A new way of identifying lung cancer that also may prevent thousands of unnecessary operations yearly was announced yesterday by doctors at Johns Hopkins University School of Medicine.

The new test uses a computer to analyze special cross-sectional X-rays and is not a substitute for the routine chest X-ray.

But in about 40,000 patients each year -- those whose X-rays show nodules that might or might not be lung cancer -- the test may prevent risky chest surgery, according to Dr. Stanley S. Siegelman, who directed the research team.

Lung cancer in the most common kind of fatal cancer in the United States. About 117,000 new cases of the disease are expected this year. The disease kills 92 percent of its victims within two years, according to Dr. Melvin S. Tockman, director of the Johns Hopkins lung project.

But if a lung tumor is discovered on a chest X-ray while it is still less than an inch in diameter and if it is removed before it can spread or produce symptoms the victims has a far better chance.

"This [the new test] is an important tool for the planning of cancer treatment," said Dr. Martin Donner, chairman of the radiology department at Johns Hopkins. "We consider [it] a breakthrough."

The new test helps doctors decide whether a possible tumor -- seen as a small, round shadow on a patient's chest X-ray -- is likely to be malignant or benign. It is based on identifying the presence of calcium in the nodule, even though there may not be enough calcium to show upon the X-ray. Studies of hundreds of patients have proved that if a lung nodule contains calcium, it is virtually always benign.

The test is used at Johns Hopkins when a patient has a small lung module that radiologists think may be cancerous. Instead of immediately having a biopsy or operation, the patient has a CT (computed tomography) scan of the lungs -- a special series of X-rays in which a computer produces detailed images of thin "slices" through the patient's chest.

The nodule will be visible on some slices, but radiologists still cannot tell if it is malignant. So the computer analyzes the nodule in each picture, and produces a series of numbers measuring its density at different points.

Measuring the densities of 91 lung nodules, the Hopkins researchers found that when calcium was present, nodules were much denser than when it was absent. The difference in density between cancers and benign scars was so consistent that they were able to identify the benign nodules every time. Most benign nodules are either scars or reactions to infections that have healed.

If the density was low, the researchers found, the nodule might still be benign -- but the chances that it was cancer were 80 percent.

Siegelman said the doctors have studies for more than 1 1/2 years more than 30 patients with nodules the computer identified as benign. The nodules have not grown or changed. Many of these patients otherwise would have had open lung surgery -- an operation with a 5 percent mortality -- or would have undergone a risky lung needle biopsy in order to learn that they did not have cancer, Siegelman said. He estimated that if the test were available nationwide, it could prevent 40,000 operations and biopsies a year.

"The majority of patients who have this operation turn out to have benign disease," Siegelman said.

"This work has two forks to it. One is to do nothing for the benign ones. But two is to get aggressive with the ones we deem malignant" instead of watching a nodule on repeated chest X-rays to see if it grows.

Patients whose lung cancer is found when the tumor is a nodule less than about an inch across can be cured of the cancer 85 to 92 percent of the time, Tockman said. For this reason, doctors order periodic chest X-rays for patients over 35, especially smokers, hoping to find tumors that are still curable.

Such X-rays uncover some 200,00 small lung nodules each year, Siegelman said. Half can be visually identified as benign, either because calcium flecks are obvious or because the same nodule can be found on a patient's previous chest X-rays.

But most of the other 100,000 patients undergo either a biopsy or an operation, Siegelman said. It is for these people that the test is potentially valuable.

The CT scan is a safe test, and at John Hopkins it costs $131. A lung biopsy costs $165, but can cause a collapsed lung that requires hospitalization. Chest surgery for removal of a nodule entails up to 10 days in the hospital, and at Johns Hopkins costs more than $4,000.

Nationally, the cost of providing the new test at many medical centers would be considerable.Siegelman said the technology of CT scanners has advanced so rapidly that only the newest machines, which produce pictures of cross-section of the body less than one-quarter inch thick, would be capable of analyzing lung nodules. He said such machines cost $600,000 to $800,-apiece, and cost a hospital $300,000 per year to run.

"If [a hospital] bought a machine five years ago, that machine is no longer applicable," he said.

He argued that at a medical center such as Johns Hopkins, which performs 10,000 CT scans a year, the scanners quickly pay for themselves and may save money by eliminating the need for other tests. But many health planners oppose the purchase of scanners by community hospitals that may not use the machines enough to justify the cost.