Detection of dermis and fascia on skin layers for liposuction surgery robot using texture and geometric information

Despite the fact that liposuction is one of the most common types of cosmetic surgery, it causes skin surface irregularity as a side effect because of the procedure's lack of systemicity and objectivity in measuring the process of regional suction. To determine a more systematic and quantitative liposuction process, the surgeon requires access to a surgical robotic system for liposuction. The first consideration in such a system, is navigation of subcutaneous fat, especially detection of the dermis and fascia skin layers. Therefore, this paper presents a method for detecting the dermis and fascia in the skin structure using an ultrasound image that could assist the surgeon's procedure during liposuction. The method proposed in this paper includes the following three steps. 1) Using the Gabor filter bank, extract the texture feature from the ultrasound image. 2) Using a k-means clustering algorithm, extract cluster areas from the texture feature, such that cluster areas contain similar texture features. 3) Detect the dermis and fascia from each cluster's geometric information as the feature after training a multi-class SVM. Using the proposed algorithm, the performance results for precision and recall for dermis are 96% and 100%, respectively. In the case of fascia, the precision is 71.11% and recall is 86.49%. The proposed algorithm would be useful as the navigation system in the development of a surgical robot for liposuction in the near future.