Is Computer-Aided Detection Becoming A Routine Mammography Analysis Tool?

The use of CAD (Computer-Aided Design) has expanded to include CADx or CADe (Computer-Aided Diagnosis) that assists radiologists and physicians in interpreting medical images.

Often, CADx is referred to as CAD (Computer Aided Detection), and its primary purpose is to reduce the likelihood that a physician makes a diagnosis of a false negative when interpreting medical images.

In this article, CADx denotes Computer Aided Diagnosis, and CAD denotes Computer-Aided Detection. The usage of CAD in this article focuses on its use for Computer-Aided Detection.

The main difference between CADx and CAD is that CADx analyses a radiographic finding to estimate the likelihood that the feature represents a specific disease process (e.g. benign versus malignant). On the other hand, CAD merely identifies suspicious features in a scanned medical image.

The main algorithms in CAD use pattern recognition to identify suspicious features in a scanned medical image. The typical procedure for examining a medical image involves three steps:

• The radiologist performs a cursory examination of the image,
• The image is processed by CAD software, which marks suspicious spots in the image for further evaluation,
• The radiologist reviews the marked suspicious areas from CAD and makes a final determination about the analysis.

Because CAD decreases the likelihood of the radiologist making false negative readings, it has been approved by the FDA and CE for use with both film and digital mammography.

The purpose of this article is to determine whether the use of CAD has become a routine mammography analysis tool. Specifically, it would be interesting to determine

• How extensively is CAD used for breast cancer detection?
• What are the pros and cons of using CAD for detecting breast cancer?
• How effective is CAD in detecting abnormalities in mammography scans?

How extensively is CAD used for breast cancer detection?

Conventional 2D mammography provides the physician with a 2-dimensional image which contains overlapping layers of tissue. The nature of the image makes it difficult to identify suspicious spots, leading to false results such as missed positives and false negatives.

An implementation of CAD called Genius 3D mammography delivers layers of breast images that enable the physician to screen for breast cancer with much greater accuracy than conventional mammography. Genius exams are FDA approved, and its performance provides (a) 41% better detection of invasive breast cancers, and (b) 40% reduction in callbacks, additional tests or biopsies.

This form of 3D mammography, called tomography is in use at over 2400 medical centers.

What are the pros and cons of using CAD for detecting breast cancer?

Published data indicates that CAD is fairly good at detecting invasive breast cancers in their early stages in older women’s mammograms.

When results from CAD are verified by the radiologist, the chances of making an accurate diagnosis are high.

However, CAD also increases the risk of providing false-positive results.

A false-positive result means that CAD identifies abnormal areas (termed DCIS or ductal carcinoma in situ) from the mammogram that look cancerous, but are not. A false-positive result means more doctor visits for the patient, and possibly the need for a biopsy.

DCIS is not invasive cancer, which means that it stays inside the milk duct, and it does not spread outside the milk duct into surrounding normal breast tissues or lymph nodes.

Although CAD is not 100% accurate in detecting breast cancer, published data clearly indicates that CAD improves the likelihood that invasive breast cancers in older women will be detected early.

From Medicare data collected from 163,000 women ranging from 67 to 89 years in age, it was determined that

• CAD-assisted mammograms had 17% improvement in DCIS diagnosis accuracy and 6% improvement in detecting early-stage invasive cancer diagnosis.
• Women whose mammograms were not analyzed with CAD needed 19% more additional tests, and 10% more biopsies.
• Detection of invasive breast cancer was about the same with or without CAD. However, the use of CAD improved the likelihood of detecting invasive cancers at an early stage, thereby improving the chance of survival.

How effective is CAD in detecting abnormalities in mammography scans?

Published data indicates that between 1990 and 2008, the mortality rate from breast cancer declined by 2.2% annually. This decrease is largely attributed to earlier detection, perhaps through CAD-assisted screening and improved treatments. When breast cancer is detected and treated at a localized stage (confined to the breast), the 5-year relative survival rate for women of all races combined is 99%. For regional disease, the rate is 84%. If the breast cancer spreads to distant organs, the 5-year survival rate drops to 23%.

Although the use of CAD is costly, its apparent effectiveness in detecting early stage breast cancer makes it a desirable tool. However, scientific evidence does not yet exist to prove that CAD-assisted mammography reduces morbidity and mortality associated with breast cancer.

Nevertheless, many medical centers adopt CAD-assisted mammography screening and about 90% of mammograms performed in the United States use CAD analysis. Some of these centers are

• The UR (University of Rochester) Medicine Breast Imaging Center in Rochester, New York. At this center, all mammograms are reviewed by CAD technology.
• CAD-assisted mammograms are also performed at the Avon Comprehensive Breast Evaluation Center at Massachusetts General Hospital in Boston.

Conclusions

CAD-assisted mammography screening is not perfect, but it provides a desirable method for early detection of breast cancer. Because early detection makes the diseases most treatable, the use of CAD improves the survival rate of breast cancer patients.

Digital mammography is more accurate than film (conventional) mammography, especially for dense breast tissue that is common in young women.

It is necessary that a radiologist reviews results obtained from CAD in order to minimize false-positive results. Double readings (more than one radiologist reviews CAD results) minimize false-positive readings. A false-positive result means unnecessary extra doctor visits for the patient, and possibly the need for a biopsy.

Insufficient scientific evidence exists at this time to establish that CAD-assisted mammography is essential for analyzing mammograms. For this reason, there is an ongoing debate about eliminating CAD because it is costly to use.

– IndiaCADworks