Mention lossy compression, and the knee-jerk reaction is anything but positive. To many in the imaging community, lossy compression means lost data, which is why it is seldom used to store or transmit critically important diagnostic images. But that doesn’t have to be the case, according to results presented at the Society for Computer Applications in Radiology meeting in Orlando.
Mention lossy compression, and the knee-jerk reaction is anything but positive. To many in the imaging community, lossy compression means lost data, which is why it is seldom used to store or transmit critically important diagnostic images. But that doesn't have to be the case, according to results presented at the Society for Computer Applications in Radiology meeting in Orlando.
Two independent reviewers concluded that lossy compression is clinically acceptable for image compression, said Peter Bak, Ph.D., project director of diagnostic imaging architecture for Canada Health Infoway. The organization promotes adoption of electronic health information systems in Canada.
Although some data loss occurs when using lossy compression, the chance that the data will critically affect the ability to make a diagnosis is extraordinarily small, if the right algorithms are used under the right circumstances.
Bak and colleagues commissioned two independent researchers to systematically review research on lossy compression. They found that the modality and size of the image matrix were key considerations in calculating compression tolerances.
Acceptable compression ratios varied from 40:1 for MRI to up to 90:1 for nuclear medicine studies. The researchers further concluded that a 50:1 compression ratio is acceptable for large-matrix images, although 25:1 is optimal. An acceptable ratio for small-matrix images is 20:1. An optimal ratio is 10:1.
They found that such tolerable compressions might be achieved by either of two main categories of algorithms: discrete cosine transform-based or wavelet-based.
The use of these algorithms may come into vogue with the increasing adoption of electronic medical records. IT systems designed for healthcare will require optimum use of networking capabilities among institutions. Compressing images through lossy algorithms can go a long way toward this optimization, according to Bak.
Canadians may be among the first to benefit from this approach, as the Canadian Association of Radiologists has endorsed lossy compression for primary reads.
"We in Canada have a good motive to use lossy compression," he said. "We see lossy compression as having a positive impact on our national EHR plans if we can make it happen."
Conclusions supporting this position presented during the SCAR meeting were based on a review of 120 papers, most over the last seven years in radiology and related journals. The research also queried radiologists in the U.S. and Canada and professional radiology organizations in both countries for position statements on the use of irreversible image compression.
This is only the start. More work must be done to determine legal risks, assess regulatory frameworks, determine economic impact, and provide clinical and performance evaluations for lossy compression to be widely embraced, Bak said.
MRI-Based AI Radiomics Model Offers 'Robust' Prediction of Perineural Invasion in Prostate Cancer
July 26th 2024A model that combines MRI-based deep learning radiomics and clinical factors demonstrated an 84.8 percent ROC AUC and a 92.6 percent precision-recall AUC for predicting perineural invasion in prostate cancer cases.
Breast MRI Study Examines Common Factors with False Negatives and False Positives
July 24th 2024The absence of ipsilateral breast hypervascularity is three times more likely to be associated with false-negative findings on breast MRI and non-mass enhancement lesions have a 4.5-fold likelihood of being linked to false-positive results, according to new research.
Can Polyenergetic Reconstruction Help Resolve Streak Artifacts in Photon Counting CT?
July 22nd 2024New research looking at photon-counting computed tomography (PCCT) demonstrated significantly reduced variation and tracheal air density attenuation with polyenergetic reconstruction in contrast to monoenergetic reconstruction on chest CT.