Tactile Imaging

Palpation is a ubiquitous skill that we use every day, from finding a button-hole in a shirt to determining if fruit is ripe in the grocery store. Nowhere is it more important than when it is used by physicians in the discovery and assessment of disease. The problem with this technique is that it can be difficult to accurately verbalize and record tactile sensations, which makes the comparison of two manual examinations extremely difficult. This thesis presents a new medical imaging technique, tactile imaging, which promises to alleviate this difficulty through more objective documentation of clinical examinations. We present one implementation of such a system, targeted at the documentation of the properties of palpable breast lumps. It is comprised of a hand held scan head with a distributed pressure sensor mounted on it and a magnetic position tracker contained within it that is stroked across the surface of the breast by a clinician. The pressure distribution on the face of the sensor and its location in space are recorded using a computer and an algorithm assembles a composite tactile map of the structure in real time. This map is the average of all pressures recorded at a particular location in the breast as the physician examines it. In order to extract features from these maps, we have developed mathematical models that relate material and geometric properties of palpable lumps to surface pressure distributions. These models were used to develop inversion algorithms that can be used to extract the size of the lumps imaged. In clinical tests on 25 surgical patients, size estimate accuracy (compared to ex vivo size measurements) was within 13% mean absolute error (MAE). This compares favorably to 34% and 47%, MAE, respectively, for ultrasound estimates and clinical breast examination estimates. Across multiple maps, size repeatability was within 7.5% (one standard deviation). Tactile imaging therefore promises to improve the objectivity of clinical breast examination.