| Article ID: | iaor19992426 |
| Country: | United Kingdom |
| Volume: | 49 |
| Issue: | 12 |
| Start Page Number: | 1233 |
| End Page Number: | 1238 |
| Publication Date: | Dec 1998 |
| Journal: | Journal of the Operational Research Society |
| Authors: | Laporte Gilbert, Semet Frdric, Dadeshidze V.V., Olsson L.J. |
| Keywords: | heuristics, programming: travelling salesman |
In the screening of cytological samples, a slide is examined using a microscope. Since the area of the specimen is very large compared with the area of a microscope field of view (FOV), a large number of FOVs is needed to cover the entire specimen area. A heuristic reducing the number of FOVs has been developed to aid in screening these specimens on the automated microscopy workstation. A smear slide is prescreened on a computer-aided optimal microscope system and locations of diagnostically significant material are stored. The heuristic computes a set of FOVs that covers this material and a route between them. As the problem is of large scale and must be solved repeatedly in real time any heuristic must be very fast. Our heuristic solves a set partitioning problem, followed by a travelling salesman problem. It produces a good solution within a very short computing time.