SSQ05-06

Denoised Ultra Low Dose for Screening Lung Cancer

Thursday, Nov. 29 11:20AM - 11:30AM Room: E353B



Participants
Edith M. Marom, MD, Tel Aviv, Israel (Presenter) Speaker, Bristol-Myers Squibb Company; Speaker, Boehringer Ingelheim GmbH;
Michael Green, MSc, Tel Aviv, Israel (Abstract Co-Author) Nothing to Disclose
Michal Eifer, MD, Ramat Gan, Israel (Abstract Co-Author) Nothing to Disclose
Eli Konen, MD, Ramat Gan, Israel (Abstract Co-Author) Nothing to Disclose
Nahum Kiryati, PhD, Tel Aviv, Israel (Abstract Co-Author) Nothing to Disclose
Arnaldo Mayer, PhD, Ramat Gan, Israel (Abstract Co-Author) Nothing to Disclose

For information about this presentation, contact:

edith.marom@gmail.com

PURPOSE

To assess the effect of a denoising method (D) for ultra low dose CT (ULDCT) LungRADS categorization.

METHOD AND MATERIALS

36 consented patients, referred for an outpatient chest CT, underwent 2 scans: a normal dose CT (NDCT), 120 kVp and automatic current modulation, with or without contrast media, immediately followed by an ULDCT, 120 kvp and fixed current at 10 mA for bmi <29 and 20 mA for bmi>=29. Reconstruction for lung and soft tissue kernels were performed for each scan. Consecutively, each ULDCT was denoised using a locally-consistent non-local-mean (LCNLM) algorithm to obtain a high signal to noise ratio (SNR) version of the ULDCT. The LCNLM algorithm leverages large databases of image patches extracted from high-SNR chest CT scans to denoise ULDCTs while enforcing local spatial consistency to preserve fine details and structures in the image. Blinded to all clinical information, a chest radiologist separately assessed the NDCT, ULDCT, and denoised ULDCT (D), documented findings, assigned a LungRADS category and a subjective suspicion for highly suspicious lesions for lung cancer (H).

RESULTS

Radiation dose using NDCT reduced the radiation for patients with a BMI > 29 by an average of 93% and for those with a BMI of up to 29 by an average of 96% . For patients with a BMI > 29 the average effective radiation dose for ULDCT was 0.41 mSv, whereas for those with a BMI of up to 29 it was 0.24mSv. For the three imaging methods, the same score was seen in 63.9% (n=23) and a different score in 36.1% (n=13). There was complete agreement on LungRADS 4A (or higher) between NDCT and D, but ULDCT categorized one of the 4A patients as LungRads 2. One lesion assigned as LungRads 4X by ULDCT was assigned LungRads2 by D and NDCT. Of the 8 patients highly suspicious for lung cancer by NDCT, D indicated so in all 8 whereas ULDCT indicated so only in 4.

CONCLUSION

Interpretation of ULDCT may cause errors in LungRADS categorization but implementation of the LCNLM algorithm for denoising improves ULDCT images so that LungRADS categorization is similar to normal dose scans.

CLINICAL RELEVANCE/APPLICATION

Denoising ULDCT with the LCNLM algorithm enables screening for lung cancer with dose reductions of greater than 90%.

Honored Educators

Presenters or authors on this event have been recognized as RSNA Honored Educators for participating in multiple qualifying educational activities. Honored Educators are invested in furthering the profession of radiology by delivering high-quality educational content in their field of study. Learn how you can become an honored educator by visiting the website at: https://www.rsna.org/Honored-Educator-Award/ Edith M. Marom, MD - 2015 Honored EducatorEdith M. Marom, MD - 2018 Honored Educator