Iterative Reconstruction Equal to Filtered Back Projection

Laird Harrison

May 09, 2014

Radiologists can interpret CT images made with iterative reconstruction as accurately as images made with filtered back projection, a new study shows.

"Iterative reconstruction improves image quality by decreasing image noise," said investigator Scott Steenburg, MD, assistant professor of radiology at Indiana University in Indianapolis. "It can also be used to decrease radiation exposure while maintaining image quality that's comparable to a full-dose filtered back projection," he told Medscape Medical News.

Many radiologists have expressed concern about iterative reconstruction, a set of algorithms used to reduce noise or reconstruct images with incomplete data, said Dr. Steenburg.

"When this package first came out in my institution in 2012, the images looked a little blotchy or smeared," he explained. "There was a lot of hand-wringing in our department that we would miss pathology because the images were not as crisp as we were used to."

He and his colleagues decided to test whether the algorithms affected the way radiologists interpret images of blunt solid organ injuries.

"In the trauma world, we may get only one chance to do the CT scan. The last thing we in radiology want to do is take up too much time creating diagnostic images," he said.

Dr. Steenburg presented the study results at the American Roentgen Ray Society 2014 Annual Meeting in San Diego.

The investigators evaluated 100 patients with various grades of solid organ injuries identified on CT scan with filtered back projection and 52 patients without organ injuries.

All scans were obtained using intravenous contrast on a 64-slice multidetector Philips CT scanner with reference 300 mAs, 120 kV peak, and a fixed 75 second delay.

They reconstructed raw CT data using filtered back projection and 3 levels of iterative reconstruction: Philips iDose levels i2, i4, and i6.

Images of a 19-year-old female with grade 4 liver injury after a motor vehicle collision. Courtesy, Dr. Scott Steenburg.

Four emergency radiologists with 2 to 6 years of experience, blinded to the reconstruction parameters and the presence or absence of solid organ injury, independently reviewed each case.

They rated image quality on a 5-point scale and recorded the presence or absence of solid organ injuries.

If the radiologists thought injuries were present, they graded the injuries on the American Association of Surgery in Trauma Solid Organ Injury Scale. They also stated whether they thought iterative reconstruction was used and, if so, at which level.

They documented 61 solid organ injuries: 20 to the liver, 33 to the spleen, and 8 to the kidney.

There was no significant difference between filtered back projection and the various iterative reconstruction levels for the diagnosis of solid organ injuries. In addition, iterative reconstruction had no effect on the grading of solid organ injuries (> .8 for all solid organ injuries).

The radiologists judged i2-level images to have the best overall quality. The difference in quality rating between i2 and other images was significant (< .01).

The sensitivity for detecting iterative reconstruction was 80%. The radiologists couldn't tell the difference between i2 and i4 levels (= .8), but they could tell the difference between i4 and i6 levels (< .001).

The results are not surprising, said Elliot Fishman, MD, professor of radiology, surgery, and oncology at Johns Hopkins Hospital in Baltimore. However, findings might be different on other brands of scanner, he cautioned.

"I think this is one more article saying iterative reconstruction has possibilities," he told Medscape Medical News. "But it needs to be customized by the machine and it has to be customized by the application."

He pointed out that the study did not show a reduction in radiation dose.

"The biggest issue with CT is dose," he said. "If you can decrease dose, it's a good thing. If you can do iterative reconstruction and improve the quality, that would be a great thing, but I have not seen anything to indicate that."

Dr. Steenburg and Dr. Fishman have disclosed no relevant financial relationships.

American Roentgen Ray Society (ARRS) 2014 Annual Meeting: Abstract 168. Presented May 9, 2014.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.