Advancing CAD Applications
Computer aided detection (CAD) tools are allowing innovative new ways to process the data captured from medical imaging and willone day play a more significant role in helping radiologists image less while obtaining more information. They make radiologists more confident in their diagnoses, while helping reduce chances of missed lesions, and they provide new ways to examine data without having to acquire additional studies.
One example of this utility in action is offered by Riverain Medical, Miamisburg, OH, with the SoftView® product, which removes bone from chest X-ray images. While Riverain has a highly sensitive and specific chest X-ray CAD product in its OnGuard™, the ability to remove bone at the click of a button is providing a new look at chest X-rays.
Similarly in breast CAD, Hologic, Bedford, MA, offers the R2™ Quantra™ software for volumetric assessment, which calculates the volumetric fraction of fibroglandular tissue. These measurements can be quantified over time to provide information about changes in breast tissue.
In magnetic resonance imaging (MRI), CAD applications such as SpectraLook® and VividLook® from iCAD, Inc., Nashua, N.H., and DynaCAD for Breast and DynaCAD for Prostate from Invivo, Orlando, FL, are providing pharmacokinetic maps that are indispensible to the reading of breast and prostate studies. These tools measure the affinity of tissue for gadolinium and they generate a map of areas that have the most malignant characteristics. New software evolutions from iCAD in particular are providing the ability to map this pharmacokinetic activity to morphology.
In the realm of computed tomography (CT) for lung imaging, CAD applications are providing more confidence to radiologists, especially when studies are read for emergency situations like pulmonary embolism and other malignancies have a potential to be missed. On the Siemens Healthcare (Malvern, PA) syngo® Lung CAD product, radiologists can incorporate CAD marks onto any thin section reconstruction and they can jump from mark to mark as they scroll through a series.
Mammography CAD
Mammography has the longest history with CAD and while the solutions are well-developed, vendors have identified 4 major goals, according to Marc Filerman, vice president of global product marketing at iCAD: Improve sensitivity, improve the number of false positives, improve the quality of the false-positive marks, and provide additional information. Mammography CAD is already reaching sensitivity rates in the mid 90th percentile. Reducing the number of false positives and improving the quality of those marks has taken centerstage. As a result of this, CAD systems are in high demand, especially when installing new full field digital mammography systems(FFDM).
“There is a high demand for the CAD application as part of the everyday workplace,” said Sibel Narin, DDS, MBA, product manager for CAD and Image Distribution at Hologic, Bedford, MA. “We ship close to 100% of our FFDM gantries with CAD applications.”
Mammography CAD is maturing in other ways. Systems are now able to better integrate the information captured in the interpretationinto the patient record. Further, new developments are happening across the industry to make the CAD marks themselves more intuitive.The R2 ImageChecker® CAD has EmphaSize™; Marks that correlate in size to the prominence of the mass or calcification. Similarly iCAD is also developing better ways to display CAD marks.
“There are various ways of displaying CAD marks so we do not block the region of interest,” said Narin. “Our peerView™ digital helps the radiologist understand why the algorithm marked a suspicious region.”
Chest X-ray CAD
While not as widely used as mammography CAD, applications are making headway in chest X-ray. The OnGuard system from Riverain can seamlessly display CAD markings on a chest, and its most recent version (5.1) has further reduced the false positives while increasing the sensitivity, according to Steve Worrell, CTO of Riverain Medical.
Just like in mammography settings, OnGuard 5.1 has the ability to use CAD marks themselves to convey more information to the reader. The marks scale in accordance with the underlying lesion size estimate.
CT CAD
Across the board CAD solutions are now providing more diagnostic confidence, especially in settings like CT chest imaging, where overlooked incidental findings can be detrimental to patients.
“Our CT CAD helps me the most when the mark is not on the area of primary consideration,” said Geoffrey H. Browne, MD, radiologist at Alamance Regional Medical Center, Burlington, N.C. “For instance, if I am reading for pulmonary embolism and a lung parenchyma is subsequently marked. We generally observe for those as a matter of course but having a secondary method to identify abnormal chest studies is helpful. The real benefit is that when you read a chest study that you consider negative, then if you apply CAD and that also comes back negative, you feel very confident in that diagnosis.”
The Burlington, N.C., area has a population of former textile workers and heavy smokers, according to Dr. Browne. In this type of population, it is not uncommon to encounter significant amounts of underlying lung disease, which can lead to false positives on CAD.
“If the lungs are not healthy to start with, you will see more false-positive findings,” Dr. Browne said. “You would expect those findings to be more numerous in somebody who has chronic disease. The bigger question is how you manage those outcomes. Some individuals may have scarring from chronic inflammation and chronic lung disease. If you find a 2 mm subpleural nodule in such a patient, it is different than if you see a healthy person with a 2 mm subpleural nodule. This points to a larger issue in terms of how we manage these types of incidental findings. You do not want to perform unnecessary follow-up studies.”
Conclusion
With CAD systems providing new ways of viewing images and of generating more useful information from fewer studies they are providing a critical first step in radiology’s important role of improving overall patient management.