This proposal uses the BIRT mechanism to initiate a collaboration between the PI of the parent R01, a biologist who studies bone healing, and the Co-PI, a physicist with expertise in medical imaging and computational image analysis. It addresses a major clinical problem: our inability to determine with precision when bone has healed. This has important implications both for the clinical management of trauma patients and for the implementation of randomized, controlled studies to evaluate new bone-healing methods. For the parent R01, the PI utilizes a rat, femoral, segmental defect model. This is well characterized and has generated a wealth of historical data, including radiographs, microCT and mechanical testing, concerning defects that healed, did not heal or partially healed.
In Specific Aim one, the Co-PI will use the historical microCT data and a software simulation technique to generate virtual radiographic images of the bone from different orientations. Using bone samples with known outcomes, the image acquisition geometry and software algorithms will be optimized to provide the maximum accuracy for detecting fractures, producing a metric that provides an accurate measure of bone healing probability. This will produce an algorithm that allows healing to be assessed on the basis of just two plain radiographs taken at different angles. Receiver operating characteristic (ROC) analysis will be used to quantify performance.
In Specific Aim 2, this metric will be tested empirically using data obtained from fresh rats whose defects are given different doses of BMP-2 that result in different degrees of healing, ranging from no healing to full healing. The performance of the metric will be evaluated by comparison with """"""""bridging"""""""" healing, as measured independently by microCT, and """"""""mechanical"""""""" healing, as measured by mechanical testing. If this project is successful, it will provide the trauma surgeon with a reliable, quantitative, objective, and inexpensive method for assessing bone healing. Although the algorithms involve sophisticated mathematical analyses, they run automatically and can be incorporated into digital radiography system equipment found commonly in most radiology departments. Because the technology uses existing hardware, clinical translation should be straightforward.
Surgeons are often unable to tell when a broken bone has healed. This can compromise clinical care and slow the development of better ways to heal bone. The proposed research will provide a new, reliable and inexpensive way to determine when bones have healed. Due to its accurate, inexpensive and easy to use mechanism, it should become widely available for clinical care efforts.
Showing the most recent 10 out of 28 publications
