• 沒有找到結果。

我們提供一個流程,以霍夫變換偵測影像中的直線,藉由為直線配對和做 connected component labeling 為配對找出可能的器械頭尾,再以機率選出連續影 像中最有可能是器械的配對,也就是將手術器械從連續影像分離出。之後取得器 械頭部移動軌跡,再與樣版軌跡算出彼此的correlation,辨認出類似的操作器械 動作。

我們的做法,欠缺對3 維空間的考慮,雖然錄像是 2 維空間,但實際上手術 是在3 維空間進行。這可以在訓練過程中多加一台攝影機,由兩錄像算出手術器 械的深度。

我們分析的錄像中的手術器械都是黑色,深色器械不易反光或背景影響,若 是淺色器械會降低準確度。可以考慮加入顏色的追蹤,既然手術中使用的器械都 已固定,加入追蹤所有同樣顏色的區塊的條件,可以提高準確度

我們的最終目的是建構一自動化的評估系統,在完成手術器械的追蹤與動向 分析後,我們的構想是拿被評估的錄像與樣版做比較,樣版也是錄像,事先錄下 一經驗豐富的人手術操作過程,並由系統將樣版錄像中的器械的軌跡找出,之後 的評估就以樣版錄像做依據,但我們不能就單純比較兩者的軌跡的不同,因為實 際上手術過程也是由許多小過程組成,這些小過程並不是只能照一種順序去做,

只以一種順序去評估也違背了當初建造客觀的系統的期望,理想的做法是分別找 出各種小過程,再拿去與樣版錄像中同類的小過程做比較。這部分需與醫學院合 作,儘可能找出各種小過程,詳細地記錄下移動速度、軌跡。然後在實際比較受 試者錄像與樣版錄像時,差得越多的小過程,就需要標記出來提醒學員改進。另 外為了讓評估結果更人性化,也可考慮加入評語或建言,這也需要跟醫學院合 作,由他們提出手術過程中容易犯的錯誤,並為每種錯誤提供改進的建議。之後

我們就可在錄像中學員犯錯的地方提供適當的建議。

參考文獻

1. http://www.bioknow.cn/html/pages/664/Paper_15664.htm

2. A.M Derossis, G.M Fried, M Abrahamowicz, et al. Development of a Model for Training and Evaluation of Laparoscopic Skills. Am. J. Surg. (1998);

175:482-487.

3. A.M Derossis, J. Bothwell, H.H. Sigman, G.M. Fried. The effect of practice on performance in a laparoscopic simulator. Surg. (1998); 12:1117-1120.

4. Traxer O, Gettman MT, Napper CA, et al. The impact of intense laparoscopic skills training on the operative performance of urology resident. J. Urol. (2001);

166:1658-1661.

5. Sara L. Waxberg, Steven D. Schwaitzberg, Caroline G.L. Cao. Effects of videogame experience on laparoscopic skill acquisition. Proceedings of the Human factors and ergonomics society 49th annual meeting, 2005, p.1047.

6. Pamela A, Derek W, Birkmeyer J et al. Laparoscopic skills are improved with LapMentro[TM] training: results of a randomized, double-blined study. Annal of surgery (2006); 243(6):854-863.

7. Tarek A. Emam, George B. Hanna, Chris Kimber, Alfred Cuschieri. Difference between Experts and Trainees in the motion Pattern of the Dominant Upper Limb during Intracorporeal Endoscopic Knotting. Digestive Surgery (2000);

17:120-125.

8. Liang Wang, Weiming Hu, Tieniu Tan. Recent developments in human motion analysis. Pattern Recognition 36 (2003); 585-601.

9. Ahmed Elgammal, David Harwood, and Larry Davis, Non-parametric Model for Background Subtraction, D. Vernon (Ed.): ECCV 2000, LNCS 1843, pp.

751–767, 2000

10. K.P K, A. B, Moving objects recognition using an adaptive background memory, in: V. Cappellini(Ed.), Time-Varying Image Processing and Moving Object Recognition, Vol. 2, Elsevier, Amsterdam, The Netherlands, 1990

11. CR Wren, A Azarbayejani, T Darrell, A.P. Pentland, Pfinder: real-time tracking of the human body, IEEE trans. Pattern Anal. Mach. Intell. 19(7)(1997) 780-782 12. M. Isard, A. Blake, Contour tracking by stochastic propagation of conditional

density, Proceedings of the European Conference on Computer Vision, 1996, pp.

343-356

13. R. Polana, R. Nelson, Low level recognition of human motion, Proceedings of the IEEE CS Workshop on Motion of Non-Rigid and Articulated Objects, Austin, TX, 1994, pp. 77-82.

14. G. Welch, G. Bishop, An introduction to the Kalman filter, from

http://www.cs.unc.edu, UNC-Chapel Hill, TR95-041, November 2000.

15. M. Isard, A. Blake, Condensation-conditional density progaion for visual tracking, Int. J. Comput. Vision 29(1)(1998) 5-28

16. Jaeyong Chun, Kwanghyun shim, Color object tracking system for interactive entertainment application. Digital content research division, ETRI 161

Gajung-dong, Yusung gu, Daejeon, south korea zip:305-350.

相關文件