Cyber-physical tactile imaging system for malignant tumor identification

Abstract

In this paper, we introduce a Cyber-Physical Tactile Imaging System (TIS) for identifying malignant tumors. The sensing is based on the principle of total internal reflection (TIR) of light. CP-TIS will estimate mechanical properties such as size and elasticity of the embedded objects (e.g. tumor). Based on the mechanical properties determined by TIS, it is possible to generate a malignancy score, since studies have shown that malignant tumors tend to be larger and stiffer. Clinical Breast Examination (CBE) is one of the qualitative techniques used by doctors for early detection of breast tumors. However, this method is effective only in detecting lesions, but not in classifying tumor malignancy. Also this is a subjective method where the performance depends on how the doctors perform CBE. Therefore, it is advantageous to quantify the mechanical properties of a tumor for objective tumor identification.Fig. 1 shows a schematic of CP-TIS. The system consists of physical and cyber components. In the physical component, there are 1, 2,..., N number of TIS with operators, laptops, and targets. The TIS includes an optical sensing probe made of Polydimethylsiloxane, a light source unit made of four light emitting diodes, a camera unit, a computer unit, a force gauge unit, as shown in Fig 2. Due to the TIR, the system does not detect anything until the contours of the sensing probe changes. However, light scatters and our TIS will capture this image when the pressure is applied to the probe. The operators use TIS on the target to capture tactile images. The images are transferred to cyber component of the system for processing and analyzing through a communication network such as Wi-Fi. The images are processed and mechanical properties are estimated in the computation unit. This information is stored in a database with patient information. Then we compute malignancy score using Machine Learning Algorithm such as Support Vector Machine. The score helps the doctors and patients to decide whether to seek further medical help.In this demonstration, a prototype TIS will be shown. A direct contact experiment was performed on various phantoms to capture tactile images. The images were then processed and analyzed to estimate size and elasticity. The phantoms were made of silicone, silicone rubber, and Polyvinyl Chloride-Plastisol (PVC-Plastisol). In the laboratory environment, the relative errors for size estimation were found to be at least 3.57% for the silicone phantom, 1.82% for the PVC-Plastisol phantom, and 0.90% for the silicone rubber. The relative errors for elasticity estimation were found to be 4.83% for the PVC-Plastisol phantom, and 5.50% for the silicone rubber. A pilot clinical study was conducted on twenty breast cancer patients in Temple University Hospital following the IRB protocol no. 13661. The estimated elasticity was correlated with the biopsy results. Preliminary results show that a sensitivity of 66.70% and a specificity of 91.70% for elasticity.The authors would like to thank Amrita Sahu, Yi Chen, and Dina Caroline, MD for their contributions and support in this work. This work was supported in part by Pennsylvania Department of Health.