• AI
  • Molecular Imaging
  • CT
  • X-Ray
  • Ultrasound
  • MRI
  • Facility Management
  • Mammography

CAD capabilities expand from detection to tissue characterization


With its roots firmly anchored in mammography, computer-aided detection has proved it can flag lesions that might otherwise be missed and, in the process, help improve the detection of breast cancer.

With its roots firmly anchored in mammography, computer-aided detection has proved it can flag lesions that might otherwise be missed and, in the process, help improve the detection of breast cancer.

CAD is now becoming a tool not only for identifying suspicious regions of interest in mammograms, but for more precisely targeting and defining clinically relevant anatomy for other imaging modalities and in any part of the body.

"Drawing a circle around a microcalcification or looking for spicules identifies only small parts of what is going on in the body," said Rich Borrelli, vice president of business development for Guardian Technologies International, which is grooming a new CAD product.

Speaking from the exhibit floor of the Healthcare Information and Management Systems Society conference Feb. 14, Borelli explained that the challenge is to understand where abnormal tissue is and then extract additional information that visually or quantitatively defines the pathological processes under way. This is done by clarifying the image and then segmenting it on the basis of data in the digital file that describe the material in the region of interest.

Guardian is developing algorithms that do these analyses, applying them first to identify weapons and explosives in baggage and other sealed containers. The next step, still possibly years away, is to apply adaptations already in development to medical imaging.

"The trend is to take some of our basic learning about how to detect microcalcifications in the breast and deploy that same kind of technology to other high-visibility disease states and help the radiologist visualize more of the process associated with disease," he said.

CAD systems in development by several manufacturers are capturing kinetic and functional data from MRI examinations of the breast, prostate, and brain to provide another dimension for characterizing and staging lesions, as well as new methods for streamlining MR imaging.

"Originally CAD involved detection, but systems are broader now. They help recognize whether a lesion is malignant or benign, and they improve workflow," said Dr. Andreas Muehler, president of CADSciences of White Plains, NY.

In both the colon and the lung, it is not the presence of an abnormality that is of most concern. It is the size of a polyp or nodule and its pattern of growth. CAD products from Siemens and GE automatically measure and segment lesions and calculate lesion doubling times.

Polyp Enhanced Viewing, a 510(k)-cleared product shown by Siemens at the 2005 meeting of the RSNA, has achieved sensitivities in the 90% range and has a false-positive rate of about 2.5% for 5-mm and 6-mm polyps. The automated polyp measurement tool allows a physician to obtain a direct measurement of individual polyps in 3D even when the lesion spans several slices or extends out of a standard projection.

Lung VCAR (volume computer-assisted reading) from GE, also introduced at the last RSNA meeting, pinpoints the precise location of lung nodules, even small ones that may otherwise go unnoticed. It eliminates manual segmentation, which can be tedious, subjective, and prone to error. The software isolates and precisely analyzes nodules, and it compares current measurements of nodules with findings from previous examinations to compute volume-doubling times.

To provide a more comprehensive evaluation of the lung, researchers from Siemens are working on algorithms that detect ground-glass opacities, and they are testing a visualization tool for pulmonary emboli that depicts filling defects even if vessels run obliquely or cross multiple slices, said Dr. Ingo Schmuecking, director of Siemens' division of marking for computer-aided diagnosis and therapy.

In the breast, it is not just microcalcifications that can indicate disease. Other factors being analyzed by CAD systems, such as underlying structures and organic compounds, as well as lesion behavior, may be early markers of pathology.

Guardian, a CAD and information management business focused on healthcare and transportation security scanning, was created in 2002 to respond to the need for analyzing information embedded in images, whether they are radiologic, pathologic, histologic, or military, Borrelli said.

Over the past few years, the company has been developing 3i (Intelligent Imaging Informatics) core technology that performs what Borrelli calls reactionary intelligent analysis. The process involves segmentation, isolation, clarification, and visualization to extract from image data a unique fingerprint of the components of the image and to identify compounds or structures that are too minute to be seen by the human eye or are obscured by bones or large amounts of tissue.

Guardian does not yet have a product on the market. It is still in scientific and clinical research mode. Borrelli therefore could not disclose the specific types of compounds or structures the company's technology would identify. He noted, however, that the firm was developing CAD systems for mammography and chest imaging.

Invivo and CADSciences are offering products that track contrast enhancement to provide information about tissue biology in MR scans.

A CAD system, in fact, is the most recent addition to the breast MRI lineup for Invivo. Marketed under the name DynaCAD, Invivo's product consists of a digital imaging workstation plus five software modules that allow radiologists to review simultaneously and synchronously the enormous volume of data associated with breast MR examinations, said Tom Tynes, director of the Interventional MRI Business Group.

In addition to modules that correct for patient movement and plan MR-guided biopsies, DynaCAD produces visual indications of contrast enhancement kinetics by graphically representing changes in MRI signal intensity over consecutive scans or time periods, providing 3D renderings that can be rotated to view an image from any vantage point and overlaying dynamic data on maximum intensity projections. Such kinetic analysis can indicate tissue vascularity, which is an early sign of tumor growth.

DynaCAD for breast imaging is being integrated into Siemens' family of Magnetom MR scanners. A DynaCAD neurological package is being developed to accompany Invivo's Eloquence functional MRI system. The neurological system, which was shown as a work-in-progress at the 2005 RSNA meeting, follows the same basic value proposition as DynaCAD (how to help radiologists process volumes of data quickly and efficiently to render a diagnosis or plan a surgery) by fusing functional MRI data and other 3D visualization techniques.

CADSciences applies a pharmacokinetic model to determine vessel permeability and extracellular volume fraction in each pixel of breast and prostate MR scans to generate a full-analysis histogram that segregates pixels of normal tissue from those of malignant or transitional areas.

When used with breast MR imaging, the pharmacokinetic analysis not only can help diagnose cancer, it may also reflect response to cancer treatment.

"Permeability is known to be an early marker of tumor response to treatment. If the tumor is responding to treatment, permeability quickly drops," Muehler said. "This happens within a week or two, maximum. Even before tumor size changes, we see the change in terms of the physiology of the tumor, which may indicate whether the tumor is or is not responding."

The ProStream CADSciences tool for MRI may make prostate scanning more economical for MR imaging centers. According to Muehler, only about 5% of these centers in the U.S. currently perform prostate scans. The drawback is the need for inserting and testing the position of endorectal coils, which greatly lengthens scan time. ProStream may make endorectal coils obsolete for imaging and detecting prostate cancer, Muehler said, by collecting and rearranging raw radiofrequency pulses from MR imaging in k-space. Borrelli agreed.

"There's a lot more to be done in the case of CAD, and it's broader than just being able to define normal tissue from abnormal," he said. "It's really about bringing all the imaging capabilities that are available to the radiologist, so he or she can analyze and further extract information from them."

Recent Videos
Emerging Research at SNMMI Examines 18F-flotufolastat in Managing Primary and Recurrent Prostate Cancer
Could Pluvicto Have a Role in Taxane-Naïve mCRPC?: An Interview with Oliver Sartor, MD
New SNMMI President Cathy Cutler, PhD, Discusses Current Challenges and Goals for Nuclear Medicine
Where the USPSTF Breast Cancer Screening Recommendations Fall Short: An Interview with Stacy Smith-Foley, MD
A Closer Look at MRI-Guided Transurethral Ultrasound Ablation for Intermediate Risk Prostate Cancer
Improving the Quality of Breast MRI Acquisition and Processing
Can Fiber Optic RealShape (FORS) Technology Provide a Viable Alternative to X-Rays for Aortic Procedures?
Does Initial CCTA Provide the Best Assessment of Stable Chest Pain?
Making the Case for Intravascular Ultrasound Use in Peripheral Vascular Interventions
Can Diffusion Microstructural Imaging Provide Insights into Long Covid Beyond Conventional MRI?
Related Content
© 2024 MJH Life Sciences

All rights reserved.