Segami is cleaning up digital images a little bit at a time. The company, which has offices in Paris and Columbia, MD, has begun marketing an algorithm that filters noise out of images by identifying and then eliminating statistical fluctuations in the
Segami is cleaning up digital images a little bit at a time. The company, which has offices in Paris and Columbia, MD, has begun marketing an algorithm that filters noise out of images by identifying and then eliminating statistical fluctuations in the data. The statistical heuristic image noise extraction (SHINE) algorithm starts by breaking the image into blocks measuring four pixels square. Standard deviations are defined for the signal inside those blocks. Data falling within those deviations are kept, and everything else is thrown out. Individual blocks are then reconstructed and assembled into an image.
"SHINE is a type of filter," said Philippe Briandet, Ph.D., president and CEO of Segami. "But it does not blur high count areas, like some filters do."
These other filters can blur a sharp edge; for example, along the myocardium. They do not distinguish well between signal and noise, Briandet said, unlike SHINE, which cleans up a typical SPECT image in three to four seconds. A whole-body nuclear exam can be processed in 20 seconds.
Using the algorithm can lead to one of two results. One is a sharper, higher quality image. The other is a reduction by half in the dose of radioisotope or data acquisition time, while quality equivalent to the raw image is maintained.
"This has tremendous potential to increase throughput in a busy department," he said. "Dose advantages are important when imaging children, as well as in developing nations that need to save money on dosages."
The algorithm was developed by two French researchers, Pascal Hannequin and Jacky Mas. They later documented that the signal-to-noise ratio in a SHINE-processed image is close to that of a raw image with twice the number of counts.
Segami further validated SHINE and, earlier this year, implemented the algorithm on its Mirage workstation, which offers image fusion, multiformat, and multimodality data exchange. It has also begun marketing SHINE to OEMs for use in their imaging equipment.
The company, which was founded in 1995, specializes in nuclear medicine but offers multimodality applications. Like its developer, SHINE caters to nuclear medicine, but the algorithm is compatible with any digital modality. Its medical imaging potential is virtually unrestricted, Briandet said.
"You input a DICOM file, and it spits out a DICOM file," he said. "There is no user interaction."
Considering its utility and broad applicability, the price is low: $5000.
"You can make a one-headed (gamma) camera give images as good as a dual head," Briandet said. "It is a ridiculously low investment."
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.