Breast MRI:How I Read It

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon

Ian Gardiner, MDBreast MRI has emerged as a powerful new tool in the fight against breast cancer. It has found wide acceptance in the past 10 years, and it appears to be one of the most rapidly growing medical studies. When breast MRI is combined with mammography and breast ultrasound, we are now able to find breast cancer at its earliest stages, when the disease can be treated most easily and successfully.

Unlike conventional mammograms and sonograms, which might generate a dozen or so images per patient, breast MRI creates 2,000 or more. This has created much additional work-up activity for clinic- and hospital-based radiologists, who are already very busy (and in short supply). The growing demand for breast MRI, combined with the stagnant supply of radiologists, created a need for new tools that would boost the efficiency of radiologists; enter computer-aided detection.

Computer-aided detection streamlines the radiologist’s workflow by automating several time-consuming tasks. First, it processes the raw data to correct for any movement that has occurred during the study, eliminating artifacts from motion. Second, it organizes the images into the radiologist’s preferred hanging protocol, so that the most relevant images are displayed initially.

Third, it analyzes the images to find the areas of tissue that have the most abnormal blood supply (an important sign that can indicate the presence of cancer). Fourth, it simplifies generation of reports, making use of ACR® BI-RADS® terminology.

A typical case will demonstrate some of the ways that I use computer-aided detection. I begin my analysis by looking at the patient’s requisition and other paperwork. It is important to understand why the patient is having a breast MRI exam. Potential reasons include assessment of breast implants, determining the true extent of disease in a newly diagnosed breast-cancer patient, and looking for evidence that treatment is effective. I also review any previous imaging that the patient has undergone, including prior MRI studies, mammograms, and breast sonograms.

image
Figure 1.Maximum-intensity projection showing recurrent breast cancer near the chest wall.

Next, I turn to my monitor and bring up the maximum-intensity projections (Figure 1). These 3D images have been reconstructed, using computer-aided detection, from the raw data. They allow me to have a global view of the breast tissue from a variety of different angles.

image
Figure 2.High-resolution axial image showing recurrent breast cancer near the chest wall.

Once I have the sense of the lay of the land, I go to the next step in my hanging protocol. These are the various high-resolution cross-sectional images that I use to analyze the anatomic characteristics of any lesions (Figure 2). Computer-aided detection includes a number of tools to assist me in my analysis.

With one click, the software can determine precisely where a lesion is located, measure its size, and calculate its distance from various landmarks, such as the chest wall and skin surface. This information is automatically added to the report. I can also select key images that demonstrate particular features of the lesions. These images can be saved to an external device that I can then take to rounds or tumor boards.

image
Figure 3.Graphical representation of the blood-flow pattern through the tumor over time.

In most instances, the high-resolution images give sufficient information to help me decide whether a given lesion is worrisome or not. At other times, I must turn to the so-called kinetic images to help me form an opinion. Kinetic images (Figure 3) can be thought of as time-lapse photography, demonstrating the flow of gadolinium dye through the breasts. This dye tends to accumulate in cancerous tissue.

We look at a variety of parameters to help distinguish between normal and suspicious tissue, including how rapidly blood flows into and out of the area of concern. These calculations are tedious when done manually, but can be rapidly and accurately done by the computer. The software generates color maps that can be superimposed on the anatomic images to highlight the areas of greatest concern. You can think of this as being like a spell checker for a document: It draws the radiologist’s attention to those areas that might require further evaluation.

After I have completely characterized the lesion, I turn my attention to the areas where breast cancer can spread. These include the axilla, skeleton, and liver. Along the