Radiology is the specialty that learned an image can save a life.

In April 1985, The Lancet published a trial about screening women for breast cancer. The screening test was an X-ray of the breast, called a mammogram. The trial asked whether regular mammograms could stop women dying of the disease. It began in 1977 in two Swedish counties. The trial divided the women at random into two groups. One group was offered a mammogram every two or three years. The other group was not offered screening. The report covered 134,867 women aged 40 to 74. Screening found more cancers at an early stage, while they were small. The group offered screening had 31% fewer deaths from breast cancer. It also had 25% fewer cancers found at an advanced stage. An X-ray could find breast cancer early enough to save lives.

Seventeen years later, in October 2002, the same journal published a trial about burst blood vessels in the brain. A ruptured aneurysm is a bulge in a brain artery that bursts and bleeds. Surgeons treated it by opening the skull and closing the bulge with a clip. A newer method reached the aneurysm from inside the blood vessels. A thin tube carried soft platinum coils up to the bulge and filled it. Nobody knew which method was better. The trial gave 2,143 patients either clipping or coiling. It was stopped early, at a planned check. Death or dependence on others at one year affected 23.7% of the patients given coils. It affected 30.6% of the patients given surgery. Coiled aneurysms bled again a little more often later, but the risk was low with both methods. A treatment through the blood vessels had beaten open brain surgery.

Fifteen and a half years later, in May 2018, the New England Journal of Medicine published a trial about finding prostate cancer. Men with a raised PSA blood test usually had a needle biopsy, with 10 to 12 samples taken across the prostate. The trial asked whether an MRI scan first could do better. It took 500 men who had never had a biopsy. One group had the standard biopsy. The other group had an MRI first. Men with a suspicious scan had a biopsy aimed at the suspicious area. Men with a normal-looking scan had no biopsy at all. The MRI spared 28% of the men in that group a biopsy. Serious cancer was found in 38% of the men in the MRI group. It was found in 26% of the men given the standard biopsy. The MRI group also had fewer low-risk cancers diagnosed. A scan before the needle had found more of the cancers that matter, with fewer biopsies.

These are three of the thousands of studies that shaped modern radiology. They taught radiology that a mammogram could find breast cancer early enough to save lives, that coils placed through the blood vessels could treat a burst brain aneurysm better than open surgery, and that an MRI before biopsy could find the prostate cancers that matter, and each time the answer was in a paper years before it was in a textbook.

It is still happening. Recently, in January 2026, The Lancet published a trial about artificial intelligence in breast screening. Standard screening in Sweden has two radiologists read every mammogram. The trial asked whether AI could help without missing more cancers. It divided 105,934 women at random into two groups. One group had the usual reading by two radiologists. The other group had their scans sorted by AI. The AI sent some scans to one radiologist and the rest to two, and it pointed out suspicious areas. The researchers counted the cancers that screening missed and that appeared before the next screen. These missed cancers affected 1.55 in every 1,000 women screened with AI. They affected 1.76 in every 1,000 women screened the usual way. Screening with AI caught 80.5% of the cancers. The usual screening caught 73.8%. Both groups had the same rate of false alarms. The AI also cut the radiologists’ reading work. A computer had helped radiologists find more cancers with less work. Practice follows the literature as it goes. It is a good culture. It has exactly one cost.