Bone can obscure lung lesions that may indicate cancer. The latest version of computer-assisted detection software from Riverain Medical, shown publicly for the first time in the RSNA 2010 exhibit hall, takes care of the problem.
Bone can obscure lung lesions that may indicate cancer. The latest version of computer-assisted detection software from Riverain Medical, shown publicly for the first time in the RSNA 2010 exhibit hall, takes care of the problem. This version automatically identifies and then draws markers encircling suspected lung cancer nodules on digital chest radiographs from which bone, such as ribs and the clavicle, have been subtracted.
The new offering is a software hybrid of two Riverain products: CAD-based OnGuard 5.1 and the bone-suppressing SoftView. As with each separate product, the bundle works as an integral part of a PACS, requiring no separate viewing station. It also works with all digital and computed radiography systems, according to Steve Worrell, chief technology officer at Riverain Medical.
“You can have the PACS or the modality push the digital chest radiograph to OnGuard or SoftView, which [will then] introduce the imagery or the markers-or marry them together-so they are all available at the point where the radiologist reads,” Worrell said.
In offering OnGuard, SoftView, and a blend of the two, Riverain provides enough choices to satisfy a range of needs, Worrell said. SoftView offers a bone-suppressed image of the chest, which can be used for applications other than nodule detection.
“In some cases, radiologists would prefer to just have a better image, as opposed to having markers on that image,” he said.
OnGuard 5.1 identifies and marks suspicious lesions for radiologists who simply want a CAD program.
As Riverain demonstrated these two separate products, some radiologists asked why they were not combined, allowing the markers to be placed on the bone-suppressed SoftView image. And so Riverain did exactly that.
Developing the two products separately permitted the company to focus on improving the underlying capabilities of both. A fundamental weakness of CAD has been its tendency to impede rather than boost productivity.
CAD typically is used as a second read. The algorithms insert markers indicating suspicious lesions after the primary read, forcing radiologists to go back over what they have already interpreted. Embedding bone suppression in OnGuard addresses this by making the primary read easier and, potentially, faster, according to Worrell.
“Bones are an obstacle to sensitivity and specificity (when interpreting chest radiographs) because lesions can hide behind them,” he said. “To the extent that we can remove the bone, we can make reading easier and better mark those lesions, which also reduces false positives.”
The bone suppression complements a powerful CAD algorithm that, in clinical trials, proved effective at detecting lung nodules ranging from 9 to 30 mm. Research published in Radiology last year documented that the algorithm can detect up to 50% of nodules that would otherwise be missed. The company asserts that improvements in this algorithm can cut false positives by 73% compared with earlier versions of the software.
Cutting the number of false positives attacks the major stumbling block to the widespread use of CAD, Worrell said. Riverain's CAD algorithm has done this while increasing sensitivity.
“As time goes on, we will continue to enrich that marker by encoding perhaps different kinds of information to aid the radiologist in the interpretation,” Worrell said. “The objective is to constantly enrich that environment with more information.”
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.