Detection of Metastases in Human Lungs from CT-Scans

Abstract

One of the possibilities to assist physicians with their diagnostic work based on CT-scans would be a tool for automatic detection of metastases in human lungs. The present project stems from the ‘Dr. Daniel den Hoedkliniek’ in Rotterdam. The problem was successfully tackled during the week and resulted in a highly efficient algorithm to detect metastases from CT-data. The algorithm is restricted to metastases that are not attached to each other or to the lung edges. A first attempt to implement it into a Fortran program yielded successful results on artificial data sets. For fine-tuning of parameters the algorithm should be applied to real data from the clinic.