目標影像之外型種類之方法流程圖； 第8圖繪示的是本發明提供之一較佳實施例之一種獲 取目標影像之邊界類別之方法流程圖； 第9圖繪示的是本發明提供之另一較佳實施例之一種 影像導引之一示意圖； 第10圖繪示的是本發明關於第9圖之一種影像導引 之方法流程圖； 第11圖繪示的是本發明關於影像導引中方位校準之 一示意圖；以及 第12圖繪示的是本發明關於第11圖之方位校準之方 法流程圖。 標號說明514515 8 5 7 8twf. Doc / 0 1 2 A7 B7 V. Description of the invention (/) The present invention relates to a method for obtaining the boundary information, appearance and boundary type of the target image, and the method of image guidance, and particularly relates to A method for automatically acquiring tumor boundary data and automatically judging tumor shape and boundary type based on ultrasound acquired tumor images, and determining the benign and malignant tumors, and then using the obtained boundary data to obtain the position and location of the tumor in the three-dimensional space. The direction in turn can guide the instrument to the target. (Cancer is one of the leading causes of death in humans9 and ultrasound testing is an important tool for tumor screening. After ultrasound scans tumors, in the past, doctors have observed the rules of tumor appearance and boundaries based on clinical experience. Judging the benign and malignant tumors, but because the images taken by ultrasound have noise in them, it is often the case that the physician will misjudge, that is, the benign tumor may have been originally, but due to the influence of noise5 The irregular shape of the tumor or the zigzag border of the tumor caused the doctor to misjudge the tumor as a malignant tumor. In the past, there is also a snake method proposed by foreigners that can detect the contour of the tumor. This method needs to be artificially followed. Select a tumor area along the outline of the tumor, otherwise it cannot continue to operate, and because the malignant tumor is a cell composition that will erode the cells, if it is a malignant tumor, the outline of the tumor 100 should be irregular as shown in Figure 1 Shape, and when the physician uses this snake method to select the target area 120, it must be traced along the outline of the tumor 100 with an irregular shape, A very difficult and imprecise approach, and when the location of the tumor is detected, the true location of the tumor is detected by a vertical detection method (shown by the arrow), but because the points collected vertically are not dense enough, The contours of the connection after detection may be other shapes, and in the same way when noise is encountered, such as noise 13, it will also cause a misjudgment, that is, the detected noise is the 3 copies of the tumor. Paper size applies to China National Standard (CNS) A4 X 297 male --- ---:; --------- · κ ------- order --------- (please (Please read the notes on the back before filling in this page) Printed by the Consumers' Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 514515 8578twf.doc / 0 12 Printed by the Consumers ’Cooperative of the Intellectual Property Bureau of the Ministry of Economics B7 5. Description of the invention (>) As a result, the scope of the tumor has been enlarged, and this method can only be used for tumor shapes with a 'regular shape.' In the application of the Republic of China application number '86106191, contouring of an organ is also performed. Enhancing the acquisition, the invention name is " "Computerized contour evaluation of medical images". The evaluation steps include the process steps of Fig. 2. First, in step 200, the ultrasonic image of the target organ is reversed to obtain a negative image. Then in step 202, it is proportional to the orthogonal direction. Enlarge the negative image to obtain an enlarged image, and then add the enlarged and negative image to the original positive image (step 204) to obtain a grayscale transfer and storage terrain image, and add edge information enhancement technology. The transfer and storage terrain is the darkroom technology used for photography. Then in step 206, the grayscale-transferred and stored terrain images in step 204 are binarized to remove the grayscale. In step 208, the expansion and erosion are binarized. Obtain a smoother outline drawing of the edge contour of the target organ using images, and finally provide an abridged outline drawing, in which the expansion and dissolution operations and abbreviation operations are performed using an image processing software package (eg, "Optimas" from Bioscan). Application in 86106191 In step 206, although the so-called dualization is used to remove the grayscale step to remove noise, only one type is mentioned in the text. This digital image of the operation, the study did not address the why digital imaging operations for dualization operation. In this application of 86106191, the technical content about solving conventional knowledge 9 is only to describe the outline of an organ more clearly to evaluate the size of the contour. If used in this case, it is only to describe the size of the contour of the tumor in detail. The exact location of the tumor in the human body was not specified. In addition, the application of 86106191 is aimed at solving the problem of prostate enlargement, but it does not require the prostate to enlarge to a certain degree. 4 The paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm). ------------- i-^ n ^ -i ------ Order --------- (Please read the notes on the back before filling this page) 514515 85 78twf.doc / 0 1 2 A7 B7 V. The description of the invention (^) is only required to be removed. Regarding the technical methods of drawing the contours of the organs, it can be seen from the above steps that it can be completed by using many technologies, including darkroom technology, expansion and erosion operations, and abbreviated operations. Due to the traditional medical ultrasound scan, the disadvantage is that the physician must judge the location of the lesion from the planar image, its accuracy is not high and it needs rich experience. During surgery, doctors use ultrasound scans to find lesions such as tumors that remain after surgery. However, the lesions found are often difficult to remove due to the obscuration of blood or other tissues. In medical diagnosis, if suspicious tissue is found in the image, the doctor must perform a puncture sample and then analyze the tissue that has been confirmed to be harmful before performing surgical resection. During the puncture sampling, the doctor must use the scale in the image and the effect on the human tissue. Knowledge estimates the location. When the tissue is too small to accurately sample, the physician often has to wait for it to expand and then take a sample. This is a very bad treatment for malignant tumors that need to be removed early. Printed by the Intellectual Property Bureau employee consumer cooperative of the Ministry of Economic Affairs, please read the note on the back 1 and then fill out this page) and in the SPANS system guided by image reconstruction and positioning (Note: The SPANS system is a system developed by Zeng Yixiu, The invention patent of the Republic of China (1 ^ 1 * 1 * 5— ^ 4 '1989— * Monthly ^ approved on the 1st), which improves the accuracy of positioning during surgery, but because the original tissues will be moved during surgery The image scanning data used by the SPANS system is obtained before the operation, and the new tissue position during the free operation cannot be displayed in real time, so the doctor must use ultrasonic scanning to assist. If the ultrasound image data can be displayed in the reconstructed stereo image, the physician can more easily and quickly determine the lesions he is looking for, and it can also make up for the shortcomings that the SPANS system cannot update the data in real time. In view of this, the present invention proposes a method to obtain the border image of the target image. The paper size is applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm) t 514515 85 78twf.doc / 0 1 2 V. Description of the invention (屮) (Please read the “Notes on the back side before filling out this page”.) The methods of material, shape and boundary, and image-guided positioning are based on the proposed simple algorithm to remove noise to learn like tumors The size of the contour in the human body, and it can be known where it is located in the human body. Even through comparison, it can be automatically identified without the need for a doctor to discern the benign and malignant tumors. However, the position known from the above is one-dimensional The spatial position can be further formed into a three-dimensional spatial position by the present invention, so that the physician can find the actual position and direction. This method of obtaining the boundary data and the shape and boundary type of the target image is suitable for use in image processing systems. The target camera contained in the rectangular area of the ROI is a tumor image if it is taken by a tumor. This method includes: obtaining a reference point with the smallest gray-scale projection in a rectangular region of the ROI, and then binarizing the rectangular region with a specific valve, and then diffusing the reference point as a starting point until it diffuses to the binarized intersection point, It is the actual range of the target image. Then, the boundary data of the target image is obtained, and whether the distance distribution between the points in the target image and the boundary points is regular and symmetrical is used to classify the external shape of the target image 5 and then use the smoothness of the second derivative of the boundary points to classify the target image. Boundary, which is the classification target. The shape and boundary of the image are used to judge the benign and malignant tumors. The method for obtaining the reference point with the smallest gray-scale projection by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs includes: calculating the gray level projection, distribution in vertical direction) to calculate the gray level distribution distribution in the horizontal direction corresponding to the pixels on the vertical axis in the rectangular area. After sequentially arranging horizontal gray distribution projections and vertical gray distribution projections ", obtain the reference coordinates corresponding to the minimum gray projection on the two-dimensional coordinate axis. This reference coordinate 6-this paper's scale applies to the Chinese National Standard (CNS) A4 Specifications (210 X 297 mm) 514515 8578twf.doc / 012 A7 B7 Printed by the Consumers ’Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs · Printed by the Society V. Description of the Invention (Evening) It is the position of the reference point in the rectangular area of the ROI. The second method for rectifying a rectangular region proposed by the present invention includes: calculating a gray average 値 and a gray standard deviation ′ in the rectangular region, and then using a linear combination of the gray average 値, the gray standard deviation, and a coefficient, Obtain a critical ^ gray level 之后. After that, set the pixels of gray level 値 below the critical gray level 设为 to black pixels, and the relative 'set the pixels 5 of gray level 値 above the critical gray level to white pixels. All pixels in a rectangular area are divided into black and white pixels. The present invention also proposes a method for obtaining boundary data of a target image, which is applicable to an ultrasonic tumor image processing system. This method package Including: Obtaining the reference point with the smallest gray level in the rectangular area, and then using the above-mentioned two dilation methods to dilate the rectangular area, and then use the reference point as a starting point to spread until the diffusion reaches the dimorphization junction, and the tumor The actual range of the image is segmented from the rectangular area to mark the coordinate positions corresponding to the several boundary points of the segmented tumor image, and then to know the location of the tumor. The present invention further proposes an acquisition target. Image type The category method is applicable to an ultrasonic tumor image processing system. The method includes: marking the coordinate position corresponding to the boundary point of the tumor image to know the tumor position, and using the distance distribution between the midpoint of the tumor image and each boundary point, Calculate the first distance average 値 and the first distance standard deviation of the distance distribution. "Calculate the first ratio 第一 of the first distance standard deviation and the first distance average 値. When the first ratio 値 is less than a preset 値 that is closer to zero, then The shape of the tumor image tends to be round. 9 indicates that the tumor is a benign tumor. When the ratio is greater than the preset value, the giant target image is divided into four images. , To calculate a second distance in each quadrant of the second distance and the average standard deviation Zhi. This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) --1 -------- · ---------- Order -------- -(Please read the notes on the back before filling in this page) * 514515 A7 8578twf.doc / 012 V. Description of the Invention (θ)--^ — Calculate the second ratio between the average distance between the first distance and the standard deviation of the second distance Alas, when the second ratio corresponding to each quadrant is approximately the same, the outer shape of the target image is symmetrical, indicating that the tumor is a benign tumor, but when the second ratio corresponding to each quadrant is not, the same The irregular shape outside the relabeled image indicates that the tumor is a malignant tumor. The present invention further proposes a method for obtaining the boundary type of the target image, which is suitable for use in an ultrasound tumor image plane system. The method includes: knowing a tumor at a plurality of coordinate positions corresponding to a number of boundary points of a tumor image of a 7IK tumor After the actual position of the human body, the second derivative distribution 边界 of the boundary points is obtained, and when the first derivative distribution 値 is smaller than the preset 値, the boundary of the target image is a smooth curve, indicating that the tumor is a benign tumor, otherwise it is distributed in the second derivative When 値 is larger than the preset 値, the boundary of the target image is an irregular curve, indicating that the tumor is a malignant tumor. In summary, the present invention is to first propose the calculation of the vertical and horizontal gray distribution projections in a rectangular region of the ROI containing the tumor (ie, tumor image) to determine the position of the tumor, and then use this position as the starting point to separate the tumor shape. And obtain the boundary data of tumor shape. According to the obtained boundary data, the distance distribution from the center of the tumor image to the boundary and the magnitude of the second derivative of the boundary point are used to distinguish the shape of the tumor and the type of boundary. The above methods can effectively express abstract spoken words, such as ^ smooth (smooth), irregular (ir regul ar), football-like (ovaq), leaf-like (10 bul at ed), etc. Effectively classify tumors as an effective tool for automatic cancer diagnosis and treatment systems. Then, the present invention proposes an image positioning and guidance method, which is applied to an image guidance system. The image guidance system includes a signal collector, and more than 8 paper sizes are applicable to the Chinese National Standard (CNS) A4 specification (210 X 297). Mm) —------ 1— · ——- 丨 丨 Order --------- (Please read the precautions on the back before filling out this page) Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs and Economics 514515 8 5 78twf.doc / 012 A7 137 V. Description of the invention (7) Sound wave transducer and positioning device fixed to the ultrasonic transducer, this image guidance method includes : Steps of capturing feature point image, measuring feature point image coordinate step, measuring feature point device coordinate step, using gene rule calculation to find transformation matrix step, step of selecting desired display pixel, displaying pixel position in three-dimensional image Steps and surgical guidance steps, placing the surgical instrument model and ultrasonic lesion image on the three-dimensional image to let the physician know the relative relationship between the surgical instrument and the lesion image position. ^ The method of integrated image guidance enables physicians to find the actual location and direction of the lesion from the position of the image and through the guidance of the positioning device. Condition. In order to make the above and other objects, features, and advantages of the present invention more comprehensible, the following describes the preferred embodiments in detail with the accompanying drawings as follows: = Brief description of the drawings: FIG. 1 Shows the method of detecting the tumor contour by the conventional snake method; Figure 2 shows the method of contour evaluation by the conventional application No. 86106191; Figure 3 shows a preferred implementation provided by the present invention An example of an image processing system; FIG. 4 shows a flowchart of a method for obtaining the boundary information, shape and boundary type of a target image according to a preferred embodiment of the present invention; FIG. 5 shows It is one of the preferred embodiments provided by the present invention. 9 —.—, 1— · ———— Order --------- (Please read the precautions on the back before filling out this Page) This paper size is in accordance with China National Standard (CNS) A4 (210 X 297 mm) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 514515 85 7 8twf.doc / 012 Λ7 B7 V. Description of the invention ($) Target Flowchart of one of the methods of image boundary data; Figures 6a ~ 6f show It is one of the corresponding intentions provided by the present invention according to the flowchart of FIG. 5; FIG. 7 is a flowchart of a method for acquiring a target image external type provided by a preferred embodiment of the present invention; FIG. 8 FIG. 9 is a flowchart of a method for obtaining a boundary type of a target image according to a preferred embodiment of the present invention; FIG. 9 illustrates one of image guidance of another preferred embodiment of the present invention. Fig. 10 shows a flowchart of a method for image guidance according to Fig. 9 of the present invention; Fig. 11 shows a schematic diagram of orientation calibration in image guidance according to the present invention; and Fig. 12 shows Shown is a flowchart of a method for azimuth calibration according to FIG. 11 of the present invention. Label description
100 :腫瘤 U 120 :目標區域 130 :雜訊 300 :影像處理系統 302 :資料庫 304 ·影像處理裝置 310 :影像饋入器 600 : ROI矩形區域 900 :訊號收集器 10 (請先閱讀背面之注意事項再填寫本頁) -ϋ I I I n 一 10, a ·1 1 ai_i Φ. 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） 514515 8578twf.doc/012 A7 B7 五、發明說明（) 902，910 :1定位裝置 904 :超音波換能器 906 :樣板 908 :手術器械 911 :影像處理裝置 912 :手術台 步驟200至步驟210係習知輪廓評估之流程步驟 步驟400至步驟412係本發明之一較佳實施步驟 步驟500至步驟514係本發明另一較佳實施步驟 步驟700至步驟724係本發明又一較佳實施步驟 • » 步驟800至步驟808係本發明再一較佳實施步驟 步驟1000至步驟1210係本發明另一較佳實施步驟 較佳實施例 請參照第3圖，其繪示的是依照本發明所提供之一種 影像處理系統300，在此影像處理系統300中包含儲存影 像資料用之資料庫302及對資料庫302中儲存之影像資料 進行處理分析之影像處理裝置304。其中影像饋入器310 即是將攝取所得之影像輸入至資料庫302,以供影像處理 裝置304進行後續處理及分析。 、 請參照第4圖，其繪示的是依照本發明一較佳實施例 的一種獲取目標影像之邊界資料及外型與邊界類別之方 法，此方法即是應用於上述之影像處理系統300中首先 當欲分析處理儲存於資料庫302中影像資料時’先自資料 庫302中取出一影像資料，然後由醫師或技術分析人員在 此影像資料中選取內含目標影像之ROI矩形區域（步驟 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） ---------.———訂—I.— (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 514515 A7 85 78twf.doc/012 五、發明說明（P ) 400)，不同於習知snake方法，需圍繞目標影像之周圍 進行影像選取’其中此矩形區域係由數個像素（ Plxel) 所組成，每一個像素皆具有灰度値且皆有二維座標位置與 其相對應。 . ’ 此獲取目標影像之邊界資料及外型與邊界類別之方 法包括：取得ROI矩形區域中具有最小灰度値之參考點（步、 驟402 )，也就是先確認目標影像中之一核心所在，之後 在步驟404中，二値化矩形區域，即將矩形區域內之像素 分爲黑白兩種色度，在步驟406中，以先前所得到之參考 點爲起點進行擴散，直至擴散至二値化交接點爲止，也就 是以參考點爲起點擴散範圍直至有黑白兩像素之交接點 爲止，即爲目標影像之實際範圍，此時即已將目標影像自 矩形區域中分離出。 之後即在步驟408中獲取目標影像之邊界資料，此邊 界資料係爲目標影像所有邊界點的座標位置（x，y)。當 已得知目標影像所有邊界點的座標位置（x，y)後，就在 步驟410中，利用目標影像之中點至所有邊界點之距離分 佈狀況是否規則且對稱分佈，以分類目標影像之外形是屬 於圓形、橄欖球形、樹葉狀或不規則型，其中在資料庫302 中即已先記載關於外形之定義，是以當由距離分佈狀況與 一預設値相比後，即可從資料庫302定義之外形來判斷出 目標影像之外形種類。同樣的，在資料庫302中亦有記載 關於邊緣之定義，是以當在步驟412中，利用邊界點之二 次導數變化與一預設値相比後即可從資料庫302定義之邊 緣來判斷出目標影像之邊緣是否平滑，以分類目標影像之 12 --------it-illi —訂--------- Φ— (請L閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印制π 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） 514515 85 78twf.doc/012 Λ7 B7 五、發明說明（// ) 邊界。 (請先閱讀背面之注意事項再填寫本頁} 其中上述之方法可適用於所有有形之物之擷取及判 斷，不管是有生命或無生命之物，凡是欲從背景中取得〜 目標影像，進而判斷目標影像之外形及邊緣，皆可使用本 發明進行影像擷取及判斷，例如：欲從超音波掃描之影像 中取得關於腫瘤之位置、大小及是否爲良惡性之判斷，胚 胎之成長情形等等情形，皆可使用本發明所提供之方法加 以獲取病灶之情況5且所獲取之資料皆爲量化之數據資 料，使得所分析出之結果皆是客觀結論而非醫師診斷般的 主觀結，論，因此正確度亦較高。 是以若是要診斷關於腫瘤之良惡性，則上述之影像饋 入器310應是超音波探測器，而儲存於資料庫302之影像 資料即爲超音波腫瘤影像。 以下即以自動分析腫瘤之位置、大小及腫瘤之良惡性 診斷爲例作技術上之更詳細說明。 經濟部智慧財產局員工消費合作社印製 請合倂參照第5圖及第6a至6f圖，其中第5圖繪示 的是本發明所提供之一種獲取目標影像之邊界資料之方 法。首先，如上所述，先自資料庫302中取出一影像資料， 假設爲圖6a，以肉眼觀看此圖，其實即可知道目標影像之 大約輪廓，但由不同的#師來解讀此腫瘤之良惡性，卻可 能因其中參雜之雜訊而有不同之結論，是以必須量化此目 標影像以取得客觀資訊，因此醫師先在此影像資料中選取 內含腫瘤影像之矩形區域600 (如6b圖），也就是以一個 矩形方塊圍住此腫瘤影像.，然後取得矩形區域600中具有 最小灰度値之參考點（步驟402)，也就是腫瘤的核心所 13 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） " ' ' ^ 經濟部智慧財產局員工消費合作社印製 514515 8 5 7 8twf. doc/O 1 2 Λ7 B7 五、發明說明（产） 在，由於以超音波照射結果，是以灰階方式呈現影像亮度 値，而由臨床得知腫瘤的核心是在腫瘤中顏色最深者，是 以取得之最小灰度値參考點即是腫瘤核心。 取得之步驟包括：步驟500，分別計算矩形區域600 內位於橫軸上之像素相對應之垂直灰度分佈投影値，如圖 6c所示，X軸係表示矩形區域600內的橫軸上之像素位 置，而y軸係表示在同一 X軸位置上之數個像素之灰度値 總合，即爲垂直灰度分佈投影値，同樣的亦分別計算矩形 區域600內縱軸上像素相對應之水平灰度分佈投影値，如 圖6d所示，X軸係表示矩形區域600內的縱軸上之像素位 置，而y軸係表示在同一 y軸位置上之數個像素之灰度値 總合，即爲水平灰度分佈投影値，之後在步驟502中，排 列水平灰度分佈投影値及垂直灰度分佈投影値，從其中選 取最小灰度分佈投影値，以取得在二維座標軸上對應最小 灰度値之參考座標，由圖6c及6d中可知對應最小灰度値 之參考座標爲（ 60, 35 )，參考座標（ 60, 35 )即爲參考 點在矩形區域600中之所在位置。 之後在圖4之步驟404中所進行二質化動作，即是將 矩形區域600內之所有像素分爲黑白兩種色度，其步驟包 括：步驟504，計算在矩形區域600中所有像素之平均灰 度値及灰度標準差，其運算式如下： (x,少）=v Σ /(x i，y - J·、 ^ w (x,j)ew (2-3) (x，少）-九 Σ l/(x，少）m(x，v)l (2-4) 14 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） ----------.11 --------訂--------- (請先閱讀背面之注意事項再填寫本頁) 514515 8578twf.doc/〇12 A7 B7 五、發明說明（/6 ) 其中，/U，J〇係表示在（X，y)位置像素之灰度値； 係表示在矩形區域內之所有像素數量； 、 係表示矩形區域內之平均灰度値；以及 duO係表示矩形區域內之灰度標準差％ 之後，需利用灰度平均値、灰度標準差 及係數之間的線性組合以取得臨界灰度値，其線性組合式 爲： g(x，少）= m(x，3；) 土 Α：χσ(χ,γ) (2-5) 其中，g(x，j〇係表示臨界灰度値，而J係表示該係數。 其中，需在步驟506中選取一適合係數J代入式子 (2-5)以決定最佳臨界灰度値以1，3〇，所謂適合之係數尤， 係指經由其所決定之最佳臨界灰度値3(1，3〇，在二値化的 狀況下，黑色像素所包含之範圍需包含腫瘤且黑色像素所 形成之黑色面積變化量需爲最小，選擇之方式是採取試誤 法。 舉例來說，由於係數π之範圍係界定於0.1至0.9之 間，是以先以係數Γ =0.1代入式子（2-5)得到臨界灰度値 gi(u)，再進行步驟508之二値化矩形區域600 5即灰度 値在臨界灰度値以上之像素設爲白色像素，灰度値在臨界 灰度値以下之像素設爲黑色像素，至此發現在係數j=〇.l 之條件下之黑色像素面積釋95，再以係數J =〇 . 2代入式 子（2-5)得到臨界灰度値g2(x，j；)，再進行步驟508之二値 化矩形區域600，發現在此條件下之黑色像素面積爲93， 之後再繼續以係數Γ =0。3代入式子（2 - 5)得到臨界灰度値 15 本紙張尺度適用中國國家標準（CNS)A‘l規格（210 X 297公釐） ---------.-----------訂··--*---I (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 經濟部智慧財產局員工消費合作社印製 W4515 8578twf.doc/012 五、發明說明（Γ>) 客3(χ，7)，再進行步驟508之二値化矩形區域600，發現在 此條件下之黑色像素面積爲70,此時可知係數Γ在0.2時 有最小之面積變化量（步驟510)，表示此條件下所產生 之灰度臨界値是一最佳灰度臨界値。 在步驟512中，即以先前所得到之參考點爲起點進行 擴散，直至有黑白兩像素之交接點爲止，即爲目標影像之 實際範圍，此時即已將目標影像自矩形區域中分離出，其 中擴散之方法可使用區域成長法比較二値化後之灰度 値，以取得目標影像之實際範圍。 之後即在步驟514中獲取目標影像之邊界資料，此邊 界資料係爲目標影像所有邊界點的座標位置（x，y)，其 中獲取目標影像之邊界資料之方法可利用梯度偵測法 (sobel method)偵測出邊界資料，其運算式如下： Δ2/(χ，加A /(x+l，少)-Δ /(X,加/(x+l，少)-2/(x，y)+/(x-1，少） (2.6) A2f(x,y) = A /(χ,γ+ϊ)-Α (2.7) 少少 y 其中，當Δ2/(χ，3；)χΔ2/(χΗ>0<〇，表示（X，y )位置是腫瘤區域100: tumor U 120: target area 130: noise 300: image processing system 302: database 304 · image processing device 310: image feeder 600: ROI rectangular area 900: signal collector 10 (Please read the note on the back first Please fill in this page for further information) -ϋ III n a 10, a · 1 1 ai_i Φ. This paper size is applicable to China National Standard (CNS) A4 (210 X 297 mm) 514515 8578twf.doc / 012 A7 B7 V. Invention Explanation (902, 910: 1 positioning device 904: ultrasonic transducer 906: template 908: surgical instrument 911: image processing device 912: operating table steps 200 to 210 are steps of the conventional contour evaluation process step 400 to step 412 is a preferred embodiment of the present invention. Steps 500 to 514 are another preferred embodiment of the present invention. Steps 700 to 724 are yet another preferred embodiment of the present invention. »» Steps 800 to 808 are another of the present invention. The preferred implementation steps 1000 to 1210 are another preferred implementation step of the present invention. For a preferred embodiment, please refer to FIG. 3, which shows an image processing system 300 provided according to the present invention. Here, the image processing system 30 0 includes a database 302 for storing image data and an image processing device 304 for processing and analyzing the image data stored in the database 302. The image feeder 310 inputs the captured images to the database 302 for subsequent processing and analysis by the image processing device 304. Please refer to FIG. 4, which shows a method for obtaining the boundary data, the shape and the boundary type of the target image according to a preferred embodiment of the present invention. This method is applied to the image processing system 300 described above. First, when you want to analyze and process the image data stored in the database 302, 'retrieve an image data from the database 302 first, and then a physician or technical analyst selects a ROI rectangular area containing the target image in this image data (step this Paper size applies Chinese National Standard (CNS) A4 (210 X 297 mm) ---------.——— Order—I.— (Please read the precautions on the back before filling this page) Economy Printed by the Consumer Cooperative of the Ministry of Intellectual Property Bureau 514515 A7 85 78twf.doc / 012 V. Description of Invention (P) 400). Unlike the conventional snake method, image selection is required around the target image. It is composed of several pixels (Plxel), each pixel has a gray scale, and each has a two-dimensional coordinate position corresponding to it. . 'This method of obtaining the boundary data, shape and boundary type of the target image includes: obtaining the reference point (step, step 402) with the smallest gray level in the rectangular area of the ROI, which is to first confirm that one of the core of the target image is Then, in step 404, the rectangular region is binarized, that is, the pixels in the rectangular region are divided into black and white chroma. In step 406, the reference point obtained previously is used as a starting point for diffusion until it is diffused to the binarization. Up to the crossover point, that is, the reference point is used as the starting point to spread the range until there is a crossover point between black and white pixels, which is the actual range of the target image. At this time, the target image has been separated from the rectangular area. After that, the boundary data of the target image is obtained in step 408. The boundary data is the coordinate positions (x, y) of all boundary points of the target image. When the coordinate positions (x, y) of all boundary points of the target image have been known, in step 410, whether the distance distribution between the points in the target image and all the boundary points is regular and symmetrical is used to classify the target image. The shape is round, rugby-shaped, leaf-shaped, or irregular. The definition of shape has been recorded in the database 302. When the distance distribution is compared with a preset frame, the shape can be changed from The database 302 defines a contour to determine the type of the target image contour. Similarly, the definition of the edge is also recorded in the database 302. When the second derivative change of the boundary point is compared with a preset value in step 412, the edge defined by the database 302 can be obtained. Determine whether the edges of the target image are smooth, and classify 12 of the target image -------- it-illi —Order --------- Φ— (Please read the notes on the back and fill in this Page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs π This paper size applies to the Chinese National Standard (CNS) A4 (210 X 297 mm) 514515 85 78twf.doc / 012 Λ7 B7 V. Description of the invention (//) Boundary . (Please read the precautions on the back before filling out this page} The above method can be applied to the capture and judgment of all tangible things, whether they are living or inanimate things, anyone who wants to obtain from the background ~ target image, Furthermore, the shape and edge of the target image can be judged. The invention can be used for image capture and judgment. For example, to obtain the position, size, and benign and malignant judgment of the tumor from the ultrasound scanned image, and the growth of the embryo. In other cases, the method provided by the present invention can be used to obtain the condition of the lesion 5 and the obtained data are quantified data, so that the results analyzed are objective conclusions rather than subjective conclusions like a doctor's diagnosis. Therefore, if the benign and malignant tumors are to be diagnosed, the above-mentioned image feeder 310 should be an ultrasonic detector, and the image data stored in the database 302 is an ultrasonic tumor image. The following is a more detailed technical description using the automatic analysis of tumor location, size, and benign and malignant diagnosis of the tumor as examples. Printed by the production bureau employee consumer cooperative, please refer to Figure 5 and Figures 6a to 6f, where Figure 5 shows a method for obtaining the boundary data of the target image provided by the present invention. First, as described above, First, take an image data from the database 302. Assume it is Figure 6a. View this image with the naked eye. In fact, you can know the approximate outline of the target image. However, different ### students can interpret the benign and malignant nature of this tumor. The mixed noise has different conclusions. The target image must be quantified to obtain objective information. Therefore, the doctor first selects a rectangular area 600 (such as Figure 6b) containing the tumor image in this image data. A rectangular box surrounds the tumor image. Then, the reference point with the smallest gray scale in the rectangular area 600 is obtained (step 402), which is the core of the tumor. 13 The paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) " '' ^ Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 514515 8 5 7 8twf. Doc / O 1 2 Λ7 B7 V. Description of the invention (production) As a result, the brightness of the image is presented in grayscale, and it is clinically known that the core of the tumor is the darkest color in the tumor, and the minimum gray level obtained is the reference point which is the core of the tumor. The obtaining steps include: Step 500 Calculate the vertical gray distribution projections corresponding to the pixels on the horizontal axis in the rectangular region 600, as shown in FIG. 6c. The X-axis system represents the pixel position on the horizontal axis in the rectangular region 600, and the y-axis system represents The grayscale sum of several pixels on the same X-axis position is the vertical grayscale distribution projection. Similarly, the horizontal grayscale distribution projections corresponding to the pixels on the vertical axis in the rectangular region 600 are also calculated. As shown in Fig. 6d, the X-axis system represents the pixel positions on the vertical axis in the rectangular region 600, and the y-axis system represents the gray-scale sum of several pixels at the same y-axis position, which is the horizontal gray-scale distribution projection. Then, in step 502, arrange the horizontal gray distribution projection 値 and the vertical gray distribution projection 値, and select the minimum gray distribution projection 値 from them to obtain the reference position corresponding to the minimum gray 値 on the two-dimensional coordinate axis. Seen from FIG. 6c and 6d corresponding to the minimum gray Zhi reference coordinates (60, 35), the location of the reference coordinates (60, 35) with the reference point in the rectangular area 600. Then, the dimorphization operation performed in step 404 of FIG. 4 is to divide all pixels in the rectangular area 600 into black and white chroma. The steps include: Step 504, calculate the average of all pixels in the rectangular area 600. The gray scale 値 and gray standard deviation are calculated as follows: (x, less) = v Σ / (xi, y-J ·, ^ w (x, j) ew (2-3) (x, less)- Nine Σ l / (x, less) m (x, v) l (2-4) 14 This paper size is applicable to China National Standard (CNS) A4 (210 X 297 mm) --------- -.11 -------- Order --------- (Please read the notes on the back before filling this page) 514515 8578twf.doc / 〇12 A7 B7 V. Description of the invention (/ 6 ) Among them, / U, J0 represents the gray scale 像素 of the pixel at the (X, y) position; represents the number of all pixels in the rectangular area;, represents the average gray scale 内 in the rectangular area; and duO represents After the gray standard deviation% in the rectangular area, a linear combination of gray average 値, gray standard deviation and coefficients is needed to obtain the critical gray 値. The linear combination is: g (x, less) = m (x, 3;) Soil A: χσ (χ, γ) (2-5) where g (x, j0 represents the critical gray scale 値, and J represents the coefficient. Among them, a suitable coefficient J needs to be selected in step 506 and substituted into the formula (2-5) to determine the optimal critical gray scale 値 to 1 , 30, the so-called suitable coefficient especially refers to the optimal critical gray level determined by it 値 3 (1,30, under the condition of binarization, the range of black pixels must include tumors and black pixels The amount of change in the black area formed needs to be minimal, and the method of choice is to use trial and error. For example, because the range of the coefficient π is defined between 0.1 and 0.9, the coefficient Γ = 0.1 is first substituted into the formula ( 2-5) Obtain the critical gray level 値 gi (u), and then proceed to step 508bis to transform the rectangular area 600 5. That is, the pixels whose gray level is above the critical gray level are set as white pixels, and the gray level is above the critical gray level. The following pixels are set as black pixels. So far, it is found that the area of the black pixels under the condition of the coefficient j = 0.1 is 95, and then the coefficient J = 0.2 is substituted into the formula (2-5) to obtain the critical gray level 値 g2. (x, j;), and then perform step 508bis to rectify the rectangular area 600, and find the black pixel area under this condition 93, and then continue to use the coefficient Γ = 0. 3 Substitute into the formula (2-5) to get the critical gray level 値 15 This paper size applies the Chinese National Standard (CNS) A'l specification (210 X 297 mm) --- ------.----------- Order ··-* --- I (Please read the precautions on the back before filling out this page) Printed by W4515 8578twf.doc / 012 from the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economics and Industry V. Description of the Invention (Γ >) Customer 3 (χ, 7), and then proceed to step 508bis to convert the rectangular area 600, and found that under this condition The black pixel area is 70. At this time, it can be known that the coefficient Γ has the smallest area change amount at 0.2 (step 510), which indicates that the gray threshold 値 generated under this condition is an optimal gray threshold. In step 512, the reference point obtained previously is used as a starting point for diffusion, until the intersection point of the black and white pixels is the actual range of the target image. At this time, the target image has been separated from the rectangular area. The diffusion method can use the area growth method to compare the gray scale after binarization to obtain the actual range of the target image. Then, in step 514, the boundary data of the target image is obtained. The boundary data is the coordinate positions (x, y) of all boundary points of the target image. The method for obtaining the boundary data of the target image can use the gradient detection method (sobel method). ) The boundary data is detected, and its calculation formula is as follows: Δ2 / (χ, plus A / (x + l, less) -Δ / (X, plus / (x + l, less) -2 / (x, y) + / (x-1, less) (2.6) A2f (x, y) = A / (χ, γ + ϊ) -Α (2.7) less or less y where Δ2 / (χ, 3;) χΔ2 / (χΗ > 0 < 〇, indicating that the (X, y) position is the tumor area
XX 在X方向上之邊緣點，同樣的9當 y y 表示U，y )位置是腫瘤區域在y方向上之邊緣點。 以上關於獲取目標影像之邊界資料之方法中’係將腫 瘤之核心及邊緣點數據化，以利後續處理’而數據化之資 料除了可知腫瘤之位置外5且可將腫瘤分離出以知道腫瘤 之實際範圍大小（如圖6e至60。雖然可由上述得到之腫 瘤位置及實際範圍大小以利醫師在作切除時方便得知正 16 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） ---丨丨丨丨丨丨·-丨·111訂-!i丨丨 (請先閱讀背面之注意事項再填寫本頁) 514515 85 78twf.doc/012 A7 B7 五、發明說明（ίΓ) 確資訊，但判斷腫瘤之良惡性仍是由醫師主觀來判斷。 (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 爲了避免醫師誤判，本發明又提出一種獲取目標影像 之外型類別之方法，在由上述之方法標示出腫瘤影像之邊 界點座標位置（步驟700 )後，利用腫瘤影像之中點至各 個邊界點之距離分佈（步驟702 )，計算距離分佈之距離 平均値meansl及距離標準差varl (步驟704 )，取得距離 平均値1^&!^1及距離標準差^1*1間的比値&=^^1*1/1^&1^1 (步驟706 )，當在步驟708中，比値a小於預設値，假設 爲0.21，表示腫瘤影像之外形係趨於圓形或是橄欖球狀 (步驟710)，也就是腫瘤係一良性腫瘤。而當在步驟708 中，比値a大於預設値=0.21，則在步驟712中先尋找腫瘤 影像之中心點及長短軸，將目標影像依據中心點及長短軸 分割成四個象限（在步驟7、14)，計算每一個象限之距離 平均値（在步驟716)，分別爲means2、means3、means4、 means5,以及距離標準差，分別爲var2、var3 、var4、var5， 再在步驟718中，各自取得四象限之比値，分別爲b= var2/means2 ^ c=var3/means3 ^ d=var4/means4 ' e= var5/means5，當比値b、c、d、e之値在步驟720中判銜爲 一致時’表示腫瘤之外形係一對稱之外形（步驟722 )， 如樹葉狀，係一良性腫瘤^否則表示腫瘤之外形係一不規 則外形（步驟724)，如圖6f所示，係一惡性腫瘤。 / 本發明另提出一種獲取目標影像之邊界類別之方 法’也是判斷腫瘤良惡性之依據，在依據上述之獲取邊界 資料方法中，標示出腫瘤影像之邊界點座標位置後（步驟 8〇〇) ’求得邊界點之二次導數分佈値（步驟802)，所謂 17 本紙張尺度適用中國國家標準（CNS)A4規格（21〇 x 297公釐） 經濟部智慧財產局員工消費合作社印製 514515 85 78twf.doc/012 A7 〆 B7 五、發明說明（tb ) 的二次導數分佈値即是將相鄰兩點之斜率相加後再除以 所有邊界點數量，即爲二次導數變化量，當在步驟804中 判斷出二次導數分佈値小於預設値，假設爲21，則表示目 標影像之邊界係一平滑曲線（步驟806)，也就是說腫瘤 係良性腫瘤，否則當二次導數分佈値大於預設値’則目標 影像之邊界係一不規則曲線（步驟808)，表示腫瘤係惡 性腫瘤。 其中，在上述之外形分類及邊緣分類方法中，判斷腫 瘤良惡行之預設値皆是臨床上之一數據。. 綜上所述，關於獲取目標影像之邊界資料及外型與邊 界類別之方法的優點包括： 1. 去除雜訊。 2. 確定目標影像之位置。 3 β可將目標影像自背景中分離出。 4。獲取目標影像外形之數據化邊界資料。 5·辨別目標影像的外形與邊界的種類，如_標影像係 一腫瘤區塊，則可由辨別出之外形與邊界的種類判斷腫瘤 之良惡性。 g靑合倂爹考弟9及第10圖，其繪示的是本發明另又 • 提供之一較佳實施例的一種影像導引方法，係由圖9之訊 號收集器900、定位裝置902及910、超音波換能器9〇4 (例如一般超音波掃描所用之超音波探頭）及上述所提供 之影像處理裝置911所構成’其中定位裝置91 〇係配置方々 一般的手術器械908，如手術刀、手術鉗等之上，定位_ 置902係配置於超音波換能器904之上，以將這些元 18 本紙張尺度適用中國國家標準（CNS)A4規格（21〇 X 297公爱) ' ------- (請先閱讀背面之注意事項再填寫本頁) ---------訂--------- 經濟部智慧財產局員工消費合作社印製 514515 8578twf.doc/012 A7 •137 五、發明說明（Q) 方位經由訊號收集器900傳送給影像處理裝置911 (即備 有本發明進行影像處理及影像導引程式之電腦）處理，當 影像處理裝置911整合影像及方位資料後，顯示導引畫面 以幫助醫師了解手術器械908及病灶的相對方位關係。 在圖10之流程圖中，先在步驟1000中讀取樣板影像’ 經由步驟1002對樣板影像進行方位校準後即可進行手術 導引步驟（步驟1004)，其中步驟1002之方位校準步驟係 由基因法則（Gene t i c a 1 go r i t hm)運算式找出影像座標與 裝置座標的轉換矩陣，之後會對方位校準有更詳細之描 述。在接續之一系列手術導引步驟中，先在步驟1〇〇6中 掃描病灶影像，即醫師會以超音波掃描來搜尋病灶’並在 找到病灶後擷取超音波影像，擷取影像的同時系統會自動 記錄超音波探頭（即超音波換能器904)上的定位裝置902 座標値，如此影像上的任何一個像素點座標都可轉換到裝 置座標。 在步驟1008的點選像素步驟中，讓醫師從超音波影像 中點選希望顯示在步驟1010之導引畫面上的像素’系統 會一一記錄被點選的像素，並依照擷取影像時的超音波換 能器904位置（即定位裝置902座標値）及方位校準步驟 中所得到的轉換矩陣，將像素點座標轉換成裝置座標’最 後將此座標記錄下來。其中，影像與超音波換能器904的 裝置座標可以在點選動作完作後刪除。 在步驟1012之顯示像素位置步驟中.，依上一步驟所 記錄的座標位置，在三維影像與二維剖面影像顯示像素位 置，讓醫師可以很容易的依據像素位置來判斷超音波影像 19 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公爱） --------------------訂--------- (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 514515 8 5 78twf. doc/0 1 2 A7 137 五、發明說明d) 中的病灶位置。 在手術導引步驟中’系統除了讓醫師可依據像素位置 的顯示來判斷病灶的位置外，還能利用裝在手術器械 上的定位裝置910隨時量測手術器械908的方位，再將此 ’方位經由裝置座標的轉換矩陣轉換成影像座標，然後同時 顯示在三維影像與二維剖面影俸中，藉此方式讓醫師了解 手術器械908的方位.，並依據上一步驟所顯示的像素位 置，來讓醫師知道手術器械908與像素位置的相對關係， 達到手術導引的目的。 ' 其中在進行影像導引之前’需先進行影像方位校準之 動作，請合倂參考第Π及第12圖，其中圖1.1是繪示出 進行影像方位校準之所需元件，而圖12是繪示進行影像 方位校準步驟之一流程圖。 圖11中之樣板906之特徵點P係作爲方位校準之媒 介，在方位校準之流程步驟中5首先需在步驟1200中， 固定定位裝置902於超音波換能器004上，定位裝置902 必需固定在超音波換能器904的剛體部份，因爲超音波換 能，器904與影像座標u〇的相對關係是固定的，所以當定位 裝置902與超音波換能器904的相對位置不變，則淀位裝 置902的裝置座標與影像座標uO的轉換關#不變，而在進 行上述之手術導引時時即不必再經過方位校準的動作。 接著在歩驟1202中擷取樣板影像及擷取定位裝置902 之座標値Μ，其中歩驟1202之擷取樣板影i象即爲圖1〇之 讀取樣板影像步驟，而爲了以基因法則運算式找出轉換關 係，必需要有三組以上的特徵點座標，每組分別有特徵點 ' 20 ----------------------訂----I----^_wi. (請先閱讀背面之注意事項再填寫本頁} 本紙張尺度適用中國國家標準（CNS)A4規格（21〇 χ 297公釐） 514515 85 7 8twf.doc/012 A7 B7 五、發明說明（θ) 的影像座標UP及裝置座標DP，因此用超音波掃描特徵點p 並擷取影像座標UP。在超音波掃描與量測裝置座標Dp時， 特徵點P相對於訊號收集器900座標値不能被移動，在擷 取超音波影像的同時，由定位裝置902得到超音波換能器 904的裝置座標，此座標在之後必需用來當作影像座標與 裝置座標轉換時的依據。 在步驟1204之量測特徵點P的影像座標$中，係以 人工判斷步驟1202所擷取的樣板影像中特徵點P的位 置，在影像中特徵點P的位置爲一個像素（pixel)點座 標（x，y)，將此像素點座標乘上超音波影像與實驗尺寸的 比例，得到新的X、Y値，再設Z値爲一任意數，則可將 點座標轉換爲影像座標UP。 在步驟1206中量測特徵點P的裝置座標Dp中，係以訊 號收集器900所用座標系量測特徵點P的裝置座標DP。由 上述所得知之定位裝置座標値ST、影像座標up及裝置座 標DP，在步驟1208中，利用基因法則運算式找出轉換矩 陣ίΤ，再將此轉換矩陣:ΓΓ存檔（步驟121〇)，其中此轉換 矩陣;ΓΓ可供本發明在進行手術導引步驟時做座標轉換之 用，來導引醫師到達病灶的實際位置。 其中量測上述定位裝置座標値ST、特徵點Ρ之影像座 標UP及裝置座標DP之方式，係如圖11所示，影像座標系 U是指固定在超音波影像上之座標系，裝置座標系d是指 訊號收集器900之座標系，超音波換能器904之座標系S 是表示固定在超音波換能器904上之定位裝置902座標 系。假設在空間中有一個固定之目標fc (如特徵點P)，此 21 —-------------—訂--------1 .^®w. (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） 514515 8578twf.doc/012 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明說明（>° )XX edge point in the X direction, the same 9 when y y represents U, y) position is the edge point of the tumor area in the y direction. In the above method of obtaining the boundary data of the target image, 'the core and edge points of the tumor are digitized for subsequent processing', and the data can be used to know the location of the tumor except for the location of the tumor.5 The actual range size (as shown in Figures 6e to 60. Although the tumor location and the actual range size can be obtained from the above, it is convenient for doctors to know that when making an excision, this paper size is applicable to the Chinese National Standard (CNS) A4 size (210 X 297 mm) ) --- 丨 丨 丨 丨 丨 丨-丨 · 111 order-! I 丨 丨 (Please read the precautions on the back before filling this page) 514515 85 78twf.doc / 012 A7 B7 V. Description of the invention (ίΓ) Accurate information, but the judgement of the benign and malignant tumors is still subjectively determined by the physician. (Please read the precautions on the back before filling out this page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs To avoid misjudgment by physicians, the present invention proposes another In the method for obtaining the external type of the target image, after the coordinate position of the boundary point of the tumor image is marked by the above method (step 700), the middle point of the tumor image is used to reach each boundary. Distance distribution (step 702), calculate the distance average 値 meansl and distance standard deviation varl (step 704), and obtain the ratio 距离 & amp between the distance average 値 1 ^ &! ^ 1 and the distance standard deviation ^ 1 * 1 ; = ^^ 1 * 1/1 ^ & 1 ^ 1 (step 706), when in step 708, the ratio 値 a is smaller than the preset 値, assuming 0.21, indicating that the shape outside the tumor image tends to be round or Rugby-like (step 710), that is, the tumor is a benign tumor. When in step 708, the ratio 値 a is greater than the preset 値 = 0.21, then in step 712, the center point and the long and short axis of the tumor image are first searched to target The image is divided into four quadrants according to the center point and the long and short axis (in steps 7, 14), and the distance average 値 (in step 716) of each quadrant is calculated, which is means2, means3, means4, means5, and distance standard deviation, respectively. Var2, var3, var4, var5, and then in step 718, respectively obtain the ratio of four quadrants 値, b = var2 / means2 ^ c = var3 / means3 ^ d = var4 / means4 'e = var5 / means5, when When the ratio of cb, c, d, and e is judged to be consistent in step 720, it means that the tumor has a symmetrical shape and a symmetrical shape. (Step 722), like a leaf shape, is a benign tumor ^ Otherwise, it indicates an irregular shape outside the tumor (step 724), as shown in FIG. 6f, it is a malignant tumor. / The present invention also proposes another method for acquiring a target image The method of the boundary category is also the basis for judging the benign and malignant tumors. In the method of obtaining the boundary data according to the above, the coordinate position of the boundary point of the tumor image is marked (step 800).値 (Step 802), the so-called 17 paper sizes are applicable to the Chinese National Standard (CNS) A4 specification (21 × 297 mm) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 514515 85 78twf.doc / 012 A7 〆B7 V. Description of the Invention The second derivative distribution (tb) is the sum of the slopes of two adjacent points and then divided by the number of all boundary points, which is the second derivative change. When the second derivative distribution is determined in step 804値 is smaller than the preset 値, assuming 21, which indicates that the boundary of the target image is a smooth curve (step 806), that is, the tumor is a benign tumor, otherwise the second derivative distribution 値 is greater than the preset 値The boundary line of the target image to an irregular curve (step 808), indicating tumor lines malignancies. Among them, in the above-mentioned external classification and edge classification methods, the preset judgments of tumor benign and evil behaviors are all clinical data. In summary, the advantages of the method of obtaining the boundary information and shape and boundary type of the target image include: 1. Remove noise. 2. Determine the location of the target image. 3 β separates the target image from the background. 4. Obtain data of the boundary of the target image. 5. Identify the shape and boundary of the target image. For example, if the target image is a tumor block, you can judge the benign and malignant tumors by identifying the type of the outer shape and boundary. Fig. 9 and Fig. 10 show the method of image guidance provided by another preferred embodiment of the present invention, which is composed of the signal collector 900 and the positioning device 902 of Fig. 9 And 910, the ultrasonic transducer 904 (such as the ultrasonic probe used for general ultrasonic scanning) and the image processing device 911 provided above, where the positioning device 91 is equipped with square or general surgical instruments 908, such as On the scalpel, surgical forceps, etc., the positioning unit 902 is placed on the ultrasonic transducer 904 to apply these 18 paper sizes to the Chinese National Standard (CNS) A4 specification (21〇X 297 public love) '------- (Please read the notes on the back before filling out this page) --------- Order --------- Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 514515 8578twf.doc / 012 A7 • 137 V. Description of the invention (Q) The direction is transmitted to the image processing device 911 (that is, a computer equipped with the present invention for image processing and image guidance program) via the signal collector 900. When the image is processed, Device 911 integrates image and orientation data and displays a guide screen to help physicians Solution surgical instruments 908 and the azimuth between lesions. In the flowchart of FIG. 10, first read the template image in step 1000 'after performing the orientation calibration on the template image in step 1002, a surgical guidance step (step 1004) can be performed, wherein the orientation calibration step in step 1002 is performed by genes The algorithm (Gene tica 1 go rit hm) finds the transformation matrix between the image coordinates and the device coordinates, and the orientation calibration will be described in more detail later. In the next series of surgical guidance steps, first scan the lesion image in step 106, that is, the physician will search for the lesion with ultrasound scans, and acquire the ultrasound image after finding the lesion, while capturing the image The system will automatically record the coordinates 902 of the positioning device on the ultrasound probe (ie, the ultrasound transducer 904), so that any pixel coordinate on the image can be converted to the device coordinates. In the step of selecting pixels in step 1008, ask the physician to select the pixels from the ultrasound image that are to be displayed on the guide screen of step 1010. The system will record the selected pixels one by one and follow the steps when the image is captured. The transformation matrix obtained in the ultrasonic transducer 904 position (ie, the coordinates of the positioning device 902) and the azimuth calibration step is used to convert the pixel coordinates into the device coordinates, and finally the coordinates are recorded. Among them, the image and the device coordinates of the ultrasonic transducer 904 can be deleted after the click operation is completed. In the step of displaying the pixel position in step 1012, the pixel position is displayed in the 3D image and the 2D cross-sectional image according to the coordinate position recorded in the previous step, so that the physician can easily judge the ultrasonic image based on the pixel position. 19 papers Standards apply to China National Standard (CNS) A4 specifications (210 X 297 public love) -------------------- Order --------- (please first Read the notes on the back and fill out this page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 514515 8 5 78twf. Doc / 0 1 2 A7 137 5. The location of the lesion in the description of the invention d). In the surgical guidance step, in addition to allowing the physician to determine the location of the lesion based on the display of the pixel position, the system can also use the positioning device 910 mounted on the surgical instrument to measure the position of the surgical instrument 908 at any time, and then use this position It is converted into image coordinates by the transformation matrix of the device coordinates, and then displayed in the three-dimensional image and the two-dimensional cross-section image at the same time. This way, the doctor understands the position of the surgical instrument 908. According to the pixel position displayed in the previous step, Let the physician know the relative relationship between the surgical instrument 908 and the pixel position to achieve the purpose of surgical guidance. 'Before performing image guidance', you need to perform image orientation calibration. Please refer to Figures Π and 12 for reference. Figure 1.1 shows the components required for image orientation calibration, and Figure 12 is for drawing. Shows a flowchart of one of the steps for image orientation calibration. The characteristic point P of the template 906 in FIG. 11 is used as a medium for azimuth calibration. In the process of azimuth calibration, step 5 is first required in step 1200. The positioning device 902 is fixed on the ultrasonic transducer 004, and the positioning device 902 must be fixed. In the rigid body portion of the ultrasonic transducer 904, the relative relationship between the transducer 904 and the image coordinate u0 is fixed because of the ultrasonic transducer. Therefore, when the relative position of the positioning device 902 and the ultrasonic transducer 904 remains unchanged, Then, the conversion relationship between the device coordinate of the lake position device 902 and the image coordinate uO is unchanged, and it is not necessary to perform the azimuth calibration operation when performing the above-mentioned surgical guidance. Then, in step 1202, the image of the sampling plate and the coordinates of the positioning device 902 are acquired. The image of the sampling plate in step 1202 is the step of reading the template image in FIG. Formula to find the conversion relationship, there must be more than three groups of feature point coordinates, each group has a feature point '20 ---------------------- order --- -I ---- ^ _ wi. (Please read the precautions on the back before filling out this page} This paper size applies to China National Standard (CNS) A4 (21〇χ 297 mm) 514515 85 7 8twf.doc / 012 A7 B7 5. In the description of the invention (θ), the image coordinates UP and device coordinates DP, so the ultrasonic scans the feature point p and captures the image coordinates UP. When the ultrasound scans and measures the device coordinate Dp, the feature point P is relative to The coordinates 900 of the signal collector 値 cannot be moved. While acquiring the ultrasound image, the device coordinates of the ultrasonic transducer 904 are obtained by the positioning device 902. This coordinate must be used later as the image coordinate and device coordinate conversion. In the measurement of the image coordinates $ of the feature point P in step 1204, it is determined manually in step 1202. The position of the feature point P in the captured sample image. The position of the feature point P in the image is a pixel (pixel) point coordinate (x, y). Multiply this pixel point coordinate by the ratio of the ultrasonic image to the experimental size. To get the new X and Y 値, and set Z 値 to an arbitrary number, then the point coordinates can be converted into the image coordinates UP. In step 1206, the device coordinates Dp of the characteristic point P measured by the signal collector 900 The coordinate system used is a device coordinate DP for measuring the characteristic point P. From the above-mentioned positioning device coordinates 値 ST, image coordinates up, and device coordinates DP, in step 1208, a transformation algorithm Γ is found by using a genetic algorithm expression, and then this Transformation matrix: ΓΓ archive (step 121), where the transformation matrix; ΓΓ can be used by the present invention to perform coordinate transformation during the surgical guidance step to guide the physician to the actual position of the lesion. The positioning device is measured The method of the coordinate Y, the image coordinate UP of the characteristic point P, and the device coordinate DP are shown in FIG. 11. The image coordinate system U refers to the coordinate system fixed on the ultrasonic image, and the device coordinate system d refers to the signal collector. 9 The coordinate system of 00 and the coordinate system S of the ultrasonic transducer 904 indicate the coordinate system of the positioning device 902 fixed on the ultrasonic transducer 904. Assume that there is a fixed target fc in the space (such as the characteristic point P), This 21 —-------------— Order -------- 1. ^ ®w. (Please read the notes on the back before filling out this page) Intellectual Property Bureau, Ministry of Economic Affairs Printed by employee consumer cooperatives. This paper is sized according to Chinese National Standard (CNS) A4 (210 X 297 mm) 514515 8578twf.doc / 012 A7 B7 Printed by employee consumer cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. )
點在座標系U及座標系D中的座標爲UP及Dp，自胃u巾 .的關係可知： D DP」muP.......(1) 其中°p、可以由訊號收集器900，定位裝置9〇2 及影像中得知，而ίΤ則爲一未知的4*4矩陣，；^中^含6 個未知數’即S座標系到U座標系的原點値 (X。f f s e t，Y。f f s e t，Z。f f S e t )與角度旋轉（α，冷，γ )。由^ ^ 可推知： ^ …-…（2) 、 其中ST爲ΓΓ的反矩陣，而要求出π，必須要有足夠數目的 校準用特徵點，並將每一特徵點對應之Dp、Up、人& 式（2)中，以求出中的未知數，而本發明係利用基因法 則運算式求得^的最佳解，所採用之合適函數（Fltness Function)定義如下： D = |：[di<P，》]2—…―⑶ 、 i=l 其中，dist()代表兩點間的距離 ，P:SDT，The coordinates of points in the coordinate system U and coordinate system D are UP and Dp, since the relationship between the stomach and the towel is known as: D DP ″ muP ....... (1) where ° p, can be determined by the signal collector 900 It is known from the positioning device 902 and the image, and Τ is an unknown 4 * 4 matrix; ^ in ^ contains 6 unknowns', that is, the origin of the S coordinate system to the U coordinate system (X.ffset, Y.ffset, Z.ffs et) and angular rotation (α, cold, γ). It can be inferred from ^ ^: ^…-… (2), where ST is the inverse matrix of ΓΓ, and π is required to have a sufficient number of calibration feature points, and each feature point corresponds to Dp, Up, In human & formula (2), to find the unknown number, and the present invention uses the genetic algorithm to find the best solution of ^. The suitable function (Fltness Function) used is defined as follows: D = |: [ di < P,》] 2—… ―⑶, i = l where dist () represents the distance between two points, P: SDT,
Sp^^Tu.P 綜合上述，此以超音波即時顯像的特性做快速的位置 導引之優點如下： Μ 吏用步驟簡單：本發明將方位校準與手術導引流程 分開，讓使用時的操作步驟簡單快速。 ‘ 2.資料可隨時更新：在手術過程中，醫師通常需要知 道病灶的位移情形，本發明符合此一要求，可隨時取得超 22 4^·氏張尺度適用中國國家標準(CNS)A4規格（210 X 297公釐） " 一 -----------<^11 裝 ί——-—丨訂--------- (請先閱讀背面之注意事項再填寫本頁) 514515 85 78twf.doc/012 A7 _ _B7____ 五、發明說明（Μ ) 音波影像並顯示像素位置做爲導引記號。 3。消除誤差：本發明所提供之方位校準單元以基因 法則運算式從多數解中找出全域最佳解（轉換矩陣），可 以消除許多方位校準時影像座標和裝置座標的量測誤 差。 雖然本發明已以較佳實施例揭露如上，然其並非用以 限定本發明，任何熟習此技藝者’在不脫離本發明之精神 和範圍內，當可作各種之更動與潤飾，因此本發明之保護 範圍當視後附之申請專利範圍所界定者爲準。 -----------·11111--訂--------- (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐）Sp ^^ Tu.P Based on the above, the advantages of fast position guidance based on the characteristics of real-time ultrasound imaging are as follows: Μ Easy to use steps: The present invention separates the orientation calibration from the surgical guidance process, allowing the The steps are simple and fast. '2. The data can be updated at any time: During the surgery, the physician usually needs to know the displacement of the lesion. The present invention meets this requirement and can be obtained at any time in excess of 22 4 ^ ·' s scales applicable to Chinese National Standard (CNS) A4 specifications ( 210 X 297 mm) " One ----------- < ^ 11 Packing ————- 丨 Order --------- (Please read the precautions on the back before (Fill in this page) 514515 85 78twf.doc / 012 A7 _ _B7____ 5. Description of the invention (M) Sound wave image and display pixel position as a guide mark. 3. Elimination of errors: The azimuth calibration unit provided by the present invention finds the global best solution (transformation matrix) from the majority of solutions by using the algorithm of genetic algorithm, which can eliminate many measurement errors of image coordinates and device coordinates during azimuth calibration. Although the present invention has been disclosed as above with preferred embodiments, it is not intended to limit the present invention. Anyone skilled in the art can make various modifications and retouches without departing from the spirit and scope of the present invention. The scope of protection shall be determined by the scope of the attached patent application. ----------- · 11111--Order --------- (Please read the notes on the back before filling out this page) Printed on paper by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs Standards apply to China National Standard (CNS) A4 (210 X 297 mm)