Ahmidi Narges, Hager Gregory D, Ishii Lisa, Gallia Gary L, Ishii Masaru
Department of Computer Science, Johns Hopkins University, Baltimore, MD 21218, USA.
Med Image Comput Comput Assist Interv. 2012;15(Pt 1):471-8. doi: 10.1007/978-3-642-33415-3_58.
We observe that expert surgeons performing MIS learn to minimize their tool path length and avoid collisions with vital structures. We thus conjecture that an expert surgeon's tool paths can be predicted by minimizing an appropriate energy function. We hypothesize that this reference path will be closer to an expert with greater skill, as measured by an objective measurement instrument such as objective structured assessment of technical skill (OSATS). To test this hypothesis, we have developed a surgical path planner (SPP) for functional endoscopic sinus surgery (FESS). We measure the similarity between an automatically generated reference path and surgical motions of subjects. We also develop a complementary similarity metric by translating tool motion to a coordinate-independent coding of motion, which we call the descriptive curve coding (DCC) method. We evaluate our methods on surgical motions recorded from FESS training tasks. The results show that the SPP reference path predicts the OSATS scores with 88% accuracy. We also show that motions coded with DCC predict OSATS scores with 90% accuracy. Finally, the combination of SPP and DCC identifies surgical skill with 93% accuracy.
我们观察到，进行微创外科手术的专家外科医生学会了尽量缩短其工具路径长度，并避免与重要结构发生碰撞。因此，我们推测，可以通过最小化一个合适的能量函数来预测专家外科医生的工具路径。我们假设，这条参考路径会更接近技能更高的专家，这是通过诸如客观结构化技术技能评估（OSATS）等客观测量工具来衡量的。为了验证这一假设，我们为功能性鼻内镜鼻窦手术（FESS）开发了一种手术路径规划器（SPP）。我们测量自动生成的参考路径与受试者手术动作之间的相似度。我们还通过将工具运动转换为与坐标无关的运动编码，开发了一种互补的相似度度量方法，我们称之为描述曲线编码（DCC）方法。我们在FESS训练任务记录的手术动作上评估我们的方法。结果表明，SPP参考路径预测OSATS分数的准确率为88%。我们还表明，用DCC编码的动作预测OSATS分数的准确率为90%。最后，SPP和DCC的组合识别手术技能的准确率为93%。
